Coronary heart disease (CHD) is a pathological condition characterized by coronary artery stenosis or occlusion caused by atherosclerosis. Acute cardiovascular events triggered by CHD remain the leading cause of mortality worldwide (1). According to the 2023 Report on Cardiovascular Health and Diseases in China, the number of patients with CHD has reached 11.39 million. The disease burden continues to grow, driven by an aging population and the increasing prevalence of metabolic disorders (2). Myocardial ischemia—the central clinical manifestation of CHD—results from an imbalance between coronary blood supply and myocardial oxygen demand, leading to insufficient oxygen delivery to cardiac tissue (3). Without timely intervention, transient ischemia may quickly evolve into acute coronary syndrome, significantly elevating the risk of sudden cardiac death (4). Accordingly, the development of reliable early diagnostic tools for myocardial ischemia is critical to improving outcomes and optimizing healthcare resource allocation.

Among current diagnostic modalities, dynamic electrocardiography (ECG) is widely employed for out-of-hospital myocardial ischemia monitoring due to its non-invasive nature and accessibility (5). Compared to conventional resting ECG, which captures only brief electrical changes, dynamic ECG enhances the detection of transient ST-T segment abnormalities through continuous 24–72-h recording, making it particularly effective for outpatient populations experiencing paroxysmal chest pain (6). Additionally, its ability to monitor cardiac activity in real time during daily life facilitates the identification of associations between myocardial workload and ischemic episodes, supporting personalized treatment approaches (7).

However, the diagnostic performance of dynamic ECG for myocardial ischemia remains controversial. Several studies have demonstrated substantial variability in sensitivity and specificity, largely attributable to: (1) bias in baseline characteristics of study populations (e.g., age distribution, comorbid conditions such as diabetes); (2) lack of standardized diagnostic thresholds (e.g., differing criteria for ST-segment depression ≥0.1 mV or ≥1 mm, lasting ≥1 min); (3) technical inconsistencies (lead configuration, motion artifact recognition); and (4) divergent reference standards (coronary angiography, myocardial perfusion imaging, or intravascular ultrasound) (8). These methodological inconsistencies hinder clinical interpretation and limit the development of evidence-based guidelines.

To address these limitations and provide a comprehensive assessment of diagnostic performance, this systematic review and meta-analysis aims to synthesize available evidence on dynamic ECG for detecting myocardial ischemic episodes in patients with CHD. By pooling results across studies, we evaluate its diagnostic accuracy, explore sources of heterogeneity, and identify key factors influencing performance.

This systematic review and meta-analysis provides a comprehensive evaluation of dynamic ECG for detecting myocardial ischemia in participants with CHD. The pooled diagnostic estimates—sensitivity, 0.75; specificity, 0.70; PLR, 2.50; NLR, 0.36; DOR, 6.64 and AUC, 0.79—indicate moderate diagnostic utility. While clinically relevant, these values fall short of ideal diagnostic thresholds, highlighting the need for complementary diagnostic tools (40). A PLR of 2.50 suggests that a positive dynamic ECG increases the post-test probability of myocardial ischemia by approximately 30% in moderate-prevalence populations, while an NLR of 0.36 decreases the probability by 40–50%. These findings support the use of dynamic ECG as a triage tool rather than a definitive diagnostic method, consistent with its established role in ambulatory monitoring of transient ischemic episodes (40).

The high heterogeneity (I² > 80% for sensitivity and specificity) is multifactorial, with unreported methodological details emerging as a critical challenge. First, population heterogeneity—including differences in age, sex distribution (46.1–70.8% male), and comorbidities such as diabetes—may have contributed to inconsistent ischemic patterns, as diabetes alters ST-segment morphology through autonomic dysfunction (41). Second, variation in ischemic threshold definitions (e.g., ST-segment depression criteria: ≥0.1 mV vs. ≥1 mm; duration: ≥1 min vs. transient episodes) introduces diagnostic inconsistency, as minor differences in cutoff values can significantly affect sensitivity and specificity trade-offs (42). Third, differences in lead configurations and artifact discrimination algorithms impact signal fidelity, particularly during patient movement—a known limitation of ambulatory monitoring. Finally, use of different reference standards introduces spectrum bias: CAG directly visualizes coronary stenosis, while MPI evaluates functional ischemia, capturing distinct pathophysiological processes.

Subgroup analyses demonstrated enhanced diagnostic performance in confirmed participants with CHD (vs. suspected cases) and CAG-based studies. The superior performance in confirmed CHD cohorts likely reflects greater atherosclerotic burden, allowing dynamic ECG to more reliably detect ischemia-induced repolarization abnormalities (43). In contrast, participants with suspected CHD may exhibit non-ischemic ST-T changes resulting from conditions such as microvascular dysfunction or electrolyte imbalance, reducing specificity. The improved performance in CAG-based studies underscores the value of anatomical correlation, as transient ECG changes may not align with perfusion defects seen in MPI—particularly in cases of balanced multivessel disease (44, 45). However, the absence of significant specificity differences across reference standards indicates ongoing challenges in distinguishing true ischemic events from physiological confounders.

Significant publication bias (Deeks’ test p = 0.02) suggests that smaller studies with less favorable diagnostic performance may have been underreported, a common issue in diagnostic meta-analyses (46). Several factors likely contribute: (1) researchers and journals may be more inclined to publish studies with “positive” findings, while studies with non-significant or lower accuracy are less likely to be submitted or accepted. This is particularly relevant for dynamic ECG research, where institutional or commercial interests in validating diagnostic tools may influence publication trends; (2) our trim-fill analysis imputed five hypothetical missing studies, which slightly reduced pooled metrics but preserved the conclusion of moderate diagnostic utility. This suggests the overestimation due to bias is modest rather than transformative; and (3) while we cannot access unpublished data, we can infer their potential characteristics: smaller sample sizes, higher risk of bias, or populations with lower disease prevalence—factors known to reduce diagnostic metric precision. Inclusion of such studies would likely widen confidence intervals but not negate the core finding that dynamic ECG has clinical utility for ischemia detection.

A notable limitation of this meta-analysis is the overrepresentation of Chinese studies (23/24), which may restrict the extrapolability of conclusions to other populations. Several factors may explain the paucity of international studies meeting our criteria: (1) Clinical practice variations: Dynamic ECG utilization patterns differ globally—while it is widely adopted as a first-line ambulatory monitoring tool in China for CHD patients, international guidelines often prioritize stress testing or coronary CT angiography for ischemia detection, potentially reducing the number of dedicated dynamic ECG diagnostic studies; and (2) Data reporting standards: International guideline focus on dynamic ECG for arrhythmia detection rather than myocardial ischemia, or lack complete diagnostic data required for meta-analysis. Thus, while our findings provide valuable evidence for Chinese clinical practice, extrapolation to other regions should be cautious. Future studies should prioritize multi-center, international collaborations to include diverse ethnicities, healthcare systems, and clinical practice patterns, thereby enhancing the generalizability of dynamic ECG’s diagnostic performance data. Moreover, the lack of standardized reporting of ST-segment depression thresholds and duration criteria in mostly included studies. Threshold variations are well-known to drive sensitivity-specificity trade-offs in diagnostic testing. Without this data, we cannot quantify their contribution to heterogeneity, highlighting a major gap in dynamic ECG research: the absence of consensus on reporting diagnostic criteria.

Despite limitations—including significant geographic bias (23 of 24 studies from China) and methodological heterogeneity—these findings reinforce the utility of dynamic ECG in non-invasive ischemia monitoring. The consistency between the single U. S. study and pooled Chinese results provides preliminary support for generalizability, but future international studies are needed to confirm these findings across diverse populations. Its strength lies in capturing transient episodes during routine activity, offering a preferable alternative to stress testing in older adults and patients with frailty. Clinicians should interpret results within the broader clinical context. Future research should prioritize standardization of ischemic criteria, adoption of advanced signal-processing technologies, and validation across diverse populations. Integrating dynamic ECG with high-sensitivity troponin assays or coronary CT angiography may enhance diagnostic accuracy, particularly in emergency department evaluations for chest pain.