Chest pain: the HEART score in practice

A 52-year-old woman arrives at the emergency department with chest pain that started three hours ago, a pressure-like pain, without radiation. She has treated hypertension and smokes. The ECG shows mild repolarization changes, and the first troponin is within normal limits. The shift is busy, there are five other patients with chest pain in the waiting room, and the question is simple: can she go home, does she need to stay for observation, or should she be admitted?

This dilemma repeats every day. Most chest pain is not acute coronary syndrome, but missing a myocardial infarction has serious consequences, and admitting everyone overloads the hospital and exposes patients to unnecessary tests. The HEART score was created to organize this decision with five items that any emergency physician can assess at the bedside. This text shows how to calculate it, what validation studies found and where its limits are.

Key points

• The HEART score adds history, ECG, age, risk factors and troponin, each worth 0 to 2 points.
• A score of 0 to 3 indicates low risk of a major cardiac event; 4 to 6, moderate risk; 7 to 10, high risk.
• Large prospective studies confirmed the score's ability to separate risk groups.
• Combined with serial troponin, the score allows safe early discharge in a large share of patients.
• Performance drops in settings very different from the one where it was created, and clinical judgment remains necessary.

Why chest pain is a challenge in the emergency department

Chest pain accounts for 9 to 10 out of every 100 visits to emergency services (Alley et al., 2015). In the United States, management of low-risk patients with chest pain involves about 8 million visits per year (Amsterdam et al., 2010). Most of these patients do not have a life-threatening condition, but the physician must separate those who need urgent treatment from those who can go home (Amsterdam et al., 2010).

Acute coronary syndrome is present in 10 to 20 out of every 100 patients who arrive with chest pain (Huis in 't Veld et al., 2017). At the same time, 2 to 5 out of every 100 patients with acute coronary syndrome are inappropriately discharged (Huis in 't Veld et al., 2017). Inadvertent discharge of a patient with acute coronary syndrome is associated with higher mortality, while admission of those without serious disease is neither indicated nor cost-effective (Amsterdam et al., 2010).

Symptoms of acute coronary syndrome are often atypical, and traditional risk factors have little diagnostic value in acute care (Alley et al., 2015). For this reason, workup is often long and costly (Alley et al., 2015). How much one accepts missing is a matter of debate. A classic study reported a rate of 2 to 4 out of every 100 for undiagnosed myocardial infarction and unstable angina, and another analysis calculated that 2 out of every 100 would be acceptable based on the test threshold (Alley et al., 2015). Even so, the most cited acceptable rate is below 1 out of every 100, and for this reason most tools aim for a negative predictive value above 99 out of every 100 (Alley et al., 2015). There are estimates that the chance of death doubles when a patient with myocardial infarction is discharged from the emergency department (Alley et al., 2015).

Clinical judgment and basic tools, such as history, physical examination and ECG, remain the basis for identifying low-risk patients (Amsterdam et al., 2010). The HEART score organizes exactly these elements, adding troponin to them. Thinking in terms of pretest probability helps to understand why such a score is useful.

How the HEART score was created

HEART was proposed in 2008 by Six and colleagues, in the Netherlands, from observing how on-call physicians decided about patients with chest pain (Six and colleagues, 2008). The authors noticed that the decision to admit was based on an overall impression, the history, the risk factors, the ECG and the infarction markers (Six and colleagues, 2008). The proposal was to turn this reasoning into a simple score (Six and colleagues, 2008).

The original study included 122 patients seen with chest pain in the first quarter of 2006 (Six and colleagues, 2008). Twenty-nine had one or more outcomes: 16 acute infarctions, 20 revascularizations and 2 deaths (Six and colleagues, 2008). The mean score was 6.51 among those who had an outcome and 3.71 among those who did not (Six and colleagues, 2008).

The risk bands came from this study. With a score of 0 to 3, the risk of an outcome was 2.5 per 100, which would support immediate discharge (Six and colleagues, 2008). With 4 to 6 points, the risk was 20.3 per 100, which would indicate observation (Six and colleagues, 2008). With 7 or more, the risk reached 72.7 per 100, which would support an early invasive strategy (Six and colleagues, 2008).

The score was not derived by logistic regression or any other multivariate technique; it was built from clinical experience and the literature, with the idea of being as simple as the Apgar score (Alley and colleagues, 2015). In the original study, about one third of the patients were classified as low risk, with only one missed outcome, but the effect of treatment or additional tests in this group was not assessed (Alley and colleagues, 2015). The authors themselves described the score as an easy, fast and reliable predictor (Six and colleagues, 2008). The small sample, however, required validation in larger populations, which came in the following years (Huis in 't Veld and colleagues, 2017).

The five components and the scoring

Each letter of the name corresponds to a component, scored from 0 to 2 (Six and colleagues, 2008). The table below summarizes the scoring proposed in the original study (Six and colleagues, 2008).

Component	0 points	1 point	2 points
History	Slightly suspicious	Moderately suspicious	Highly suspicious
ECG	Normal	Nonspecific repolarization abnormality	Significant ST depression
Age	Under 45 years	45 to 65 years	65 years or older
Risk factors	None known	1 or 2 factors	3 or more factors, or known atherosclerotic disease
Troponin	Up to the normal limit	1 to 2 times the limit	More than 2 times the limit

Since each item is worth 0, 1 or 2, the score is easy to memorize and to calculate without a computer (Alley and colleagues, 2015). The history is the most subjective component, because the original table only asks that it be classified as slightly, moderately or highly suspicious (Six and colleagues, 2008). Two people can give different ratings to the same account, and this is a known source of variation.

The risk factors considered include treated diabetes, current or recent smoking, diagnosed or treated hypertension, diagnosed hypercholesterolemia, family history of coronary disease and obesity (Alley and colleagues, 2015). A history of significant atherosclerosis, such as prior revascularization, infarction, stroke or peripheral arterial disease, is worth 2 points on its own (Alley and colleagues, 2015).

Troponin is interpreted in relation to the upper limit of normal of the test used (Six and colleagues, 2008). This point gained importance with the arrival of high-sensitivity troponins and serial measurement protocols.

A detail worth checking: in the published version of the original table, the 2-point band for age appears printed as "≤65", and the reading consistent with the other bands is 65 years or older (Six and colleagues, 2008).

What the validation studies show

The first major validation came from Backus and colleagues. In a retrospective study of 880 patients from four Dutch hospitals, major cardiac events at 6 weeks occurred in 17.95 per 100, and among the 303 low-risk patients the rate was 0.99 per 100 (Alley and colleagues, 2015). Huis in 't Veld and colleagues summarize this result as 34 per 100 of the patients classified as low risk, with events in 0.99 per 100 (Huis in 't Veld and colleagues, 2017). Other validation studies showed consistent results, with event rates below 2 per 100 at a score of 0 to 3 (Alley and colleagues, 2015). In many services, however, an event rate above 1 per 100 is considered unacceptable (Alley and colleagues, 2015).

The next prospective study included 2440 patients in 10 Dutch hospitals (Huis in 't Veld and colleagues, 2017). The C statistic of HEART was 0.83, versus 0.75 for TIMI and 0.70 for GRACE (Huis in 't Veld and colleagues, 2017). In the low-risk group, the rate of major cardiac events was 1.7 per 100 (Huis in 't Veld and colleagues, 2017).

Later studies tested the score in other populations. In a study of 809 patients in the derivation cohort and 298 in the validation cohort, the area under the curve of HEART for events at 3 months was 0.88 in both groups (Kim and colleagues, 2021). In this study, age was not statistically significant in the multivariate analysis, and a modified version without the age component reached 0.91 (Kim and colleagues, 2021). TIMI was 0.83 in derivation and 0.81 in validation (Kim and colleagues, 2021). With a cutoff chosen to keep the negative predictive value above 99 per 100, the modified version had a sensitivity of 97.4 and a negative predictive value of 99.3 per 100 (Kim and colleagues, 2021).

Context matters a great deal. In an emergency department in Tanzania, 927 patients with chest pain were followed for 30 days (Grisel and colleagues, 2024). Major cardiac events occurred in 23.3 per 100, and deaths accounted for 17.6 per 100 (Grisel and colleagues, 2024). The recommended cutoff of 4 points had a sensitivity of 59.4 per 100, a specificity of 52.8 per 100 and a negative predictive value of 74.7 per 100, with an area under the curve of 0.61 (Grisel and colleagues, 2024). The positive likelihood ratio ranged from 1.023 at the 1-point cutoff to 3.556 at the 7-point cutoff (Grisel and colleagues, 2024). The authors concluded that the score did not perform as observed in high-income countries (Grisel and colleagues, 2024).

This contrast teaches something central about scores: performance depends on the population. Where the prevalence of severe disease is high and access to exams is different, the same cutoffs may not protect the patient. It is a reminder that clinical decision-making is not just about applying a number.

The HEART Pathway and serial troponin

The score alone uses a single troponin, and the HEART Pathway was created precisely to address the rate of missed events above 1 per 100 (Alley and colleagues, 2015). It combines the score with troponins measured on arrival and after 3 hours (Huis in 't Veld and colleagues, 2017). In this model, the patient with a low score and negative troponins can be discharged early (Huis in 't Veld and colleagues, 2017).

Alley and colleagues describe a validation of the HEART Pathway with 1070 patients (Alley and colleagues, 2015). These patients stayed in an observation unit for stress testing, and nearly all underwent some objective test, one third of them coronary CT angiography (Alley and colleagues, 2015). The strategy had sensitivity of 100 per 100 for major cardiac events, and 82 per 100 of patients would be eligible for early discharge (Alley and colleagues, 2015). The HEART score alone, without serial troponin, would have missed 5 patients, equivalent to 0.6 per 100 (Alley and colleagues, 2015). A retrospective validation in a multicenter cohort of 1107 patients from 18 American services, with acute coronary syndrome in 22 per 100, found sensitivity of 99 per 100 and identified 20 per 100 of patients as eligible for early discharge (Alley and colleagues, 2015).

A randomized trial with 282 patients compared the HEART Pathway with usual care (Alley and colleagues, 2015). Early discharge occurred in 39.7 per 100 in the protocol group, versus 18.4 per 100 in usual care, and median hospital stay fell by about 12 hours (Alley and colleagues, 2015). No patient classified as low risk by the protocol had a major cardiac event within 30 days (Alley and colleagues, 2015). The protocol was used as decision support, not as a substitute for clinical judgment (Alley and colleagues, 2015). Within 30 days, 8 patients in the low-risk group, equivalent to 12 per 100, underwent stress testing, and another 19 considered low risk were tested during the initial visit (Alley and colleagues, 2015). Of these two groups, 2 patients had inducible ischemia (Alley and colleagues, 2015).

The current trend is to go further and reduce the number of measurements. The French eCARE protocol, a trial in 10 emergency departments, tests a strategy in which the physician does not order troponin when the HEAR score, without troponin, is less than 2 (Moumneh and colleagues, 2022). When HEAR is 2 or more and HEART is less than 4, the strategy skips the second measurement (Moumneh and colleagues, 2022). The primary outcome is non-inferiority in the rate of major cardiac events between discharge and 30 days (Moumneh and colleagues, 2022).

Another front is the moderate-risk patient. In a retrospective study, 333 of 547 patients with HEART of 4 to 6 were sent home with a rapid cardiology appointment, after shared decision-making (Pawlikowski and colleagues, 2023). Median time to the appointment was 2.9 business days, and 264 patients, or 79 per 100, attended (Pawlikowski and colleagues, 2023). There was 1 death within 30 days, before the appointment, and no myocardial infarction (Pawlikowski and colleagues, 2023).

Impact on practice and limits of the score

Adopting a HEART-based protocol changes emergency department flow. In one hospital, comparison of 900 records before and 900 after adoption showed an increase in emergency department discharges, with an odds ratio of 1.56 (Bylund and colleagues, 2021). Admissions for cardiac investigation fell, with an odds ratio of 0.46, and emergency department stay was 17 minutes shorter (Bylund and colleagues, 2021). Stress tests also decreased, with an odds ratio of 0.47 (Bylund and colleagues, 2021). The study included patients with chest pain and patients with atypical symptoms only (Bylund and colleagues, 2021).

The score fits an older model of organization. Low-risk patients with chest pain are increasingly seen in chest pain units, with accelerated protocols of ECG and serial markers of cardiac injury to rule out acute coronary syndrome (Amsterdam et al., 2010). When these results are negative, the protocol usually ends with a confirmatory test to rule out ischemia, generally an exercise stress test or cardiac imaging (Amsterdam et al., 2010). It was this type of test that decreased after the adoption of the HEART-based protocol described above (Bylund et al., 2021).

Stratification may even begin before the hospital. An economic evaluation in the state of Victoria, Australia, modeled the care of 188551 patients brought by ambulance with chest pain (Dawson et al., 2023). Risk stratification by paramedics with on-site troponin resulted in a lower annual cost than current care, both without and with pre-hospital discharge (Dawson et al., 2023). The authors note that the safety of this model still needs to be confirmed in prospective studies (Dawson et al., 2023).

The limits also appear in specific groups. According to the 2021 American guideline, summarized by Riley et al., women seek the emergency department more often with chest pain, but receive less timely and adequate care than men (Riley et al., 2022). For this reason, the guideline recommends, in women, a history that values the associated symptoms most common in them (Riley et al., 2022). In patients over 75 years old, the guideline recommends considering that falls, syncope or acute delirium may be manifestations of acute coronary syndrome (Riley et al., 2022).

Conscious and unconscious biases linked to sex, age, ethnicity and culture can influence how symptoms are classified (Riley et al., 2022). Because the history component of HEART is subjective, these biases can enter the score directly. A low score calculated from a poorly taken history does not protect the patient.

How NEXO helps train this reasoning

The limits of use of the game are described on the NEXO economy page. If you research the teaching of clinical reasoning, the page for researchers explains how to get in touch. Professionals who write or review cases find information in curation and in creators, and institutions can learn about the proposal on the partnerships page. Training differential diagnosis with clues and seeking the right ICD-10 code complete this training.

Frequently asked questions

Does a low HEART score guarantee that it is not a heart attack?

It does not guarantee it. In the Dutch prospective study, the low-risk group had major cardiac events in 1.7 per 100 (Huis in 't Veld et al., 2017). Combining the score with serial troponin increases safety (Alley et al., 2015).

What is the difference between the HEART score and the HEART Pathway?

The HEART Pathway adds troponins measured on arrival and after 3 hours to the score (Huis in 't Veld et al., 2017).

Does the score work in any country?

Not necessarily. In an emergency department in Tanzania, the area under the curve was 0.61 (Grisel et al., 2024).

Does a moderate-risk patient always need to be admitted?

Not always. In one study, patients with a score of 4 to 6 who went home with a quick cardiology appointment had 1 death and no heart attack in 30 days (Pawlikowski et al., 2023).

Is it possible to skip troponin in some patients?

That is the question of the eCARE trial, in 10 emergency departments: not measuring troponin when the HEAR score is less than 2 and measuring only once when HEART is less than 4 (Moumneh et al., 2022). The result still depends on the conclusion of the study.

How much is acceptable to let slip by?

The most cited acceptable missed event rate is below 1 per 100 (Alley et al., 2015).

Does the score work for older adults and women?

It works, but it requires attention to the history. The guideline recommends giving weight to associated symptoms in women and considering falls, syncope or delirium above 75 years of age (Riley et al., 2022).

Take-home messages

• HEART adds up five components scored from 0 to 2: history, ECG, age, risk factors and troponin.
• A score of 0 to 3 indicates low risk, 4 to 6 moderate and 7 to 10 high.
• Combined with serial troponin, it allows safe early discharge in many patients.
• Protocols based on HEART increase discharges and reduce admissions and tests.
• Performance depends on the population, and a poorly taken history compromises the score.

References

1. Alley W, Mahler SA. Clinical decision aids for chest pain in the emergency department: identifying low-risk patients. Open Access Emerg Med. 2015. PMID 27147894.
2. Amsterdam EA, Kirk JD, Bluemke DA, et al. Testing of low-risk patients presenting to the emergency department with chest pain: a scientific statement from the American Heart Association. Circulation. 2010. PMID 20660809.
3. Bylund WE, Cole PM, Lloyd ML, et al. Effect of Implementation of HEART Chest Pain Protocol on Emergency Department Disposition, Testing and Cost. West J Emerg Med. 2021. PMID 33856317.
4. Dawson LP, Nehme E, Nehme Z, et al. Chest Pain Management Using Prehospital Point-of-Care Troponin and Paramedic Risk Assessment. JAMA Intern Med. 2023. PMID 36715993.
5. Grisel B, Adisa O, Sakita FM, et al. Evaluating the performance of the HEART score in a Tanzanian emergency department. Acad Emerg Med. 2024. PMID 38400615.
6. Huis In 't Veld MA, Cullen L, Mahler SA, et al. The Fast and the Furious: Low-Risk Chest Pain and the Rapid Rule-Out Protocol. West J Emerg Med. 2017. PMID 28435499.
7. Kim MJ, Ha SO, Park YS, et al. Validation and modification of HEART score components for patients with chest pain in the emergency department. Clin Exp Emerg Med. 2021. PMID 35000355.
8. Moumneh T, Penaloza A, Charpentier S, et al. Efficacy of HEAR and HEART score to rule out major adverse cardiac events in patients presenting to the emergency department with chest pain: study protocol of the eCARE stepped-wedge randomised control trial. BMJ Open. 2022. PMID 36600358.
9. Pawlikowski A, Hubbard E, Krauss J, et al. Early emergency department discharge for intermediate heart score patients presenting for chest pain. J Am Coll Emerg Physicians Open. 2023. PMID 37692195.
10. Riley RF, Batchelor WB, Goldstein JA, et al. The 2021 AHA/ACC Guideline for the Evaluation and Diagnosis of Chest Pain: An Interventionalist's Viewpoint. J Soc Cardiovasc Angiogr Interv. 2022. PMID 39131964.
11. Six AJ, Backus BE, Kelder JC. Chest pain in the emergency room: value of the HEART score. Neth Heart J. 2008. PMID 18665203.