Correspondence to Hongwu Chen; [email protected] ; Dr Yilian Wang; [email protected]
STRENGTHS AND LIMITATIONS OF THIS STUDY
The present study was based on a comprehensive search of large databases including PubMed, MEDLINE, the Cochrane Library and Web of Science.
The Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool was used to evaluate the strength and quality of the evidence.
Only studies that utilised electrophysiologic study (EPS) as the gold standard were included in the analysis.
The accuracy could not be comparable to EPS even though different algorithms were combined when diagnosing wide QRS complex tachycardia.
Background
Wide QRS complex tachycardia (WCT), as detected by a 12-lead ECG, is a frequently encountered condition in emergency rooms and other acute care settings.1 2 The differential diagnosis typically involves distinguishing between ventricular tachycardia (VT) and supraventricular tachycardia (SVT) with aberrant conduction.3–6 Of the two, VT poses a greater risk to patients. While electrophysiology study (EPS) can facilitate the diagnosis of VT, accurately distinguishing VT from WCT in acute care settings is challenging.1 7 8
To improve the accuracy of differentiating VT from WCT, various traditional ECG-based criteria have been established. In the early 1990s, Brugada9 and colleagues developed a multistep decision tree algorithm for the differential diagnosis of WCT, providing clinicians with a systematic approach to reach a definitive diagnosis. Subsequently, several other multistep algorithms have emerged, demonstrating good diagnostic accuracy in identifying VT or SVT, including the Brugada9 algorithm (figure 1A), Griffith10 algorithm, Vereckei-pre11 algorithm (figure 1B) and Vereckei-aVR12 algorithm (figure 1C). Recently, the R wave peak time of lead II criteria13 (RWPT-II, figure 1D) has been established, exhibiting high diagnostic accuracy compared with previous algorithms. Additionally, some novel algorithms employ VT scores or machine learning techniques to improve diagnostic performance.14–20
Enlarge this image.Figure 1. Four algorithms included in this meta-analysis. AV, atrioventricular; BBB, bundle branch block; FB, fascicular block; RWPT, R wave peak time; SVT, supraventricular tachycardia; VT, ventricular tachycardia.
Due to the abundance and heterogeneity of these analysis criteria or algorithms, confusion may arise, and the quantitative review of their diagnostic accuracy has been lacking. Therefore, we performed a systematic review and meta-analysis of various ECG-based analysis algorithms to comprehensively evaluate their diagnostic accuracy in regular WCT, hoping to assist clinicians in selecting appropriate algorithms to diagnose WCT quickly and accurately.
Methods
This meta-analysis was reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis statement21(online supplemental tables S1 and S2). The protocol for this review was submitted before the main analyses were begun and was registered in the International Prospective Register of Systematic Reviews before the analyses were completed, as shown in Supplemental Protocol.
Search methodology
We searched the PubMed, MEDLINE, Cochrane Library and Web of Science bases between January 1990 and May 2022 for articles reporting on the diagnostic accuracy of ECG-based criteria for WCT.
The search terms used were ‘diagnosis’ OR MeSH term ‘diagnostic technique and procedure’ OR ‘Sensitivity’ OR ‘Specificity’ in conjunction (AND) with ‘wide QRS complex tachycardia’ OR ‘WCT’ (AND) ‘ECG’ OR ‘electrocardiogram’. The full search terms used in this study are presented in online supplemental table S3.
The inclusion criteria were adult studies reporting the diagnostic accuracy of ECG-based criteria in patients with WCT. The exclusion criteria were as follows: paediatric studies, studies not related to the analytic criteria of ECG in regular WCT, those that did not refer to a gold standard and those not in the English language. We also obtained primary sources from tracking references by hand searches of review papers and original articles. Only original data were collected in this meta-analysis.
Data extraction
Test performance data were extracted as a 2×2 table of true positive, true negative, false positive and false negative values directly from tabulated results.22 If these data were not available, they were calculated from reported sensitivity, specificity, positive predictive value and/or negative predictive values; if this step was not possible or there was doubt about the 2×2 calculation, the study was excluded from subsequent analysis.
Assessment of study quality
Study quality was assessed by two reviewers using the QUADAS-2 (Quality Assessment of Diagnostic Accuracy Studies) tool,23 24 resolving differences by discussion. Evaluation items for risk of bias were organised into four domains: patient selection, index test, reference standard and flow and timing. The applicability of studies was evaluated for the first three key domains in each study and judged as ‘yes, no or unclear’; ‘yes’ indicated a low risk of bias, ‘no’ indicated a high risk of bias and ‘unclear’ indicated a lack of sufficient information.
Data analysis
When pooled estimates are calculated in diagnostic meta-analyses, the use of DerSimonian-Laird random-effects model is recommended to reflect interstudy heterogeneity.25 Accordingly, pooled sensitivity, specificity, likelihood ratios (LR) and diagnostic ORs (DOR) were generally analysed with DerSimonian-Laird random-effects model and Mantel-Haenszel fixed-effect model if necessary.26 27 DOR reflects the degree of association between the results of diagnostic tests and the disease. When the value is greater than 1, the larger the value, the better the discrimination effect of the diagnostic test. CIs for sensitivity and specificity were calculated using the F-distribution method for the binomial proportion.
A summary receiver operator characteristics curve (SROC) was used to graphically determine performance following testing for correlation between sensitivity and specificity to explore for threshold effects and subsequent assessment for constant DOR.28 A symmetrical or asymmetrical SROC was used depending on whether the DOR was constant. Due to between-study variation in thresholds, we used the hierarchical summary receiver operating characteristics (HSROC) model to estimate SROC curves.29 For HSROC model statistics, the beta value was a scale parameter used to evaluate the symmetry of the SROC curve, and the lambda value was used as the effect index of the discriminative ability of the test.
Heterogeneity was investigated using preplanned subgroup analysis (study quality, location, era of publication, number of patients, etc, see Supplemental Protocol for details) and calculated by the I2 method.30 We also performed a sensitivity analysis in which one study was removed from the analysis and the other studies were analysed to estimate whether a single study was likely to have a significant effect on the results.
A funnel plot and effective sample size regression analysis were used to investigate publication bias.31 Quality assessment of studies was performed using Review Manager (RevMan) software, V.5.4.1, and data analyses were performed using Meta-Disc software, V.1.4, and STATA software, V.16.
Patient and public involvement
Patients and the public were not involved in this study.
Results
Included studies
A total of 467 articles were identified by the search strategy, and 408 (87.4%) were excluded based on the title and abstract. Fifty-nine articles underwent full-test evaluation, and among them, 45 were further excluded based on the prestated criteria (non-English, non-EPS, etc). Finally, 14 observational studies were included in the meta-analysis.9 11–14 32–40 Figure 2 shows the flowchart of studies identified in the systematic review.
Enlarge this image.Figure 2. Flow chart of studies identified in the systematic review. EPS, electrophysiological study. *Other: incorrect data, research purpose does not meet the requirements, number of studies involving one algorithm<=2.
The 14 studies evaluated nine different ECG-based algorithms: Brugada algorithm (11 studies involving 3422 patients), Vereckei-aVR algorithm (nine studies involving 2473 patients), RWPT-II criteria (five studies involving 1656 patients), Vereckei-pre algorithm (three studies involving 1050 patients), Bayesian algorithm (one study), Griffith algorithm (one study), VT score algorithm (one study), limb lead criteria (one study) and RS/QRS ratio criteria (one study). Most of the studies included more than one algorithm, but only the first four algorithms were assessable. Online supplemental table S3(not S3, but S4-S5) shows the characteristics of the 14 studies included in this meta-analysis.
Data from 3966 patients were available. The quality assessment results of the 14 articles are shown in online supplemental figure S1, while nine of 14 studies9 13 14 33–35 38–40 were deemed to be of relatively low quality in selecting patients (the judgement of the first domain of QUADAS-2 was ‘no’ or ‘unclear’), and the other five11 12 32 36 37 were deemed to be of high quality (the judgement was ‘yes’). Six studies noted containing structural heart disease patients in the published article, and eight did not. Nine studies were published in the early period (1990–2013), and five were published in the late period (2014–2022). The median (range) number of patients was 215 (51–587), seven studies had fewer than 215 patients and seven studies had 215 or more patients. Seven studies were conducted in Europe, whereas seven were performed elsewhere.
Meta-analysis
Sensitivity, specificity, likelihood ratio and diagnostic OR
As shown in table 1 and figure 3A, four algorithms were included in the final meta-analysis: the pooled sensitivity was 88.89% (95% CI: 85.03 to 91.86), pooled specificity was 70.55% (95% CI: 62.10 to 77.79), pooled positive LR was 3.02 (95% CI: 2.30 to 3.95), pooled negative LR was 0.15 (95% CI: 0.11 to 0.22) and pooled DOR was 19.17 (95% CI: 11.45 to 32.10). A symmetrical SROC was therefore an appropriate representation of the diagnostic accuracy. The area under the SROC (AUC) was 0.90 (SE=0.02) with a Q statistic of 0.83 (SE=0.03). The HSROC model was used to estimate SROC curves. The beta of the HSROC was 0.10 (95% CI: −0.33 to 0.53), and the Z value was 0.45, with p=0.65. The lambda value was 3.02 (95% CI: 2.43 to 3.61) (figure 3B).
Enlarge this image.Figure 3. Forest plots for sensitivity, specificity and HSROC for all algorithms with total studies. (A) Forest plots for sensitivity and specificity for all algorithms with total studies in the diagnosis of WCT. (B) HSROC curve with 95% confidence region and prediction region for all algorithms with total studies in the diagnosis of WCT. HSROC, hierarchical summary receiving operating characteristic; WCT, wide QRS complex tachycardia.
Assessment of diagnostic accuracy and heterogeneity in subgroup analysis
| Algorithm | Subgroups | No. of studies | Pooled sensitivity (95% CI)/% | Pooled specificity (95% CI)/% | Pooled positive LR (95% CI) | Pooled negative LR (95% CI) | Pooled DOR (95% CI) | I² (%) DOR |
| All algorithms | 14 | 88.89 (85.03 to 91.86) | 70.55 (62.10 to 77.79) | 3.02 (2.30 to 3.95) | 0.15 (0.11 to 0.22) | 19.17 (11.45 to 32.10) | 89.1 | |
| Brugada | All studies | 11 | 90.25 (85.40 to 93.62) | 64.02 (49.34 to 77.48) | 2.51 (1.65 to 3.81) | 0.15 (0.09 to 0.27) | 16.48 (6.25 to 43.48) | 89.9 |
| Period | ||||||||
| Early (1990–2013) | 6 | 90.7 (89.1 to 92.2) | 78.4 (74.5 to 82.0) | 3.59 (1.92 to 6.71) | 0.13 (0.07 to 0.26) | 29.79 (9.03 to 98.29) | 92.6 | |
| Late (2014–2022) | 5 | 91.0 (89.2 to 92.6) | 57.3 (53.1 to 61.4) | 1.83 (1.39 to 2.42) | 0.21 (0.12 to 0.35) | 9.15 (4.35 to 19.24) | 84.1 | |
| Quality assessment | ||||||||
| Low | 5 | 92.3 (90.2 to 94.1) | 72.7 (68.0 to 77.1) | 3.08 (1.24 to 7.66) | 0.15 (0.04 to 0.50) | 22.16 (3.06 to 160.41) | 95.6 | |
| High | 6 | 90.2 (88.8 to 91.6) | 63.7 (59.9 to 67.3) | 2.38 (1.92 to 2.93) | 0.16 (0.12 to 0.20) | 16.38 (11.52 to 23.28) | 49.7 | |
| Location | ||||||||
| Euro | 5 | 92.0 (90.6 to 93.2) | 72.4 (68.9 to 75.7) | 3.71 (2.10 to 6.57) | 0.10 (0.06 to 0.18) | 41.95 (13.82 to 127.37) | 93.4 | |
| Non-Euro | 6 | 88.7 (86.4 to 90.7) | 56.5 (51.1 to 61.7) | 1.86 (1.38 to 2.51) | 0.25 (0.15 to 0.42) | 7.82 (3.72 to 16.41) | 79.7 | |
| Number | ||||||||
| <218 | 5 | 84.6 (80.6 to 88.0) | 48.6 (40.4 to 57.0) | 1.64 (1.21 to 2.23) | 0.31 (0.20 to 0.48) | 5.91 (2.84 to 12.31) | 59.0 | |
| >218 | 6 | 92.0 (90.8 to 93.1) | 70.0 (66.9 to 73.0) | 3.36 (2.31 to 4.89) | 0.11 (0.07 to 0.17) | 34.06 (14.90 to 77.89) | 92.2 | |
| SHD | ||||||||
| With | 5 | 89.1 (87.4 to 90.7) | 62.7 (58.8 to 66.5) | 2.27 (1.59 to 3.26) | 0.19 (0.14 to 0.27) | 11.84 (6.30 to 22.26) | 84.2 | |
| Without | 6 | 93.1 (91,5 to 94.5) | 73.7 (69.1 to 77.9) | 2.92 (1.50 to 5.65) | 0.13 (0.05 to 0.34) | 24.30 (5.52 to 107.07) | 93.1 | |
| Vereckei-aVR | All studies | 9 | 90.35 (85.87 to 93.52) | 60.88 (49.71 to 71.01) | 2.31 (1.73 to 3.09) | 0.16 (0.10 to 0.25) | 14.57 (7.18 to 29.54) | 90.3 |
| Period | ||||||||
| Early (1990–2013) | 4 | 92.8 (90.5 to 94.6) | 69.4 (63.2 to 75.1) | 2.81 (1.86 to 4.25) | 0.13 (0.05 to 0.35) | 22.33 (5.91 to 84.37) | 90.0 | |
| Late (2014–2022) | 5 | 86.8 (84.7 to 88.7) | 58.3 (54.2 to 62.4) | 2.00 (1.32 to 3.01) | 0.20 (0.11 to 0.39) | 10.16 (4.05 to 25.46) | 89.3 | |
| Quality assessment | ||||||||
| Low | 4 | 88.7 (85.2 to 91.7) | 48.0 (41.2 to 54.8) | 2.04 (1.29 to 3.22) | 0.22 (0.12 to 0.38) | 9.83 (4.04 to 23.91) | 74.3 | |
| High | 5 | 89.0 (87.3 to 90.6) | 66.8 (62.8 to 70.6) | 2.52 (1.70 to 3.74) | 0.14 (0.06 to 0.33) | 18.54 (5.83 to 59.03) | 94.0 | |
| Location | ||||||||
| Non-Euro | 6 | 90.5 (88.3 to 92.4) | 57.4 (52.1 to 62.5) | 2.14 (1.29 to 3.53) | 0.19 (0.11 to 0.34) | 11.32 (4.12 to 29.02) | 85.7 | |
| Number | ||||||||
| <218 | 5 | 88.9 (85.5 to 91.8) | 61.3 (53.9 to 68.3) | 2.13 (1.46 to 3.09) | 0.19 (0.10 to 0.36) | 11.73 (4.82 to 28.54) | 73.5 | |
| >218 | 4 | 89.0 (87.2 to 90.5) | 61.8 (57.9 to 65.6) | 2.52 (1.46 to 4.34) | 0.14 (0.06 to 0.35) | 18.12 (5.07 to 64.69) | 95.5 | |
| SHD | ||||||||
| With | 5 | 88.5 (86.6 to 90.2) | 58.9 (54.8 to 62.8) | 2.29 (1.54 to 3.40) | 0.15 (0.06 to 0.32) | 16.00 (5.68 to 45.04) | 92.1 | |
| Without | 4 | 90.0 (87.3 to 92.3) | 69.4 (62.8 to 75.5) | 2.36 (1.20 to 4.65) | 0.20 (0.09 to 0.46) | 12.17 (3.15 to 46.99) | 88.4 | |
| RWPT-II | All studies | 5 | 78.15 (61.04 to 89.09) | 88.30 (77.30 to 94.36) | 6.68 (2.86 to 15.59) | 0.25 (0.12 to 0.51) | 27.00 (5.91 to 123.40) | 79.1 |
| Vereckei-pre | All studies | 3 | 94.80 (93.30 to 96.20) | 75.40 (69.60 to 80.60) | 3.79 (2.86 to 5.03) | 0.08 (0.04 to 0.14) | 60.70 (38.98 to 94.54) | 0 |
DOR, diagnostic OR; LR, likelihood ratio; RWPT II, R wave peak time of lead II; SHD, structural heart disease.
Brugada algorithm
For the Brugada algorithm, the pooled sensitivity was 90.25% (95% CI: 85.40 to 93.62), pooled specificity was 64.02% (95% CI: 49.34 to 77.48) (online supplemental figure S2A), pooled positive LR was 2.51 (95% CI: 1.65 to 3.81), pooled negative LR was 0.15 (95% CI: 0.09 to 0.27) and pooled DOR was 16.48 (95% CI: 6.25 to 43.48). The AUC was 0.94 (SE=0.05) with a Q statistic of 0.87 (SE=0.06). The beta of the HSROC was 0.30 (95% CI: −0.26 to 0.86), and the Z value was 1.04, with p=0.30. The lambda value was 3.08 (95% CI: 1.97 to 4.19) (online supplemental figure S2B).
Vereckei-aVR algorithm
For this algorithm, the pooled sensitivity was 90.35% (95% CI: 85.87 to 93.52), pooled specificity was 60.88% (95% CI: 49.71 to 71.01) (online supplemental figure S3A), pooled positive LR was 2.31 (95% CI: 1.73 to 3.09), pooled negative LR was 0.16 (95% CI: 0.10 to 0.25) and pooled DOR was 14.57 (95% CI: 7.18 to 29.54). The AUC was 0.89 (SE=0.07) with a Q statistic of 0.82 (SE=0.07). The beta of the HSROC was 0.08 (95% CI: −0.69 to 0.84), and the Z value was 0.19, with p=0.85. The lambda value was 2.75 (95% CI: 1.76 to 3.74) (online supplemental figure S3B).
RWPT-II criteria
For the RWPT-II criteria, the pooled sensitivity was 78.15% (95% CI: 61.04 to 89.09), pooled specificity was 88.30% (95% CI: 77.30 to 94.36) (online supplemental figure S4A), pooled positive LR was 6.68 (95% CI: 2.86 to 15.59), pooled negative LR was 0.25 (95% CI: 0.12 to 0.51) and pooled DOR was 27.00 (95% CI: 5.91 to 123.40). The AUC was 0.88 (SE=0.11) with a Q statistic of 0.81 (SE=0.11). The beta of the HSROC was −0.17 (95% CI: −1.27 to 0.94), and the Z value was −0.30, with p=0.77. The lambda value was 3.37 (95% CI: 1.83 to 4.91) (online supplemental figure S4B).
Vereckei-pre algorithm
The pooled sensitivity of this algorithm was 94.80% (95% CI: 93.30 to 96.20) (online supplemental figure S5A), pooled specificity was 75.40% (95% CI: 69.60 to 80.60) (online supplemental figure S5B), pooled positive LR was 3.79 (95% CI: 2.86 to 5.03), pooled negative LR was 0.08 (95% CI: 0.04 to 0.14) and pooled DOR was 60.70 (95% CI: 38.98 to 94.54). The AUC was 0.95 (SE=0.02) with a Q statistic of 0.89 (SE=0.03) (online supplemental figure S5C). Because the Vereckei-pre algorithm was included in only three studies, the HSROC was unable to be derived.
Heterogeneity analysis
Heterogeneity was detected in all pooled indices (table 1). For all 14 studies of the four algorithms, the heterogeneity (I2) was 96.30% (sensitivity), 91.87% (specificity), 92.10% (positive LR), 94.80% (negative LR) and 89.1% (DOR).
For the Brugada algorithm, the heterogeneity was 86.00% (sensitivity), 82.77% (specificity), 92.00% (positive LR), 87.10% (negative LR) and 89.90% (DOR); for the Vereckei-aVR algorithm, the heterogeneity was 90.48% (sensitivity), 89.03% (specificity), 92.20% (positive LR), 90.50% (negative LR) and 90.30% (DOR); for the RWPT-II criteria, the heterogeneity was 96.07% (sensitivity), 82.77% (specificity), 58.40% (positive LR), 93.00% (negative LR) and 79.10% (DOR) and for the Vereckei-pre algorithm, the heterogeneity was 78.20% (sensitivity), 60.90% (specificity), 29.30% (positive LR), 73.80% (negative LR) and 0% (DOR).
Subgroup analysis
Subgroup analysis was performed to assess differences in heterogeneity and diagnostic accuracy among the prespecified groups (table 1). For the Brugada algorithm, the heterogeneity of the DOR was lower in the high-quality studies (I2, 49.7% vs 95.6%) and in the smaller studies (I2, 59.0% vs 92.2%). There was no significant difference in the heterogeneity of the other subgroup analyses.
Meta-regression
There was no significant correlation between any of the covariates and the DOR in the univariate meta-regression analysis (online supplemental table S6). Given that there were only 14 studies included, the power of the multivariate meta-regression is low, which would limit the overall value of the meta-regression.
Sensitivity analysis
We investigated the influence of a single study by excluding one study at a time and found that omitting the study ‘Brugada 1991’ with the Brugada algorithm, ‘Vereckei 2008’ with the Vereckei-aVR algorithm and ‘Pava 2010’ with the RWPT-II algorithm had the smallest OR for each algorithm (online supplemental figure S6–S8).
Publication bias
No significant publication bias was found in the studies included in this meta-analysis since linear regression analysis indicated that p>0.05 (figure 4 and online supplemental figure S9).
Enlarge this image.Figure 4. Deeks’ funnel plot asymmetry test for assessment of publication bias for all algorithms with total studies. ESS, effective sample size.
Discussion
This study is the first meta-analysis to evaluate the accuracy of different ECG-based diagnostic algorithms in WCT. The pooled specificity of the included algorithms was high (88.89%), while the pooled sensitivity was moderate (70.55%). The symmetrical SROC curve demonstrated an AUC of 0.90, indicating the effectiveness of ECG-based algorithms in distinguishing VT from SVT.
The diagnosis of WCT, particularly in emergency cases, has posed challenges in clinical practice, especially in emergency cases. While EPS is considered the gold standard,41 it is not readily available in emergency situations. Although there are a number of ECG-based diagnostic options, it is not easy to remember all the algorithms even for professionals, so many clinicians are confused regarding which to choose. Consequently, clinicians may resort to using multiple algorithms simultaneously, which prolongs diagnostic time without necessarily improving accuracy.
Among the evaluated algorithms, the Brugada and Vereckei-aVR algorithms are widely accepted. The RWPT-II algorithm is simpler than the Brugada and Vereckei-aVR algorithms, with only one standard, and it has been broadly accepted.
With respect to each algorithm individually, the Brugada algorithm and Vereckei-aVR algorithm had similar pooled sensitivity (90.25% vs 90.35%) and specificity (64.02% vs 60.88%). Regarding the RWPT-II algorithm, the pooled sensitivity (78.15%) was relatively low, but the pooled specificity (88.30%) was greater than that of the other three algorithms. The Vereckei-pre algorithm had the highest pooled sensitivity (94.80%) and relatively high specificity (75.40%); however, such excellent data might not be convincing.
In terms of the Vereckei-pre algorithm, two of the three studies had similar participants and a similar period, which had a high pooled sensitivity (95.70%). This point would explain why the higher pooled accuracy of the Vereckei-pre algorithm contrasted with that of the Vereckei-aVR algorithm, which was proposed by Vereckei based on the previous algorithm and was originally thought to be a simpler and more accurate algorithm.
Subgroup analysis revealed that studies published in the early period generally exhibited higher DOR than those published in the later period, suggesting that the diagnostic accuracy of the algorithms might have declined over time. This observation held true for both the Brugada (DOR: 29.79 vs 9.15) and Vereckei-aVR algorithms (DOR: 22.33 vs 10.16). In this meta-analysis, we found that there was high diagnostic accuracy when each algorithm was designed, but a series of studies yielded large gaps in contrast with the initial study when validating the algorithm afterwards. Similarly, we also conducted sensitivity analyses separately for the various algorithms and found that the designated study in which each algorithm was proposed had the highest heterogeneity for each algorithm (online supplemental figure S6–S8).
Furthermore, studies conducted in Europe demonstrated higher diagnostic accuracy for both the Brugada (DOR: 41.95 vs 7.82) and Vereckei-aVR algorithms (DOR: 22.37 vs 11.32). It remains unclear why these objective, language-independent ECG-based algorithms would perform better in the European subgroups, but it is an important observation and could have implications for the role of Brugada and Vereckei-aVR algorithms as worldwide tools for identifying WCT, particularly when the specificity is only 56.5% to 57.4% in non-Europeans compared with 65.2% to 72.4% in Europeans. Furthermore, the studies with larger numbers of participants had a higher DOR for both the Brugada (DOR: 34.06 vs 5.91) and Vereckei-aVR algorithms (DOR: 18.12 vs 11.73). This contrasts with the results of other meta-analyses involving fewer patients, which were prone to having a higher DOR than those with a larger number of patients.22 The underlying reasons for this observation merit further exploration.
Due to the limited number of studies, we did not conduct subgroup analysis on the RWPT-II algorithm, and because of the relatively low sensitivity and highest specificity, we thought the RWPT-II algorithm would be more suitable as a supplement to other algorithms. Considering the relatively high sensitivity and low specificity of the Brugada and Vereckei-aVR algorithm, it would be a good idea to combine Brugada or Vereckei-aVR with RWPT-II algorithm, especially when analysing non-European data. Furthermore, given that the RWPT-II algorithm only has one criterion and is easy to implement, the combination would improve diagnostic accuracy without significantly increasing the workload of clinicians.
The limitations of this study were as follows: (1) only 14 studies with reference tests for EPS were included. This fact might have limited the number of included studies and had an impact on the final consolidated results. However, this limitation improved the overall quality of the included studies, and no publication bias was detected. (2) This meta-analysis did not include the latest algorithms, such as the VT-score algorithm, because the number of studies were less than two studies. Therefore, readers might be misled into the impression that the new algorithm is not as good as the typical algorithms. (3) Regardless of the combination of various algorithms, the diagnostic accuracy cannot be comparable to the gold standard EPS.
Conclusion
ECG-based algorithms exhibit high sensitivity and moderate specificity in the diagnosis of WCT. A combination of Brugada or Vereckei-aVR algorithm with RWPT-II may be considered to diagnose WCT.
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information.
Ethics statements
Patient consent for publication
Not required.
XS and YT contributed equally.
Contributors HC and YW obtained the funding, developed the research design, and interpreted the results. HC had primary responsibility for the overall content. XS, YT and SM analysed the data and interpreted the results. XS, YT and HC drafted manuscript. All authors critically reviewed and approved the manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.
1 Littmann L, Olson EG, Gibbs MA. Initial evaluation and management of wide-complex tachycardia: A simplified and practical approach. Am J Emerg Med 2019; 37: 1340–5. doi:10.1016/j.ajem.2019.04.027
2 Wellens HJ, Bär FW, Lie KI. The value of the electrocardiogram in the differential diagnosis of a tachycardia with a widened QRS complex. Am J Med 1978; 64: 27–33. doi:10.1016/0002-9343(78)90176-6
3 Brady WJ, Mattu A, Tabas J, et al. The differential diagnosis of wide QRS complex tachycardia. The American Journal of Emergency Medicine 2017; 35: 1525–9. doi:10.1016/j.ajem.2017.07.056
4 Alzand BSN, Crijns HJGM. Diagnostic criteria of broad QRS complex tachycardia: decades of evolution. Europace 2011; 13: 465–72. doi:10.1093/europace/euq430
5 Wellens HJ. Electrophysiology: ventricular tachycardia: diagnosis of broad QRS complex tachycardia. Heart 2001; 86: 579–85. doi:10.1136/heart.86.5.579
6 Delbridge TR, Yealy DM. Wide complex tachycardia. Emerg Med Clin North Am 1995; 13: 903–24.
7 Attin M. Electrophysiology study: a comprehensive review. Am J Crit Care 2001; 10: 260–73.
8 Akhtar M, Shenasa M, Jazayeri M, et al. Wide QRS complex tachycardia. reappraisal of a common clinical problem. Ann Intern Med 1988; 109: 905–12. doi:10.7326/0003-4819-109-11-905
9 Brugada P, Brugada J, Mont L, et al. A new approach to the differential diagnosis of a regular tachycardia with a wide QRS complex. Circulation 1991; 83: 1649–59. doi:10.1161/01.CIR.83.5.1649
10 Griffith MJ, de Belder MA, Linker NJ, et al. Multivariate analysis to simplify the differential diagnosis of broad complex tachycardia. Br Heart J 1991; 66: 166–74. doi:10.1136/hrt.66.2.166
11 Vereckei A, Duray G, Szénási G, et al. Application of a new algorithm in the differential diagnosis of wide QRS complex tachycardia. Eur Heart J 2007; 28: 589–600. doi:10.1093/eurheartj/ehl473
12 Vereckei A, Duray G, Szénási G, et al. New algorithm using only lead aVR for differential diagnosis of wide QRS complex tachycardia. Heart Rhythm 2008; 5: 89–98. doi:10.1016/j.hrthm.2007.09.020
13 Pava LF, Perafán P, Badiel M, et al. R-wave peak time at DII: A new criterion for differentiating between wide complex QRS tachycardias. Heart Rhythm 2010; 7: 922–6. doi:10.1016/j.hrthm.2010.03.001
14 Jastrzebski M, Sasaki K, Kukla P, et al. The ventricular tachycardia score: a novel approach to electrocardiographic diagnosis of ventricular tachycardia. Europace 2016; 18: 578–84. doi:10.1093/europace/euv118
15 Pachón M, Arias MA, Salvador-Montañés Ó, et al. A scoring algorithm for the accurate differential diagnosis of regular wide QRS complex tachycardia. Pacing Clin Electrophysiol 2019; 42: 625–33. doi:10.1111/pace.13658
16 Mazandarani FN, Mohebbi M. Wide complex tachycardia discrimination using dynamic time warping of ECG beats. Comput Methods Programs Biomed 2018; 164: 238–49. doi:10.1016/j.cmpb.2018.04.009
17 May AM, DeSimone CV, Kashou AH, et al. The WCT formula: A novel algorithm designed to automatically differentiate wide-complex tachycardias. Journal of Electrocardiology 2019; 54: 61–8. doi:10.1016/j.jelectrocard.2019.02.008
18 Kashou AH, DeSimone CV, Deshmukh AJ, et al. The WCT formula II: an effective means to automatically differentiate wide complex tachycardias. Journal of Electrocardiology 2020; 61: 121–9. doi:10.1016/j.jelectrocard.2020.05.004
19 May AM, DeSimone CV, Kashou AH, et al. The VT prediction model: A simplified means to differentiate wide complex tachycardias. J Cardiovasc Electrophysiol 2020; 31: 185–95. doi:10.1111/jce.14321
20 Kashou AH, LoCoco S, McGill TD, et al. Automatic wide complex tachycardia differentiation using mathematically synthesized Vectorcardiogram signals. Noninvasive Electrocardiol 2022; 27. doi:10.1111/anec.12890 Available: https://onlinelibrary.wiley.com/toc/1542474x/27/1
21 Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 2021; 372: n71. doi:10.1136/bmj.n71
22 Torlot FJ, McPhail MJW, Taylor-Robinson SD. Meta-analysis: the diagnostic accuracy of critical flicker frequency in minimal hepatic encephalopathy. Aliment Pharmacol Ther 2013; 37: 527–36. doi:10.1111/apt.12199
23 Whiting PF, Rutjes AWS, Westwood ME, et al. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med 2011; 155: 529–36. doi:10.7326/0003-4819-155-8-201110180-00009
24 Schueler S, Schuetz GM, Dewey M. The revised QUADAS-2 tool. Ann Intern Med 2012; 156: 323; doi:10.7326/0003-4819-156-4-201202210-00018
25 Park SH. Tools for assessing fall risk in the elderly: a systematic review and meta-analysis. Aging Clin Exp Res 2018; 30: 1–16. doi:10.1007/s40520-017-0749-0
26 Reitsma JB, Glas AS, Rutjes AWS, et al. Bivariate analysis of sensitivity and specificity produces informative summary measures in diagnostic reviews. J Clin Epidemiol 2005; 58: 982–90. doi:10.1016/j.jclinepi.2005.02.022
27 Wang F, Gatsonis CA. Hierarchical models for ROC curve summary measures: design and analysis of multi-reader, multi-modality studies of medical tests. Stat Med 2008; 27: 243–56. doi:10.1002/sim.2828
28 Moses LE, Shapiro D, Littenberg B. Combining independent studies of a diagnostic test into a summary ROC curve: data-analytic approaches and some additional considerations. Stat Med 1993; 12: 1293–316. doi:10.1002/sim.4780121403
29 Rutter CM, Gatsonis CA. A Hierarchical regression approach to meta-analysis of diagnostic test accuracy evaluations. Stat Med 2001; 20: 2865–84. doi:10.1002/sim.942
30 Higgins JPT, Thompson SG, Deeks JJ, et al. Measuring inconsistency in meta-analyses. BMJ 2003; 327: 557–60. doi:10.1136/bmj.327.7414.557
31 Deeks JJ, Macaskill P, Irwig L. The performance of tests of publication bias and other sample size effects in systematic reviews of diagnostic test accuracy was assessed. J Clin Epidemiol 2005; 58: 882–93. doi:10.1016/j.jclinepi.2005.01.016
32 Isenhour JL, Craig S, Gibbs M, et al. Wide-complex tachycardia: continued evaluation of diagnostic criteria. Acad Emerg Med 2000; 7: 769–73. doi:10.1111/j.1553-2712.2000.tb02267.x
33 Lau EW, Ng GA. Comparison of the performance of three diagnostic Algorithms for regular broad complex tachycardia in practical application. Pacing Clin Electrophysiol 2002; 25: 822–7. doi:10.1046/j.1460-9592.2002.00822.x
34 Lin T, Ma Y, Muhu Y, et al. Value of aVR lead four steps algorithm on differential diagnosis of wide QRS complex tachycardia. Zhonghua Xin Xue Guan Bing Za Zhi 2011; 39: 69–72.
35 Baxi RP, Hart KW, Vereckei A, et al. Vereckei criteria as a diagnostic tool amongst emergency medicine residents to distinguish between ventricular tachycardia and supra-ventricular tachycardia with Aberrancy. J Cardiol 2012; 59: 307–12. doi:10.1016/j.jjcc.2011.11.007
36 Szelényi Z, Duray G, Katona G, et al. Comparison of the "real-life" diagnostic value of two recently published electrocardiogram methods for the differential diagnosis of wide QRS complex tachycardias. Acad Emerg Med 2013; 20: 1121–30. doi:10.1111/acem.12247
37 Kaiser E, Darrieux FCC, Barbosa SA, et al. Differential diagnosis of wide QRS tachycardias: comparison of two electrocardiographic Algorithms. Europace 2015; 17: 1422–7. doi:10.1093/europace/euu354
38 May AM, Brenes-Salazar JA, DeSimone CV, et al. Electrocardiogram Algorithms used to differentiate wide complex tachycardias demonstrate diagnostic limitations when applied by non-Cardiologists. J Electrocardiol 2018; 51: 1103–9. doi:10.1016/j.jelectrocard.2018.09.015
39 Chen Q, Xu J, Gianni C, et al. Simple electrocardiographic criteria for rapid identification of wide QRS complex tachycardia: the new limb lead algorithm. Heart Rhythm 2020; 17: 431–8. doi:10.1016/j.hrthm.2019.09.021
40 Kim M, Kwon CH, Lee JH, et al. Right bundle branch block-type wide QRS complex tachycardia with a reversed R/S complex in lead V6: development and validation of electrocardiographic differentiation criteria. Heart Rhythm 2021; 18: 181–8. doi:10.1016/j.hrthm.2020.08.023
41 Hammill SC, Sugrue DD, Gersh BJ, et al. Clinical Intracardiac electrophysiologic testing: technique, diagnostic indications, and therapeutic uses. Mayo Clin Proc 1986; 61: 478–503. doi:10.1016/s0025-6196(12)61984-3
© 2023 Author(s) (or their employer(s)) 2023. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ. http://creativecommons.org/licenses/by-nc/4.0/ This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/ . Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Details
; Teng, Yanling 2 ; Mu, Shengnan 3 ; Wang, Yilian 1
; Chen, Hongwu 4
1 Department of Cardiology, The Second People's Hospital of Lianyungang, Affiliated to Kangda College of Nanjing Medical University, Lianyungang, China; Department of Cardiology, The First Affiliated Hospital of Nanjing Medical University, Nanjing, China
2 Department of Cardiology, The First people’s Hospital of Lianyungang, The First Affiliated Hospital of Kangda College of Nanjing Medical University, Lianyungang, China
3 Department of Cardiology, The Second People's Hospital of Lianyungang, Affiliated to Kangda College of Nanjing Medical University, Lianyungang, China
4 Department of Cardiology, The First Affiliated Hospital of Nanjing Medical University, Nanjing, China