Study population

Between August 2020 and August 2023, 456 patients with suspected NSTE-ACS were screened. Of them, 383 (84%) consented and 379 (99%) completed follow-up, and pre- and in-hospital HEART scores were available for 331 patients (86.4%), see Fig. 3.

Fig. 3The alternative text for this image may have been generated using AI.

Flowchart of patient inclusion and analytical cohorts

Baseline characteristics

The mean age of the 331 complete patients was 65.4 years; 48.0% were female. Cardiovascular risk factors were commonly observed; 28.0% had prior MIBased on pre-hospital HEART-scores 26.0% were low risk(n = 86) and 74.0% intermediate-to-high-risk (n = 245; Tab. 1). The 30-day MACE rate was 4.7% in the low-risk group (all index NSTEMIs) and 12.1% in the higher-risk group.

Table 1 Baseline characteristics based on pre-hospital HEART scoreAgreement between pre-hospital and in-hospital HEART scores

In-hospital assessment reclassified 35 patients to low-risk (n = 121/331, 36.6%) (ESM Table S1). Outcomes in the pre-hospital low-risk group are summarised in ESM Table S2. The overall concordance between pre- and in-hospital scores was moderate (ICC = 0.653, with lowest concordance for history and ECG; ESM Table S3). Differences between pre-hospital and in-hospital HEART risk classification are summarised in ESM Table S4 and Table S5, showing concordant low-risk classification in 66 patients and discordant classification in 75 patients. Outcomes in the false-negative subgroup are detailed in ESM Table S6. Among pre-hospital low-risk patients (n = 86), 66 (76.7%) remained low risk after in-hospital assessment. Interobserver agreement for the total HEART score within this pre-hospital low-risk subgroup was low (ICC 0.16). Component-level interobserver agreement within the pre-hospital low-risk group is provided in ESM Table S3a. A sensitivity analysis correcting the age component based on actual age at inclusion for both pre-hospital and in-hospital assessments resulted in only minimal changes in overall agreement (ICC 0.653 to 0.668) and led to reclassification of four patients (1.2%), all shifting from a HEART score of 3 to 4 (i.e., from low-risk to intermediate-risk). None of whom experienced a MACE during follow-up (ESM Table S3b).

Diagnostic accuracy in predicting MI and MACE at 30 days follow-up

The diagnostic performance of the HEART score in predicting MI and MACEs within 30 days is summarised in Tab. 2 and 3.

Table 2 Diagnostic accuracy of prehospital strategies for diagnosing myocardial infarctionTable 3 Diagnostic accuracy of prehospital strategies using the HEART score for diagnosing major adverse cardiac eventsPrimary analysis

The pre-hospital HEART score of ≤ 3 obtained in the ambulance with POC conventional cTnI yielded a sensitivity of 91.3% and specificity of 28.8% for MI (NPV 95.3%, area under the curve (AUC) of 0.60 (95% confidence interval [CI]: 0.55–0.65). For predicting MACE, the sensitivity was 91.7%, the specificity was 28.1% (NPV 95.2%, AUC 0.72 [95% CI: 0.64–0.81]). The in-hospital ESC 0/1‑h hs-cTnT algorithm (reference standard) achieved 100% sensitivity and NPV for MI and MACE (specificity was 93.4% and 95, and AUC was 0.97 and 0.98, respectively).

Exploratory sub-analyses

Alternative strategies, including pre-hospital HEART scores incorporating serial hs-cTnT and hs-cTnI measurements, or the use of a stricter definition requiring HEAR ≤ 3 and troponin sub-score of 0, showed higher sensitivity but at the cost of reduced specificity (ESM Box S1; Tab. 2 and 3, clinical endpoints are described in ESM Tables S10, S11 and S15).

Clinical outcomes in the pre-hospital low-risk group

Among the 86 pre-hospital low-risk patients (26% of the cohort), four MACEs occurred (4.7%); all were index-admission NSTEMIs, with no urgent revascularization or cardiovascular mortality during follow-up (ESM Tables S2 and S7). After in-hospital reassessment by Cardiac Emergency physicians, 35 patients were reclassified as low-risk, yielding a total of 121 patients. In this extended low-risk group, only two experienced MACEs (1.7%); both patients were already identified as low-risk at the pre-hospital assessment stage (ESM Table S8).

Influence of symptom duration

Among 331 patients, 193 (58%) had symptoms lasting ≥ 2 h (2 h). Based on the pre-hospital HEART score, 46 patients (23.8%) were low risk, and two (4.3%) experienced MACE (ESM Table S12). Further duration-based hs-troponin analyses are shown in ESM Tables S13 (hs-cTnT) and Table S14 (hs-cTnI), with overall diagnostic performance in ESM Table S15.

Discharge diagnoses

ACS was confirmed in 62/331 participants (18.7%). Four events occurred in the pre-hospital low-risk group (4.7%, all NSTEMIs), whereas the remaining 58 occurred in patients classified as intermediate-to-high (23.7%).

Benign thoracic pain was the most common discharge diagnosis (121/331, 36.6%), followed by non-ischaemic cardiac conditions (52/331, 15.7%) and non-cardiac disorders (92/331, 27.8%), with the latter occurring nearly twice as often in the pre-hospital low-risk cohort (43.0% vs. 22.0%), see ESM Table S9). Additional subgroup details are provided in ESM Box S4.