Acute non-traumatic chest pain (ANTCP) is the second cause of consultation in emergency departments (ED) in industrialized countries. About 70% of patients with ANTCP who develop major adverse cardiac events (MACE) present as a non persistent ST-elevation acute coronary syndrome (NSTE-ACS) in the electrocardiogram (ECG). Managing ANTCP suspected of NSTE-ACS correctly in EDs and primary care emergency centers (PCECs) is essential to avoid inappropriate discharges and costs related to complementary tests and unnecessary admissions^1–3.

On the one hand, inadequate discharges lead to new consultations for ANTCP, and some patients end up developing MACE within six weeks. In a study conducted in the US in 2010⁴, 5–10% of patients were discharged from the ED because of non-coronary ANTCP, and in subsequent visits they were diagnosed with an acute myocardial infarction (AMI).

On the other hand, Owens et al.⁴observed that 20% of the consultations for ANTCP and probable NSTE-ACS result in acute coronary syndrome (ACS). Half of the patients admitted for suspected ACS who underwent diagnostic cardiac tests were ultimately not diagnosed with ACS. These tests amount to 10 billion dollars/year,or 3,000–6,000 dollars/patient,butonly 10% of these indicate a cardiac issue.

In Spain², ACS is one of the main causes of morbidity, mortality, and costs². There are approximately 120,000 cases of ACS per year, 41.67% of which are hospitalized with NSTE-ACS.

Finally, because diagnosing MACE in the context of an NSTE-ACS is complex, it is important to have a rapid and reproducible tool for both high-intermediate-risk patients, who require additional tests, and low-risk patients, who can be discharged quickly and safely.

In an attempt to improve theaccuracy in the diagnosis of MACE in patients with probable NSTE-ACS, the European Society of Cardiology (ESC)³ and the National Institute for Health and Care Excellence (NICE)⁵ recommend the use of the Global Registry of Acute Coronary Events (GRACE 2.0)⁶ and the thrombolysis in myocardial infarction (TIMI) risk scores⁸. However, according to validation studies, these recommendations are more useful in patients with a high probability of presenting ACS^10,11,14.

To help identify low-risk patients, Six AJ et al. (2008)⁷ developed the medical history, ECG, age, risk factors, and troponin (HEART) score,a risk stratification tool for MACE in the context of NSTE-ACS. In 2013 Backus et al.⁹ externally validated the HEART score in a prospective multicenter study, in which the c-statistic of the HEART score (0.83) was significantly higher than that of the TIMI (0.75) and GRACE (0.70) respectively (; p < 0.0001) scores.In their systematic review and meta-analysis, Ke J et al. (2021)²⁶ observed that TIMI and HEART were superior to GRACE for predicting MACE risk in acute chest pain patients admitted to the ED.

Our aim was to prospectively compare the performance of the HEART, TIMI, and GRACE scores relative to the diagnosis and prognosis of MACE at six weeks in patients presenting with ANTCP and suspected NSTE-ACS in the ED of a Hospital and of a PCEC in Lleida, Spain.

Figure 1 shows the flow of patients in the study. A total of 3112 patients attended the HUAV ED and PCEC for ANTCP. Using consecutive systematic sampling and after applying the inclusion, exclusion, and follow-up loss criteria, we obtained a representative sample of 317 patients. The mean age was 58 years (Table 1), and 67.5% of them were male.

Table 1 shows the variables that were associated with MACE: 48.9% (p < 0.001) of patients had a clinical history of chest pain which was moderately suspicious; 50.0% (p = 0.016) were hypertensive; 93.7% (p < 0.001) had a Killip class 1; 64.1% (p < 0.001) had a normal ECG; 87.0% (p < 0.001) had troponin within normal limits; 51.5% (p < 0.001) had one or two cardiovascular risk factors; 78.7% (p = 0.107) were men; 57.4% (p < 0.012) were under 65 years of age;55.3% lived in urban areas (p < 0.803); 46.8% (p < 0.064) had an education level equivalent to an unskilled worker; and 55.3% (p < 0.318) lived with their partner.

The HEART score had the highest AUC (0.743; 95% CI: 0.674–0.812; Fig. 2).Figure 2

ROC curves for the HEART, TIMI, and GRACE risk scales.

Overall, 13.8% of patients had MACE at six weeks: 17.4% in the PCEC and 15.4% in the HUAV ED. MACE were most frequent in patients classified as high-risk according to the HEART score (35.9%; with a 95% CI: 21.20–52.82), followed by high-risk patients according to the GRACE (34.3%; with a 95% CI: 19.13–52.21) and TIMI scales (33.3%; with a 95% CI:13.34–59.01; Table 2).Table 2

MACE according to the risk stratification of the HEART, GRACE, and TIMI scales.

The mean HEART score was4 in the PCEC (95% CI: 2.00–5.00) and the ED (95% CI: 3.00–6.00). The GRACE scale scored lower in the PCEC than in the ED 77 (95% CI: 59.5–101) vs 99 (95% CI: 77.0–126). The TIMI score obtained a 1 (95% CI: 0.00–2.00) in the PCEC and a 2 (95% CI: 2 1.00–3.00) in the ED.

The most frequent MACE were those that required PCI (10.7%;). The most frequent type of MACE in high-risk patients differed according to the risk stratification score: MACE requiring PCI, according to the HEART score (25.6% and DDTAC, according to the GRACE (17.1%) and TIMI (16.7%) scores (Table 2).

The balanced accuracy in the risk stratification of MACE in the low- and moderate–high risk patients was 0.67 (95% CI: 0.58–0.69) between the HEART and GRACE scores, 0.68 (95% CI: 0.59–0.69) between the HEART and TIMI scores, and 0.67 (95% CI: 0.67–0.77) between the TIMI and GRACE scores.

The ability of the HEART scale to diagnose MACE was very high, with an SE of 100% (95% CI: 100–100) and an NPV of 100% (95% CI:100–100) in low-risk patients, and an SP of 90.74% (95% CI:87.28–94.2) and a PPV of 88.13% (95% CI:84.33–91.93) in high-risk patients (Table 3).Table 3

Performance characteristics of the TIMI, GRACE, and HEART risk scales.

Regarding the likelihood of MACE occurring, the HEART score performed significantly better than the TIMI and GRACE scores: the risk of MACE in moderate-risk patients was 5 times higher than in low-risk patients (OR 5.32[95% CI: 2.07–15.79]); Table 4).Table 4

Comparison of the Odds Ratio—MACE according to the risk stratification of the HEART, GRACE, and TIMI scales.

The mean time to MACE was one day (95% CI:0.00–8). The mean time to DDTAC was nine days (95% CI: 7.50–24), to AMI was one day (95% CI: 0.00–1), to PCI was one day (0.25–9.50), and to CABG was one day (95% CI: 0.25–8.75) (Fig. 3).Figure 3

Survival analysis: global Kaplan Meier curve.a: major acute cardiac event; b:acute non-traumatic chest pain.

The kappa index in the categorical classification of patients as low- and moderate–high-risk was 0.32 (95% CI: 0.58–0.69) according to the HEART and GRACE scores, according to the HEART and TIMI scores, and 0.31 (95% CI: 0.67–0.77) according to the GRACE and TIMI scores.

In this study, we compared the effectiveness of the HEART, GRACE, and TIMI scales for the diagnosis and prognosis of MACE at six weeks in patients with ANTCP and suspected NSTE-ACS. The HEART scale was superior to TIMI and GRACE, with results similar to previous studies^10,11,21. Specifically, the HEART scale is more effective in stratifying low-risk patients, with an excellent level of safety. In this group, 2.18% of patients (95% CI 1.2–3.2) had MACE at six weeks, similar to the 2.5% observed by Six et al.⁷, and our own previous observations from a retrospective study¹⁵.

The HEART score is currently validated for use in some medical emergency services, and it is considered superior to TIMI and GRACE (GRACE 2.0) for determining MACE risk at six weeks, with highSE and a NPV^9,11,26. Poldervaart et al.¹⁰ observed that the HEART score identified low-risk patients better than the TIMI and GRACE scores, with only a 0.8% incidence of MACE in this group of patients. Stopyra et al.¹⁴ integrated the HEART score with the clinical history of patients and observedthat 97.5% of low-risk patients did not die of AMI within one year, with a 7% reduction in the yearly hospitalization rate.

The superior diagnostic and predictive ability for MACE at six weeks of the HEART score compared to the TIMI and GRACE scores was expected. Indeed, HEART was designed to predict MACE at six weeks in patients with ANTCP^7,9, while TIMI and GRACE were created to determine the prognosis of patients with unstable angina and NSTE-ACS and are not effective for patients with a medium–low probability of MACE¹⁹. Therefore, the efficacyof the TIMI and GRACE scores in predicting MACE is questioned and subjected to constant validity studies^14,19,21,26.

The higher the SE and NPV, the better the clinical scoring system for risk stratification²². In our study, the HEART score stratified low-risk patients for MACE at six weeks better than the TIMI and GRACE scores, with the best SE and NPV. Similar results were obtained in previous validation studies, in which the SE ranged from 99% (95% CI 97%–100%) to 99.5% (95% CI 97.1%–99.9%) and the NVP between 99% (98%–100%) and 99.6% (95% CI 97.3%–99.9%) ^9–11,14,19.

We also observed a higher AUC and a better PPV in high-risk patients (35.9% [95% CI: 20.84–50.95]) for the HEART score compared to the TIMI and GRACE scores, reinforcing the superior diagnostic performance of this clinical tool. Similar results were obtained by others for the AUC (0.83 [95% CI 0.81–0.85]) for HEART regarding MACE at 30 days⁹ and PPV (36%[30%–41%]–46% [40%–52%])^8,9,19.

Kline JA et al.²¹ evaluated the risk score for MACE at 30 days for chest pain, suggesting arange of0.5%–3.0% for a valid FNR, with an average estimationof 2.0% for low-risk patients. Wamala et al.¹⁹ compared nine coronary risk scores and alsoobtained an FNR of 2% for the HEART score in low risk patients, while Stark et al.²² found an FNR of 0.09% to 3.2% for low-riskMACE patients according to the HEART scale. We obtained similar results, with an FNR of 0 to 2.13% for low-risk MACE patients according to the HEART scale.Additionally, the OR and the Kapplan Meir we obtained highlight the better risk stratification by the HEART score and are similar to that of Six et al.⁷.

Scales that do not include chest pain would be expected to have better reproducibility and reliability and less interobserver bias y¹². On the contrary, we observed that chest pain is a relevant component and does not reduce the effectiveness of the HEART score, with each of its variables being statistically significant (p < 0.001).

The kappa index between the HEART, GRACE, and TIMI scores was acceptable in the categorical classification of patients as low- and moderate–high-risk, which reinforces the high degree of reliability of the HEART scale.

The effective risk stratification of ANTCP patients in EDs and PCECs is always a challenge. Recent attempts to further improve the accuracy of predictive models include more specific cardiac markers and the combination of risk scores and other complementary tests that are not present in some hospital EDs and most PCECs. In this sense, the HEART score is easier to calculate compared to other risk scales because its elements are more affordable. From the GRACE score, the Killip class component requires the physician’s judgment for the diagnosis of heart failure, generating inter-rater variation. Very few patients undergo additional tests for heart failure, such as ECG and natriuretic peptide level, because they are not available in PCECs and some hospital EDs. In our study, only 1.2% of the patients assisted in the HUAV ED had their natriuretic peptide level measured, and 31.9% underwent an ECG.In the CUAP, none of the patients underwent these tests because the resources were not available.

The HEART score is better than the TIMI and GRACE scores for the diagnosis and prognosis of MACE at six weeks for low-risk patients in Spain attending PCECs and EDs with ANTCP and suspected NSTE-ACS.This eliminates the need for more complementary tests without compromising patient health.