Diagnostic Accuracy of Electrocardiographic Criteria in Differentiating Arrhythmogenic Right Ventricular Cardiomyopathy From Idiopathic Outflow Tract Ventricular Arrhythmia: A Systematic Review and Meta‐Analysis

ABSTRACT

Background

Differentiating arrhythmogenic right ventricular cardiomyopathy (ARVC) from idiopathic outflow tract ventricular arrhythmias (OT VA) remains a challenge due to overlapping electrocardiographic (ECG) features. However, their pathophysiology and prognosis differ. Given the limited accessibility of advanced imaging, ECG parameters could facilitate early diagnosis. This review and meta‐analysis evaluate the diagnostic accuracy of ECG parameters in distinguishing ARVC from OT VA.

Methods

A systematic database search was conducted from inception to March 2025. Observational studies assessing ECG markers for differentiating ARVC from OT VA in adults were included. Diagnostic accuracy metrics (sensitivity, specificity, diagnostic odds ratio, and area under the receiver operating characteristic curve [AUC]) were synthesized using a bivariate random‐effects model. Risk of bias was assessed using the QUADAS‐2 tool.

Results

Ten studies enrolling 723 patients were included. T‐wave inversion in V1–V3 was the most commonly assessed parameter, with a pooled sensitivity of 0.55 and specificity of 0.88 (AUC = 0.798). Epsilon waves demonstrated low sensitivity (0.12) but high specificity (0.98). Among depolarization‐based parameters, QRS duration in V1–V3 (AUC = 0.832) and the ratio of QRS duration in precordial leads (AUC = 0.938) demonstrated strong diagnostic performance. S‐wave upstroke and terminal activation delay also showed high specificity (0.95) but moderate sensitivity.

Conclusion

Several ECG parameters, particularly depolarization‐based parameters, exhibit high specificity and moderate sensitivity in differentiating ARVC from OT VA. These findings support the use of ECG as a reliable and accessible diagnostic tool, particularly in settings without advanced imaging. A multi‐marker ECG approach may enhance early recognition of patients with suspected ARVC.

ECG parameters, particularly depolarization‐based markers, demonstrate high specificity for differentiating ARVC from idiopathic outflow tract arrhythmias. A multi‐marker ECG approach may facilitate accessible early screening and clinical decision‐making, particularly in settings with limited access to advanced imaging.

1. Introduction

Ventricular arrhythmias constitute major global health concerns, contributing substantially to morbidity and sudden cardiac death (SCD). Their spectrum encompasses premature ventricular complexes, sustained ventricular tachycardia (VT), and ventricular fibrillation. Among non‐ischemic etiologies, outflow tract ventricular arrhythmias (OT VA) are relatively frequent and may arise from two clinically and pathophysiologically distinct entities: idiopathic OT arrhythmias and arrhythmogenic right ventricular cardiomyopathy (ARVC). Although these conditions can exhibit similar surface electrocardiogram (ECG) patterns, they differ fundamentally in underlying mechanisms, prognosis, and therapeutic strategies. Misclassification carries significant clinical consequences, including unnecessary implantable cardioverter–defibrillator (ICD) implantation or delayed recognition of a potentially life‐threatening cardiomyopathy [1, 2].

ARVC is a genetically determined cardiomyopathy characterized by the progressive fibrofatty replacement of the right ventricular myocardium, predominantly affecting young and athletic individuals and predisposing them to malignant arrhythmias and SCD. In contrast, idiopathic OT arrhythmias typically occur in structurally normal hearts, are non‐genetic in origin, and usually follow a benign clinical course. Despite these differences, overlapping ECG features—such as left bundle branch block (LBBB) morphology and inferior axis—may obscure accurate clinical diagnosis, particularly in younger patients.

The diagnosis of arrhythmogenic right ventricular cardiomyopathy is established using the 2010 Revised Task Force Criteria, which integrate findings from electrocardiography, cardiac imaging, histopathology, genetic testing, and family history [3]. Because ARVC is a progressive disease, its electrocardiographic manifestations may evolve over time, and ECG findings alone are insufficient to establish a definitive diagnosis. Advanced imaging modalities, particularly cardiac magnetic resonance imaging (MRI), provide important structural and functional assessment and play a central role in confirming ARVC. However, these modalities may be costly and not universally available, especially in resource‐limited settings. In such contexts, ECG remains the most widely accessible, non‐invasive, and cost‐effective tool for screening and early detection, allowing clinicians to raise clinical suspicion and guide further diagnostic evaluation. Importantly, ECG parameters should be interpreted as supportive markers within a multimodal diagnostic framework, with greater utility as rule‐in rather than rule‐out tools. Establishing which ECG parameters offer the strongest discriminatory value could refine current diagnostic algorithms, guide risk stratification for sudden cardiac death, inform guideline updates, and enable earlier recognition of ARVC across diverse clinical environments.

Multiple ECG markers have been proposed to differentiate ARVC from idiopathic OT arrhythmias. Recognized features include epsilon waves in leads V1–V3, prolonged QRS duration (> 110 ms), T‐wave inversions in V2–V3 (in patients over 12 years without right bundle branch block), and VT with an LBBB morphology and superior axis. These reflect perturbations in myocardial depolarization and repolarization caused by underlying structural remodeling. Additional indices, such as QT dispersion and the T _peak–T _end interval, have been investigated as non‐invasive surrogates of arrhythmic risk [4, 5]. Such features are typically absent in idiopathic OT arrhythmias, which occur in otherwise structurally normal hearts.

While prior systematic reviews addressed the prognostic significance of ECG findings in ARVC [6] or their diagnostic value in specific populations with limited access to imaging [7], no meta‐analysis has yet systematically quantified and compared the diagnostic accuracy of individual ECG parameters for distinguishing ARVC from idiopathic OT arrhythmias. This study represents the first meta‐analytic synthesis specifically designed to evaluate ECG‐based diagnostic markers for ARVC, addressing a critical gap in current diagnostic algorithms and guideline criteria.

Accordingly, this systematic review and meta‐analysis aim to synthesize available evidence on the diagnostic performance of ECG parameters in differentiating ARVC from idiopathic OT VA. By quantifying sensitivity, specificity, and overall discriminative value, this study provides a foundation for developing structured multi‐marker ECG approaches or artificial intelligence‐based screening strategies. Ultimately, these findings may facilitate the initial screening process, timely triage, and optimize referral decisions for further diagnostic investigations to establish a definitive diagnosis of ARVC, and improve clinical outcomes for patients presenting with ventricular arrhythmia.

2. Methods

2.1. Study Protocol Registration

This systematic review and meta‐analysis was prospectively registered in the PROSPERO international prospective register of systematic reviews (registration ID: CRD420251040225). The updated 2020 Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) guidelines were used to review diagnostic test accuracy studies [8].

2.2. Literature Search Strategy

A comprehensive and systematic literature search was conducted across three electronic databases (PubMed, Scopus, and Embase) from their inception until March 2025. The aim was to identify studies evaluating ECG parameters for differentiating ARVC from idiopathic OT VA, including arrhythmias originating from the right ventricular outflow tract (RVOT) or left ventricular outflow tract (LVOT).

The search strategy employed a combination of Medical Subject Headings (MeSH) and free‐text keywords such as “arrhythmogenic right ventricular cardiomyopathy,” “ARVC,” “RVOT,” “LVOT,” “idiopathic ventricular arrhythmia,” “diagnosis,” and “electrocardiogram.” Boolean operators (AND, OR) were applied to optimize sensitivity and specificity, and the search syntax was tailored to each database's indexing system. The complete search syntax is provided in Appendix.

Two independent reviewers (E.M. and N.M.R.) initially screened titles and abstracts, followed by a full‐text review of potentially eligible studies. Any disagreements were resolved through discussion and consensus; if consensus could not be reached, a third reviewer was consulted for arbitration. In addition, the reference lists of all included articles and relevant reviews were manually searched to identify further eligible studies.

2.3. Eligibility Criteria

Inclusion criteria were defined a priori to ensure methodological rigor and transparency. Studies were eligible for inclusion if they met the following conditions: they were original, peer‐reviewed observational studies (cross‐sectional, case–control, or cohort designs); enrolled adult participants (≥ 18 years) presenting with VA suspected to represent either ARVC or idiopathic OT VA; employed ECG parameters as the primary index test to differentiate ARVC from idiopathic OT VA; and used a validated reference standard for ARVC diagnosis (e.g., the Revised TFC or advanced imaging modalities) while providing sufficient diagnostic accuracy data to allow construction of 2 × 2 contingency tables.

Studies were excluded if they met any of the following criteria: they were case reports, case series, editorials, expert commentaries, review articles (including systematic reviews and meta‐analyses), or conference abstracts; they comprised unpublished manuscripts, preprints, or other forms of gray literature; or they lacked extractable ECG data or did not clearly distinguish diagnostic outcomes between ARVC and OT VA groups.

All retrieved records were imported into Rayyan platform (https://www.rayyan.ai) for deduplication and screening. To assess eligibility, two reviewers independently screened titles, abstracts, and full texts. Discrepancies were resolved by discussion and consensus, with involvement of a third reviewer when necessary. The PRISMA flow diagram details the complete study selection process (Figure 1).

FIGURE 1.

PRISMA 2020 flow diagram for the study selection process This flow diagram illustrates the identification, screening, eligibility, and inclusion of studies in this systematic review and meta‐analysis. A total of 10 studies were included in the qualitative synthesis and quantitative meta‐analysis. ARVC = arrhythmogenic right ventricular cardiomyopathy; ECG = electrocardiogram; PRISMA = Preferred Reporting Items for Systematic Reviews and Meta‐Analyses.

2.4. Data Extraction

Two reviewers independently extracted data using a standardized Google Sheets form to ensure consistency and accuracy. Extracted information included key study characteristics (first author's last name, year of publication, study location, and study design), as well as patient demographics (total sample size, age distribution, and sex proportions). Diagnostic criteria used to confirm ARVC and OT VA were documented, together with the specific ECG parameters evaluated in each study.

For the assessment of diagnostic performance, data on true positives, false positives, true negatives, and false negatives were extracted. Additionally, calculated measures such as sensitivity, specificity, positive predictive value, negative predictive value, and their corresponding confidence intervals (CI) were recorded. When critical data were missing or unclear, attempts were made to contact the corresponding authors for clarification. If no response was obtained and the missing data were essential for quantitative synthesis, the study was excluded from the meta‐analysis. Any discrepancies between reviewers were resolved through discussion and consensus, with adjudication by a senior investigator when necessary.

2.5. Risk of Bias Assessment

The methodological quality of each included study was evaluated using the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS‐2) tool, a validated framework for appraising the risk of bias and concerns regarding applicability in diagnostic accuracy research. This tool systematically assesses four domains: Patient Selection, Index Test, Reference Standard, and Flow and Timing. Each domain is evaluated for potential sources of bias and concerns about the study's applicability to the review question [9].

Two reviewers independently performed the QUADAS‐2 assessment for each study. Discrepancies were resolved through discussion and consensus; a third reviewer provided adjudication if consensus could not be achieved. To ensure the robustness of the meta‐analytic findings, sensitivity analyses were conducted by excluding studies assessed as having a high risk of bias in one or more QUADAS‐2 domains.

2.6. Definitions of ECG Parameters and Clinical Outcomes

In this review, ECG parameters were categorized into two groups: depolarization markers (QRS duration in V1–V3, epsilon waves, ratio of sum of QRS duration in precordial leads, difference of sum of QRS duration in precordial leads, terminal activation delay, and S‐wave upstroke), repolarization markers (T‐wave inversion in V1–V3, T_peak–T_end interval, JT interval dispersion, JT interval ratio, and QT interval and dispersion).

The diagnostic accuracy of each parameter was assessed using standard metrics, including sensitivity, specificity, and diagnostic odds ratio (DOR), calculated from 2 × 2 contingency tables when available. Additionally, relevant clinical outcomes such as sustained or non‐sustained VT, appropriate ICD therapies, and SCD were documented if reported. SCD was defined as an unexpected death occurring within 1 h of symptom onset, while ICD therapies were deemed appropriate if delivered in response to documented ventricular arrhythmias. Unless otherwise specified, all ECG recordings were assumed to follow standard 12‐lead configurations (10 mm/mV, 25 mm/s paper speed).

2.7. Statistical Analysis

Meta‐analyses were performed using R software (version 4.4.2). Pooled estimates of sensitivity and specificity were calculated using a bivariate random‐effects model, which accounts for both within‐study and between‐study variability. To provide a summary measure of diagnostic accuracy, diagnostic odds ratios were calculated for each study using DerSimonian‐Laird method and log‐transformed. Pooled logDOR values were estimated using a random‐effects model, as logarithmic transformation stabilizes variance and reduces skewness. Diagnostic performance was visualized using forest plots and summary receiver operating characteristic (SROC) curves. Heterogeneity was assessed using the I ² and τ ² statistics, with I ² values greater than 50% indicating substantial heterogeneity [10].

3. Results

3.1. Study Selection

A total of 382 records were identified through database searches: PubMed (n = 97), Embase (n = 197), and Scopus (n = 88). After removing 157 duplicates, 225 unique articles remained for title and abstract screening. Of these, 198 were excluded due to irrelevant title or abstract, leaving 27 articles for full‐text assessment. Four full texts could not be retrieved due to access limitations. Efforts were made to obtain inaccessible full‐text articles by contacting the corresponding authors. One study was successfully retrieved; however, it did not meet the predefined inclusion criteria upon full‐text assessment and was therefore excluded. The remaining articles could not be obtained despite these attempts. Among the 23 accessible full‐text articles, 13 were excluded based on the following reasons: single‐arm study design (n = 6), inclusion of healthy controls instead of comparative disease groups (n = 3), analysis limited to ARVC family members (n = 1), or ECG parameters evaluated only during ventricular ectopic activity (n = 3). Ultimately, 10 studies met all inclusion criteria and were included in both qualitative synthesis and quantitative meta‐analysis. Figure 1 presents the PRISMA 2020 flow diagram.

3.2. Study Characteristics

The 10 included studies comprised a total of 723 participants, representing both ARVC and idiopathic OT VA populations. Most studies were retrospective (8/10), with two prospective in design. Sample sizes ranged from 23 to 200 participants. Geographically, the studies originated from diverse regions—including Europe, North America, and Asia—indicating a broad representation of clinical settings. All ECG parameters included in this meta‐analysis were derived from standard 12‐lead ECG recordings during sinus rhythm. Studies focusing primarily on ventricular ectopy or ventricular tachycardia were not included to ensure methodological consistency and comparability (Table 1).

TABLE 1.

Baseline characteristics of the included studies.

Study (Author, Year)	Country	Study design	Sample size (ARVC/OT VA)	Age (Mean ± SD or Range) (ARVC/OT VA)	% Male (ARVC/OT VA)	ECG type and rhythm context	History of syncope (%) (ARVC/OT VA)	History of NSVT (%) (ARVC/OT VA)	History of sustained VT (%) (ARVC/OT VA)	History of SCD (%) (ARVC/OT VA)	ICD (%) (ARVC/OT VA)	RV wall motion abnormalities (by Echo/MRI) (%) (ARVC/OT VA)	Gene mutations (%) (ARVC/OT VA)	Diagnostic criteria of ARVC
Peters et al. [11]	Germany	R	42/25	NR	NR	12‐lead ECG, Sinus rhythm	NR	33/40	62/60	4/0	NR	NR	NR	TFC
Nasir et al. [12]	USA	R	39/28	38 ± 15	54/NR	12‐lead ECG, Sinus rhythm	17/NR	28/NR	42/NR	NR	NR	56/NR	NR	TFC
Cox et al. [13]	Netherlands	R	42/27	35 ± 13/52 ± 15	76/37	12‐lead ECG, Sinus rhythm	NR	100/100 a		NR	NR	NR	60/0 (PKP2)	TFC
Morin et al. [14]	USA	R	79/121	40 ± 14/48 ± 14	58/41	12‐lead ECG, Sinus rhythm	NR	100/100 a		NR	NR	NR	NR	TFC
Hoffmayer et al. [15]	USA	P	37/49	38.3 ± 2.4/46.6 ± 2.2	51/49	12‐lead ECG, Sinus rhythm	NR	59/8 a		NR	NR	NR	NR	TFC
Samol et al. [16]	Germany	P	10/13	45 ± 12/42 ± 13	100/31	12‐lead ECG, Sinus rhythm	NR	60/0 a		NR	60/0	NR	50/0 (PKP2)	TFC
Golcuk et al. [17]	Turkey	R	25/13	42 ± 16/45 ± 14	64/53	12‐lead ECG, Sinus rhythm	60/0	100/100 a		12/0	68/0	NR	NR	TFC
Emkanjoo et al. [18]	Iran	R	29/30	41.7 ± 14.4/42.7 ± 12.7	70/23	12‐lead ECG, Sinus rhythm	48/33	NR	56/13	37/7	NR	NR	NR	TFC
Novak et al. [19]	UK	R	20/33	43 ± 21/39 ± 11	70/70	12‐lead ECG, Sinus rhythm	25/24	NR	NR	NR	NR	65/0	100/0 (Desmosome)	TFC
Chung et al. [20]	Taiwan	R	31/30	44.7 ± 13.9/43.6 ± 15.8	42/3	12‐lead ECG, Sinus rhythm	19/20	19/17	52/17	10/3	NR	16/0	NR	TFC

Note: This table presents the baseline characteristics from the 10 studies included in the review and meta‐analysis. It summarizes the study design, sample sizes for ARVC and OT VA, demographic features of the subjects, imaging findings, and genetic mutations.

Abbreviations: ARVC = arrhythmogenic right ventricular cardiomyopathy; ECG = electrocardiogram; ICD = implantable cardioverter‐defibrillator; MRI = magnetic resonance imaging; NR = not reported; NSVT = non‐sustained ventricular tachycardia; OT VA = outflow tract ventricular arrhythmia; P = prospective; PKP2 = plakophilin‐2; R = retrospective; RV = right ventricle; SCD = sudden cardiac death; SD = standard deviation; TFC = Task Force Criteria; VT = ventricular tachycardia.

^a The type of VT (sustained or non‐sustained) was not specified.

Participants were generally middle‐aged, with mean ages clustered in the fourth decade of life. A male predominance was noted in ARVC cohorts, with several studies reporting > 60% male subjects. Clinical histories varied but frequently included sustained VT, particularly in ARVC patients, with reported rates exceeding 60% in multiple cohorts. Reports of syncope, non‐sustained VT, and SCD were inconsistent, though several studies described notable rates of SCD and ICD implantation in high‐risk groups.

Right ventricular structural abnormalities were reported using echocardiography or cardiovascular magnetic resonance. Genetic testing, when available, most frequently identified pathogenic variants in PKP2 and other desmosomal genes. Notably, Samol et al. Cox et al., and Novak et al. reported genetic mutation positivity rates between 50% and 100% among ARVC patients tested. All studies applied the 2010 Revised Task Force Criteria for ARVC diagnosis. Table 1 presents the baseline characteristics of the included studies.

ECG parameters assessed varied across studies and included both established and exploratory indices, such as T‐wave inversion (TWI) in V1–V3, epsilon waves, QRS duration, S‐wave upstroke, and terminal activation delay (TAD). Some studies, including those by Emkanjoo et al. and Chung et al. extended their analysis to repolarization indices such as the Tp–Te interval, PR/QTc intervals, and JT dispersion. A detailed list of the ECG parameters assessed in each study is presented in the Appendix.

This heterogeneity in study design, population characteristics, and ECG parameters highlights the complexity of electrocardiographic diagnosis in ARVC and the rationale for integrative, multimodal analysis.

3.3. ECG Parameters Evaluated

Across the included studies, various ECG parameters were evaluated for their ability to differentiate ARVC from idiopathic OT VA. To enable a structured and clinically meaningful synthesis, the electrocardiographic findings reported across the included studies were grouped into 11 predefined parameter categories. This grouping was based on physiological relevance, frequency of reporting, and their association with distinct electrophysiological mechanisms. The categories included: TWI in V1–V3, epsilon waves, QRS duration in V1–V3, ratio of sum of QRS duration in precordial leads, S‐wave upstroke duration, TAD, Tp–Te interval, difference of sum of QRS duration in precordial leads, JT interval dispersion, JT interval ratio in right precordial/left precordial, and QT interval and dispersion. These groupings allowed preservation of parameter specificity while ensuring sufficient data for meta‐analysis. Importantly, the categorization also reflects three principal domains of ARVC‐related electrical abnormalities—namely, depolarization delay, repolarization disturbances, and intraventricular conduction heterogeneity—which are central to the pathophysiological distinction between ARVC and idiopathic OT arrhythmias. Table 2 summarizes the grouping of ECG criteria and the number of studies assessing each parameter.

TABLE 2.

ECG parameters investigated across included studies to distinguish ARVC from OT VA.

No	ECG parameters	Number of studies	Reported in
1	T‐wave inversion V1–V3	10	All 10 studies
2	Epsilon wave	6	Peters et al. [11], Nasir et al. [12], Cox et al. [13], Hoffmayer et al. [15], Samol et al. [16], Golcuk et al. [17]
3	QRS duration in V1–V3	5	Peters et al. [11], Nasir et al. [12], Cox et al. [13], Golcuk et al. [17], Emkanjoo et al. [18]
4	Ratio of sum of QRS duration in precordial leads	4	Peters et al. [11], Nasir et al. [12], Cox et al. [13], Emkanjoo et al. [18]
5	S‐wave upstroke	4	Nasir et al. [12], Cox et al. [13], Emkanjoo et al. [18], Novak et al. [19]
6	Terminal activation delay	2	Cox et al. [13], Golcuk et al. [17]
7	Precordial Tp–Te interval	1	Golcuk et al. [17]
8	Difference of sum of QRS duration in precordial leads	1	Emkanjoo et al. [18]
9	JT interval dispersion	1	Emkanjoo et al. [18]
10	JT interval ratio in right precordial/left precordial	1	Emkanjoo et al. [18]
11	QT interval and dispersion	1	Nasir et al. [12]

Note: This table summarizes the electrocardiographic parameters evaluated across the 10 included studies, the number of studies assessing the parameters, and the specific studies in which they were evaluated. T‐wave inversion in leads V1‐V3 was the most commonly evaluated parameter, assessed in all 10 studies. Epsilon waves and QRS duration in leads V1‐V3 were assessed in 5 and 6 studies, respectively. Difference of sum of QRS duration in precordial leads, JT interval dispersion, ratio JT interval, and QT interval and dispersion were only assessed in 1 study. Tp–Te = T_peak–T_end.

Abbreviations: ARVC = arrhythmogenic right ventricular cardiomyopathy; ECG = electrocardiogram; OT VA = outflow tract ventricular arrhythmia.

The most frequently assessed parameter was TWI in V1–V3 in all 10 studies, followed by epsilon waves in six studies and prolonged QRS duration in right precordial leads in five studies. The ratio of sum of QRS duration in precordial leads and S‐wave upstroke was each evaluated in four studies, indicating growing interest in subtle conduction delays. Parameters reflecting repolarization abnormalities, such as TAD and precordial Tp–Te interval, were each assessed in two studies. Less commonly reported indices included the difference of sum of QRS duration in precordial leads, JT interval dispersion, JT interval ratio, and QT interval and dispersion, each of which was evaluated in only one study. The variation in ECG marker selection underscores ongoing efforts to refine non‐invasive diagnostic strategies for distinguishing ARVC from OT VA.

3.4. Quality of Study

3.4.1. Risk of Bias

Risk of bias (RoB) was evaluated using the QUADAS‐2 tool. Overall, the majority of studies demonstrated a low risk of bias across all four domains. The Flow and Timing domain had the best performance, with all 10 studies (100%) rated as low risk. However, some concerns were noted in other domains: Reference Standard showed the highest proportion of unclear risk (5/10 studies), followed by Patient Selection (3/10) and Index Test (2/10). These unclear ratings were mainly due to insufficient information regarding blinding procedures or selection criteria. Importantly, no high‐risk judgments were observed in any domain. In terms of applicability, all studies were judged to have low concerns across all three domains (Patient Selection, Index Test, and Reference Standard) indicating that the included studies were highly relevant to the review question. The figure illustrating the RoB is presented in Appendix.

3.4.2. Sensitivity Analysis

A leave‐one‐out sensitivity analysis was conducted to assess the robustness of the pooled diagnostic performance. Each study was sequentially excluded from the meta‐analysis, and the corresponding pooled log odds ratio (OR) was recalculated. The results demonstrated consistent effect sizes, with log OR estimates ranging from 2.16 (95% CI: 1.10–3.21) to 2.73 (95% CI: 1.68–3.77). All recalculated models remained statistically significant (p < 0.001), and no single study exerted a disproportionate influence on the overall diagnostic effect. These findings confirm the stability and robustness of the pooled estimates. The figure illustrating the leave‐one‐out sensitivity analysis is presented in Appendix.

3.5. Subgroup Analyses

3.5.1. T‐Wave Inversion in Leads V1–V3

A total of 10 studies evaluated the diagnostic performance of TWI in V1–V3 to differentiate ARVC from OT VA. Sensitivity estimates varied considerably, ranging from 0.20 to 0.90, while specificity remained consistently high in nine studies, between 0.77 and 1.00. Using a bivariate random‐effects model, the pooled sensitivity was 0.55 (95% CI: 0.37–0.72) (Figure 2) and specificity was 0.88 (95% CI: 0.76–0.95) (Figure 3), indicating that TWI provides moderate discriminative power.

FIGURE 2.

Forest plots of pooled sensitivity for the evaluated ECG parameters. The panels show forest plots of pooled sensitivity for: (A) TWI, (B) epsilon wave, (C) QRS duration in leads V1–V3, (D) the ratio of the sum of QRS duration in precordial leads, (E) S‐wave upstroke, and (F) terminal activation delay in distinguishing ARVC from OT VA. Each forest plot displays sensitivity estimates for each study, along with overall pooled sensitivity evaluated using the random‐effects meta‐analysis. Each square shows the point estimate for an individual study, with the horizontal bars depicting its 95% CI. The pooled estimates appear as diamonds at the bottom of each panel, whose width reflects the corresponding 95% CI. ARVC = arrhythmogenic right ventricular cardiomyopathy; CI = confidence interval; ECG = electrocardiogram; OT VA = outflow tract ventricular arrhythmia; TWI = T‐wave inversion.

FIGURE 3.

Forest plots of pooled specificity for the evaluated ECG parameters. The panels show forest plots of pooled specificity for: (A) TWI, (B) epsilon wave, (C) QRS duration in leads V1–V3, (D) the ratio of the sum of QRS duration in precordial leads, (E) S‐wave upstroke, and (F) terminal activation delay in distinguishing ARVC from OT VA. Each forest plot displays specificity estimates for each study, along with overall pooled specificity evaluated using the random‐effects meta‐analysis. Each square shows the point estimate for an individual study, with the horizontal bars depicting its 95% CI. The pooled estimates appear as diamonds at the bottom of each panel, whose width reflects the corresponding 95% CI. ARVC = arrhythmogenic right ventricular cardiomyopathy; CI = confidence interval; ECG = electrocardiogram; OT VA = outflow tract ventricular arrhythmia; TWI = T‐wave inversion.

The diagnostic odds ratio of TWI was 9.54 (95% CI: 3.41–26.66), as presented in Figure 4, and the log DOR was 2.26 (95% CI: 1.23–3.28), suggesting TWI's meaningful overall accuracy in distinguishing ARVC from OT VA. Between‐study heterogeneity of DOR was modest (I ² = 9.96%, τ ² = 1.96), and no significant threshold effect was observed (ρ = 0.025), as presented in Table 3. The area under the SROC curve (AUC) was 0.798 (Figure 5), and the normalized partial AUC was 0.707, reflecting good model fit. These findings indicate that TWI is a highly specific and moderately sensitive ECG marker for ARVC.

FIGURE 4.

Forest plots of pooled diagnostic odds ratios for the evaluated ECG parameters The panels show forest plots of pooled DORs for: (A) TWI, (B) epsilon wave, (C) QRS duration in leads V1–V3, (D) the ratio of the sum of QRS duration in precordial leads, (E) S‐wave upstroke, and (F) terminal activation delay. Each forest plot displays the study‐specific DOR for the respective ECG parameter across all included studies, as well as the pooled DOR estimated using the DSL method. Squares represent individual study DOR, with horizontal lines indicating the corresponding 95% CI; square size is proportional to the study's weight in the meta‐analysis. The blue diamond represents the pooled DOR, and its width indicates the 95% CI. CI = confidence interval; DOR = diagnostic odds ratio; DSL = DerSimonian‐Laird; ECG = electrocardiogram; TWI = T‐wave inversion.

TABLE 3.

Summary results of various ECG parameter in differentiating ARVC from idiopathic OT VA.

ECG parameter	Number of studies	DOR (95% CI)	LogDOR (95% CI)	Heterogeneity	Sensitivity (95% CI)	Specificity (95% CI)	AUC (95% CI)
T‐wave inversion in leads V1–V3	10	9.54 (3.41–26.66)	2.26 (1.23–3.28)	I 2 = 9.96%	0.55 (0.37–0.72)	0.88 (0.76–0.95)	0.798 (0.700–0.896)
Epsilon wave	6	7.05 (2.03–24.49)	1.95 (0.71–3.20)	I 2 = 0%	0.12 (0.05–0.26)	0.98 (0.94–0.99)	0.972 (0.920–0.990)
QRS duration in leads V1–V3	5	23.48 (7.22–76.38)	3.16 (1.98–4.34)	I 2: 0%	0.59 (0.41–0.75)	0.95 (0.80–0.99)	0.832 (0.770–0.890)
Ratio of sum of QRS duration in precordial leads	4	38.83 (14.93–101.03)	3.66 (2.70–4.62)	I 2: 0%	0.70 (0.40–0.89)	0.93 (0.86–0.97)	0.938 (0.860–0.980)
S‐wave upstroke	4	24.89 (3.74–165.60)	3.21 (1.32–5.11)	I 2: 0%	0.54 (0.16–0.87)	0.95 (0.88–0.98)	0.945 (0.860–0.990)
Terminal activation delay	2	27.51 (4.86–155.72)	3.31 (1.58–5.05)	I 2: 0%	0.61 (0.37–0.80)	0.95 (0.81–0.99)	0.922 (0.800–0.980)

Note: This table presents the pooled diagnostic accuracy measures for each ECG parameters to distinguish ARVC from idiopathic OT VA across the included studies. Higher DOR and AUC values indicate better discriminatory performances. Heterogeneity is express as I².

Abbreviations: ARVC = arrhythmogenic right ventricular cardiomyopathy; AUC = area under the receiver operating characteristic curve; CI = confidence interval; DOR = diagnostic odds ratio; ECG = electrocardiogram; OT VA = outflow tract ventricular arrhythmia.

FIGURE 5.

SROC curves for the evaluated ECG parameters. The panels show SROC curves for: (A) TWI, (B) epsilon wave, (C) QRS duration in leads V1–V3, (D) the ratio of the sum of QRS duration in precordial leads, (E) S‐wave upstroke, and (F) terminal activation delay. Each triangle represents the sensitivity and false‐positive rate from an individual study. The solid line shows the fitted SROC curve derived from the bivariate random‐effects model. The open circle denotes the summary estimate of sensitivity and false‐positive rate, and the surrounding ellipse represents its 95% confidence region. ECG = electrocardiogram; SROC = Summary receiver operating characteristic; TWI = T‐wave inversion.

3.5.2. Epsilon Wave

Six studies investigated the diagnostic performance of the epsilon wave in differentiating ARVC from OT VA. Sensitivity across studies was uniformly low, ranging from 0.02 to 0.33, with a pooled estimate of 0.12 (95% CI: 0.05–0.26) based on the bivariate random‐effects model, as presented in Figure 2. In contrast, specificity was consistently high across all included studies, with a pooled value of 0.98 (95% CI: 0.94–0.99), as presented in Figure 3.

The pooled diagnostic odds ratio, estimated using the DerSimonian–Laird method, was 7.05 (95% CI: 2.03–24.49), reflecting moderate diagnostic utility (Figure 4). There was no detectable between‐study heterogeneity in DOR (τ ² = 0.00; I ² = 0%) (Table 3), indicating that the study‐specific estimates were highly consistent. The SROC curve demonstrated excellent overall performance, with an AUC of 0.972 (Figure 5). Despite its low sensitivity, the epsilon wave remains a highly specific electrocardiographic marker in the diagnosis of ARVC.

3.5.3. QRS Duration in Leads V1–V3

Five studies evaluated the diagnostic accuracy of QRS duration in V1–V3 for distinguishing ARVC from OT VA. Sensitivity estimates ranged from 0.26 to 0.72, while specificity remained high across studies, ranging from 0.77 to 1.00. The bivariate random‐effects model yielded a pooled sensitivity of 0.59 (95% CI: 0.41–0.75) (Figure 2) and pooled specificity of 0.95 (95% CI: 0.80–0.99) (Figure 3).

Diagnostic odds ratios varied widely, from 7.93 to 111.44, with a pooled DOR of 23.48 (95% CI: 7.22–76.38), indicating strong overall discriminative power (Figure 4). There was no significant between‐study heterogeneity for DOR (I ² = 0%, τ ² = 0.47), as presented in Table 3. The SROC curve demonstrated good model fit, with an AUC of 0.832 and a normalized partial AUC of 0.661 (Figure 5). These findings indicate that prolonged QRS duration in V1–V3 is a moderately sensitive yet highly specific marker for identifying ARVC.

3.5.4. The Ratio of the Sum of QRS Duration in Precordial Leads

Four studies assessed the diagnostic utility of the ratio of the sum of QRS durations in precordial leads for differentiating ARVC from OT VA. Sensitivity varied considerably, ranging from 0.36 to 0.93, while specificity was consistently high between 0.93 and 1.00. The bivariate random‐effects model analysis yielded a pooled sensitivity of 0.70 (95% CI: 0.40–0.89), as follows: as presented in Figure 2, and a pooled specificity of 0.93 (95% CI: 0.86–0.97), as presented in Figure 3.

The pooled diagnostic odds ratio, calculated using the DerSimonian–Laird method, was 38.83 (95% CI: 14.93–101.03), suggesting excellent discriminatory capacity (Figure 4). No substantial heterogeneity was observed for DOR (I ² = 0%, τ ² = 0.12) (Table 3). The SROC curve demonstrated strong overall accuracy, with an AUC of 0.938 (Figure 5). These findings suggest that the ratio of the sum of QRS duration in precordial lead is a highly specific marker for ARVC diagnosis.

3.5.5. S‐Wave Upstroke

Four studies evaluated the diagnostic utility of S‐wave upstroke duration in distinguishing ARVC from OT VA. Sensitivity estimates varied markedly, ranging from 0.20 to 0.95, with a pooled sensitivity of 0.54. In contrast, specificity was consistently high across studies, ranging from 0.94 to 1.00. The bivariate random‐effects model analysis yielded a pooled sensitivity of 0.54 (95% CI: 0.16–0.87) (Figure 2) and a pooled specificity of 0.95 (95% CI: 0.88–0.98) (Figure 3).

The pooled diagnostic odds ratio, based on the DerSimonian–Laird model, was 24.89 (95% CI: 3.74–165.60), reflecting strong overall diagnostic capacity (Figure 4). No substantial heterogeneity was observed for DOR (I ² = 0%, τ ² = 2.61), as presented in Table 3. The SROC curve showed an AUC of 0.945 (Figure 5), indicating excellent overall diagnostic performance. These findings suggest that S‐wave upstroke is a highly specific marker with variable sensitivity.

3.5.6. Terminal Activation Delay

Two studies assessed the diagnostic value of TAD for differentiating ARVC from OT VA. Sensitivity ranged from 0.48 to 0.71, while specificity was consistently high, between 0.92 and 0.96. The bivariate random‐effects model yielded a pooled sensitivity of 0.61 (95% CI: 0.37–0.80) (Figure 2) and a pooled specificity of 0.95 (95% CI: 0.81–0.99) (Figure 3).

The pooled diagnostic odds ratio was 27.51 (95% CI: 4.86–155.72), reflecting strong overall accuracy (Figure 4). No substantial heterogeneity was observed for DOR (I ² = 0%, τ ² = 0.37) (Table 3). The AUC was 0.922 (Figure 4), suggesting excellent discriminative performance. These findings suggest that TAD is a highly specific parameter with moderate sensitivity.

4. Discussion

Differentiating ARVC from idiopathic OT VA remains a major clinical challenge due to overlapping electrocardiographic manifestations. Both entities commonly present with LBBB morphology and inferior axis orientation, features that obscure diagnostic clarity and often delay appropriate risk stratification. Early or concealed ARVC, in particular, may mimic the benign phenotype of OT VA, leading to frequent misclassification and underestimation of arrhythmic risk [2]. Such diagnostic ambiguity carries significant consequences, as ARVC is associated with a malignant prognosis and sudden cardiac death, in contrast to the generally favorable course of OT VA.

Current diagnostic frameworks, including the 2010 TFC and its revisions, emphasize multimodal assessment, yet their overall sensitivity remains imperfect. Importantly, not all criteria contribute equally to diagnostic accuracy; ECG and arrhythmic features consistently demonstrate superior discriminative value compared with imaging or family history [21]. These observations reinforce the enduring relevance of ECG as a frontline diagnostic tool—particularly in early disease stages, borderline cases, or in settings where advanced imaging and genetic testing are limited. Thus, systematic evaluation of ECG markers represents a crucial step toward improving early recognition and management of ARVC.

While ECG offers a practical and widely accessible screening modality, it should not be interpreted as a standalone diagnostic tool. The progressive nature of ARVC and variability in ECG manifestations across disease stages limit its diagnostic completeness. Moreover, reliance on ECG alone may lead to false‐positive findings and unnecessary downstream interventions. Therefore, ECG‐based findings should be integrated with imaging modalities, particularly cardiac MRI, to establish an accurate diagnosis and guide optimal management.

Among repolarization abnormalities, TWI in V1–V3 consistently emerges as the most clinically relevant ECG marker in distinguishing ARVC from RVOT‐VA. Our pooled analysis demonstrated moderate sensitivity (0.55) but high specificity (0.88), with an AUC of 0.798, underscoring its role as a confirmatory rather than screening tool. Particularly, TWI extending beyond lead V2 is exceptionally uncommon in idiopathic OT VA or healthy individuals, and strongly suggests right ventricular free wall involvement characteristic of ARVC [14, 22]. In practice, such patterns should serve as a red flag, prompting escalation to advanced imaging or genetic testing in otherwise equivocal cases. TWI's limited sensitivity necessitates integration with depolarization indices and multimodal assessment for robust case identification. According to the 2010 Revised Task Force Criteria, T‐wave inversion in leads V1–V3 is considered a major repolarization abnormality in the absence of complete right bundle branch block (RBBB), and a minor criterion when present with RBBB. However, most included studies did not consistently report RBBB status, preventing stratified analysis.

The wide variability in sensitivity of TWI (0.20–0.90) reported across studies likely reflects differences in disease stage, patient age, and study population characteristics. TWI is more frequently observed in overt or advanced ARVC, where progressive fibrofatty replacement of the right ventricular myocardium leads to repolarization abnormalities [12]. Conversely, TWI may be absent in early or concealed stages of the disease, which may explain the lower sensitivity observed in some cohorts. Age‐related physiological TWI patterns, particularly in younger individuals and athletes, may further influence ECG interpretation, especially in the right precordial leads [23]. In addition, heterogeneity in study design, diagnostic criteria, and ECG interpretation methods may also contribute to the wide range of TWI sensitivities reported in the literature.

Epsilon waves, a depolarization abnormality long regarded as pathognomonic for ARVC, similarly displayed very high specificity (0.98) but extremely poor sensitivity (0.12). Therefore, these findings support the role of ECG markers as rule‐in tools for ARVC. Their presence may substantially increase the likelihood of the disease, whereas their absence cannot reliably exclude ARVC, particularly in early or concealed stages. Their rarity in early‐stage disease and susceptibility to technical limitations, such as filtering or lead placement, reduce their reliability in routine screening. Modified techniques, including high precordial lead placement, may improve detection [24], but their absence cannot rule out ARVC.

Apart from epsilon waves, depolarization abnormalities demonstrated overall stronger and more consistent discriminatory performance compared with repolarization abnormalities. In our meta‐analysis, QRS duration in V1–V3 achieved an AUC of 0.832 with a DOR of 23.48, while the ratio of sum of QRS duration in precordial leads demonstrated excellent diagnostic performance (AUC 0.938, DOR 38.83). These indices reflect conduction slowing due to fibrofatty myocardial replacement, a hallmark of ARVC pathology that is absent in idiopathic RVOT‐VA [12, 18]. Additional depolarization parameters, including S‐wave upstroke and TAD, also showed high specificity (> 94%) with moderate sensitivity. Notably, the absence of statistical heterogeneity across studies (I ² = 0%) underscores the robustness and external validity of these markers. This consistency suggests that depolarization indices may represent useful ECG markers for ARVC across different populations and recording conditions, particularly in supporting rule‐in assessment. However, their limited sensitivity necessitates interpretation within a comprehensive, multi‐modal diagnostic framework.

From a clinical perspective, the strong performance of depolarization markers highlights their potential as a cornerstone of diagnostic evaluation. Unlike repolarization features that may be age‐ or sex‐dependent, conduction abnormalities appear tightly linked to the structural substrate of ARVC, thereby providing more stable discriminative power. Systematic incorporation of these markers into routine ECG interpretation may therefore substantially enhance early recognition and risk stratification in patients with suspected ARVC.

Our findings highlight an important sensitivity–specificity trade‐off in the interpretation of several ECG markers for ARVC, particularly epsilon waves and prolonged QRS duration in the right precordial leads. These ECG markers demonstrated very high specificity but relatively low sensitivity, indicating that they function primarily as rule‐in markers rather than rule‐out markers [25]. In clinical practice, the presence of such findings strongly supports the diagnosis of ARVC, whereas their absence cannot reliably exclude the disease. This limitation is particularly relevant in early or concealed stages of ARVC, where structural and electrical abnormalities may still be subtle and incompletely expressed on ECG. Accordingly, these individual markers should not be interpreted in isolation, but rather in combination with other ECG markers to improve the diagnostic accuracy.

Building on the diagnostic utility of individual ECG markers, several studies have proposed composite scoring systems to systematically differentiate ARVC from idiopathic RVOT‐VA. Novak et al. [19] highlighted QS morphology in V1, intrinsicoid deflection time > 80 ms, and negative QRS in lead I as predictive features, yielding an AUC of 0.84. Similarly, Hoffmayer et al. [15] developed a practical scoring system incorporating anterior TWI, QRS duration ≥ 120 ms in lead I, QRS notching in multiple leads, and delayed precordial transition, achieving excellent diagnostic accuracy with 84% sensitivity and 100% specificity. Ren et al. [26] further corroborated these findings, emphasizing prolonged QRS duration in lead I, lateral QRS notching, and delayed R‐wave transition as robust discriminators.

Based on the diagnostic performance observed in this meta‐analysis, a stepwise interpretation of ECG findings may help guide the evaluation of suspected ARVC. Highly specific depolarization abnormalities (epsilon waves, QRS duration in leads V1–V3, the ratio of the sum of QRS durations, S‐wave upstroke, and TAD) demonstrated strong rule‐in value and therefore provide substantial diagnostic confidence when present. However, because these markers exhibit only moderate or low sensitivity, their absence does not exclude ARVC. In such cases, clinicians should further evaluate repolarization abnormalities such as T‐wave inversion in the right precordial leads and consider integrating multiple ECG features to improve overall detection. This multi‐marker approach may be particularly important for identifying early or concealed forms of ARVC before overt structural abnormalities become evident.

Collectively, these frameworks highlight the value of structured, multi‐marker ECG analysis in overcoming the limitations of single parameters. By weighting both repolarization and depolarization features, composite scores can enhance diagnostic precision, reduce misclassification, and facilitate early identification of patients requiring further testing or intervention. Our findings strengthen the rationale for developing a standardized ECG‐based risk score, which could serve as a cost‐effective triage tool, particularly in low‐resource settings where access to advanced imaging and genetic testing remains limited.

Our findings highlight potential refinements to current diagnostic frameworks. While the 2010 TFC and 2020 Padua Criteria [27] emphasize repolarization markers and integrate imaging/genetics, they omit high‐performing depolarization indices such as the ratio of sum of QRS duration in precordial leads, TAD, and QRS duration in V1–V3. These parameters in our analysis consistently showed strong specificity and diagnostic value, suggesting they could enhance early ARVC detection, especially in the concealed phase. Given its accessibility and reproducibility, ECG should not remain a supportive tool but rather serve as a central gatekeeper in diagnostic algorithms, particularly where advanced testing is limited.

Compared with prior reviews, our study provides a more targeted contribution by focusing on the critical differentiation between ARVC and idiopathic RVOT‐VA. While Iqbal et al. [6] broadly synthesized ventricular arrhythmic risk markers in ARVC, they did not specifically address this diagnostic overlap, limiting direct clinical applicability. Similarly, Alekseeva et al. [7] highlighted repolarization and depolarization abnormalities in pediatric cohorts, findings that parallel adult data but remain descriptive. By systematically dissecting 11 electrocardiographic categories and appraising their diagnostic weight, our meta‐analysis advances beyond these broader syntheses, delivering a clinically actionable framework for frontline decision‐making.

The available evidence base remains relatively small, comprising only 10 eligible studies with a total sample size of 723 participants. Furthermore, several ECG parameters were evaluated in only one or two studies, which may limit the precision and generalizability of the pooled estimates. Accordingly, these findings should be interpreted as exploratory and hypothesis‐generating rather than definitive diagnostic benchmarks. Further large‐scale, well‐designed studies are required to validate these observations and enhance their clinical applicability. Potential publication bias and reliance on observational studies further restrict generalizability, while unmeasured confounders—including age, sex, and disease stage—may influence marker performance. These factors underscore the need for prospective, multicenter validation before integration into standardized diagnostic algorithms.

Another limitation relates to the reference standard used in the included studies. Most studies used the 2010 Revised Task Force Criteria for ARVC diagnosis. Because these criteria include several electrocardiographic findings, there is a potential risk of incorporation bias, which may lead to overestimation of the diagnostic accuracy of certain ECG markers. However, it is important to note that several parameters evaluated in this meta‐analysis are not directly included in the Task Force Criteria, which partially mitigates this concern.

An important limitation of this study is the lack of ARVC stage stratification. Electrocardiographic abnormalities in ARVC evolve with disease progression, with early or concealed stages often demonstrating minimal or absent findings. None of the included studies consistently reported disease stage or provided standardized severity proxies, precluding stage‐specific analyses. This limitation is clinically significant and may have influenced the observed diagnostic performance, particularly sensitivity estimates, which are likely underestimated due to the inclusion of patients in early‐stage disease.

Some included studies were conducted prior to the introduction of the 2010 Revised Task Force Criteria. As individual patient‐level data were not available, reclassification according to the updated criteria was not feasible, and diagnoses were accepted as reported in the original studies. This may have introduced misclassification bias, as earlier diagnostic criteria differ from the 2010 Task Force Criteria, potentially affecting the accuracy of case definition and the estimated diagnostic performance.

Future work should focus on prospective multicenter validation of ECG‐based markers, alongside the development of standardized composite scoring systems that combine repolarization and depolarization indices. Integrating artificial intelligence (AI) for automated ECG interpretation and merging ECG‐derived markers with imaging and genetic data into multimodal algorithms holds promise for earlier, more accurate ARVC detection and personalized risk stratification.

Recent research has investigated machine learning–based ECG analysis for diagnosing ARVC. AI‐based ECG models, including convolutional neural networks [28] and deep‐learning tools [29], have demonstrated promising diagnostic performance in identifying ARVC. Another study reported that machine‐learning (ML) models achieved high sensitivity, specificity, and accuracy, suggesting that AI‐enhanced ECG interpretation may support early ARVC detection and screening [30]. However, one of the limitations of AI is that subtle electrical abnormalities in early or concealed stages are still difficult to discern [28]. The field remains diverse regarding study design, dataset size, technical approaches, and levels of clinical validation, indicating that research on AI‐based ECG models in ARVC is still in its early stages.

5. Conclusion

This systematic review and meta‐analysis highlight the diagnostic value of specific ECG parameters—particularly those related to depolarization abnormalities such as QRS duration in V1–V3, QRS ratio across precordial leads, S‐wave upstroke, and terminal activation delay—in differentiating ARVC from idiopathic OT‐VA. While individual markers show variable sensitivity, their consistently high specificity supports their utility in ruling in ARVC, especially in early or concealed stages. These findings reinforce the role of ECG as a frontline, accessible diagnostic tool and suggest the potential for developing structured multi‐marker ECG approaches or AI‐based screening strategies in clinical practice.

Author Contributions

Conceptualization and design: Erika Maharani, Lucia Kris Dinarti, Dyah Wulan Anggrahini, and Yoga Yuniadi. Data Acquisition: Erika Maharani, Abida Hasna Laila, Nisrina Maulida Rozanti. Data extraction and quality assessment: Erika Maharani, Abida Hasna Laila, and Nisrina Maulida Rozanti. Analysis and interpretation: Erika Maharani, Abida Hasna Laila, and Nisrina Maulida Rozanti. Writing and revision: Erika Maharani, Lucia Kris Dinarti, Dyah Wulan Anggrahini, Yoga Yuniadi, Abida Hasna Laila, and Nisrina Maulida Rozanti.

Funding

The authors have nothing to report.

Ethics Statement

The authors have nothing to report.

Consent

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Appendix Table 1 ECG parameters assessed in each included study.

Appendix Figure 1. QUADAS‐2 risk of bias and applicability assessment.

Appendix Figure 2. Leave‐one‐out sensitivity analysis across included studies.

Data Availability Statement

All data used in this study are available within the articles included in the systematic review and meta‐analysis. The extracted data may be provided upon request.