In this study, we analyzed the accuracy of HEART score in predicting short-term risk of major adverse cardiovascular events in a large contemporary cohort of Tunisian patients presenting to the ED for acute undifferentiated chest pain. We found that HEART score has an excellent prognostic value and could serve as an effective risk stratification tool.
It is well known that patients admitted to the ED with chest pain are at risk for several life-threatening conditions. Given the difficulties associated with accurate risk stratification of these patients, and the potential consequences associated with inappropriate discharge, clinicians often elect to admit patients even whom they believe to be at low risk of MACE [
14]. As a result, the AHA/ACC guidelines have recommended that risk stratification scores should be used to aid in clinical decision-making [
1], but the best one to be used is still unknown. Among the widely used risk stratification scores, we mention TIMI score and the GRACE score. However, neither TIMI nor GRACE scores were designed for ED chest pain risk stratification. The relevance of their use for MACE prediction and patient disposition in ED undifferentiated chest pain is up for debate [
15‐
18]. Boubaker et al. have shown that both scores had low prognostic value and do not serve as an effective risk stratification tool in Tunisian chest pain population [
19]. Recently, the HEART score has emerged as a reliable alternative to risk stratify patients presenting with undifferentiated chest pain in emergency departments. It was firstly introduced in Europe. It is easy to use, intuitive and includes well-established factors associated with the probability of having an ACS. It helps identifying low risk patients in whom immediate further cardiac testing can be safely forgone. It was developed and validated initially in Netherlands, then validated in American and Asian populations but has not been tested in North African populations. Ethnic validation of risk scores is paramount, since inter-ethnic differences in cardiovascular risk factors potentially influence risk score performances. Thus, extrapolation of risk stratification models to different settings cannot be anticipated [
20‐
23]. This is the first prospective study assessing HEART score risk stratification performance in a large sample of Tunisian patients with acute chest pain. Our study population is comparable to the original study population [
7] and other previous cohorts validating HEART score [
3,
24,
25]. However, our patients were younger. In addition, we had higher frequency of diabetes and lower rate of hypertension in our population. Like the previous validating HEART score studies, our incidence of MACE was low in the three HEART score categories, compared to the original study population [
26]. Our results suggest that HEART score can be used in risk stratifying chest pain patients. A c-statistic of 0.87 for the HEART score indicates a good to excellent ability to discriminate the short-term prognosis in ED chest pain patients. Our findings are in accordance with the results of many studies that tested and validated the clinical value of the HEART score in many countries. Matthew et al. [
27] reported that the AUROC of HEART score was as good as 0.88, 95% CI (0.84–0.93); they showed that there was no MACE in their low risk group patients. In a stepped-wedge, cluster randomized trial; Poldervaart et al. [
28] also reported that HEART score was an accurate risk stratification instrument and safe to use when assessing patients with chest pain in the ED. In our low risk group patients, incidence of MACE was 1.2%. This finding is quite similar to that shown in the meta-analysis of Berg et al. [
29]. They showed that the pooled incidence of “missed” MACE was 1.6% in the low risk group. It is noteworthy that none of our low risk patients died during the 30-day follow up. Even though, we should acknowledge that to safely rule-out ACS, the HEART score should achieve higher sensitivity and NPV [
30]; its poor calibration must still urge to more caution in its use in clinical practice.
There are several potential limitations to our study that should be acknowledged. First, although our study has the largest sample size, it may not reflect the full spectrum of patients with acute non traumatic chest pain because we did not include those admitted to nonteaching hospitals or patients treated in an ambulatory setting. However, many of these patients were secondarily transferred to our emergency departments to be investigated in our chest pain units. Second, the relative short duration of follow-up in current study could not allow us to observe more clinical events. Future study is warranted to evaluate whether the HEART score is also useful for longterm risk prediction in our population. Third, the HEART score utilized conventional troponins as cardiac biomarker component; the question whether using highly sensitive troponins would improve the performance of the score is a relevant issue requiring a specific study. Fourth, we need to know more about the real effects of using the HEART score in routine clinical practice. Such experience should be investigated to provide the evidence for its potential benefit. Fifth, there were 76 (1.8%) patients lost to follow. Despite that such rate is considered low, assuming that these patients reached endpoint, there could be an impact on the estimated prognostic accuracy.