This commentary refers to ‘Mortality after drug-eluting stents vs. coronary artery bypass grafting for left main coronary artery disease: a meta-analysis of randomized controlled trials’ by Y. Ahmad et al., doi:10.1093/eurheartj/ehaa135.
A recent systematic review and meta-analysis study was published in the European Heart Journal by Ahmad et al.1
Innovative investigation: This study treads novel ground in delineating the comparative effectiveness of each mode of treatment, with focus on prognosis/mortality of the patient post-treatment. Nevertheless, there are some aspects of the meta-analysis that we would like to require addressing.
Publication bias: Ahmad et al. used PRISMA guidelines for conducting of the analysis, which requires that meta-analysis studies also conduct publication bias assessment.2 Therefore, we would like to recommend publication bias analysis using Egger’s bias indicator test.3
Variance of true effect size: We recommend the use of the Tau2 parameter, in addition to the I 2 parameter for the assessment of between study heterogeneity, in order to present a more robust analysis (Table 1).
Table 1.
Heterogeneity and hypothesis testing of the included studies
| Subgroups | Heterogeneity |
HR | 95% CI |
HR | 95% CI |
Fixed effects model |
Random effects model |
||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Q | P | I 2 | Low | High | Low | High | Z | P | Studies | Z | P | Studies | |||
| Risk of death | 7.12 | 0.13 | 43.89 | 1.08 | 0.90 | 1.30 | 1.03 | 0.78 | 1.35 | 0.87 | 0.38 | 5 | 0.22 | 0.83 | 5 |
| Risk of cardiac death | 3.80 | 0.28 | 21.07 | 1.03 | 0.78 | 1.35 | 1.01 | 0.74 | 1.39 | 0.22 | 0.83 | 4 | 0.10 | 0.92 | 4 |
| Risk of stroke | 9.2 | 0.06 | 56.51 | 0.75 | 0.52 | 1.08 | 0.73 | 0.36 | 1.47 | −1.53 | 0.13 | 5 | −0.87 | 0.37 | 5 |
| Risk of MI | 1.45 | 0.69 | 0 | 1.24 | 0.95 | 1.62 | 1.25 | 0.95 | 1.63 | 1.61 | 0.10 | 4 | 1.61 | 0.11 | 4 |
| Risk of revascularization | 0.72 | 0.95 | 0 | 1.88 | 1.58 | 2.25 | 1.89 | 1.58 | 2.25 | 7.09 | 0 | 5 | 7.09 | 0 | 5 |
Pooled effect size: In addition, we observe that the study conducted used relative risk as the effect size metric for meta-analysis. However, while combining randomized controlled trials in a meta-analysis, the standard mean difference may be a more appropriate effect size metric to represent the pooled data.4
Sample size: The lack of sufficient number of studies is also an issue, as the meta-analysis conducted here only uses a total of five studies, which is not sufficient to provide a result of sufficient power to be used in clinical decision-making. A limitation that requires highlighting.
Survival endpoints: We would also like the authors to describe their reasoning behind comparison of studies with different endpoints, as the studies are split between a 5-year follow-up, and a 1-year follow-up, and comparing studies with variable endpoints may introduce heterogeneity into the study.1
Statistical significance or estimated effect size: It is also recommended that the results of the study be described purely in terms of the effect size metric in the meta-analysis, and not using ‘statistical significance’, as it has shown to be limited in describing statistical results.5
It also worth noting that as this study is a literature based meta-analysis, the results may guide clinical decision making, but it cannot present any recommendations for treatment.
Conflict of interest: none declared.
Contributor Information
Rama Jayaraj, Department of Artificial Intelligence, Nanjing University of Information Science and Technology (NUIST), Jiangsu, China; Northern Territory Medical Program (NTMP), College of Medicine and Public Health, Flinders University, CDU Campus, Ellengowan Drive, Darwin, Northern Territory 0909, Australia.
Chellan Kumaraswamy, School of Public Health, The University of Adelaide, North Terrace Campus, Adelaide, SA 5005, Australia.
Peter Shaw, Department of Artificial Intelligence, Nanjing University of Information Science and Technology (NUIST), Jiangsu, China.
References
- 1. Ahmad Y, Howard JP, Arnold AD, Cook CM, Prasad M, Ali ZA, Parikh MA, Kosmidou I, Francis DP, Moses JW, Leon MB, Kirtane AJ, Stone GW, Karmpaliotis D. Mortality after drug-eluting stents vs. coronary artery bypass grafting for left main coronary artery disease: a meta-analysis of randomized controlled trials. European Heart Journal 2020;doi:10.1093/eurheartj/ehaa135. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Moher D, Altman DG, Liberati A, Tetzlaff J, Prisma S. PRISMA statement. Epidemiology 2011;22:128. [DOI] [PubMed] [Google Scholar]
- 3. Egger M, Smith GD, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997;315:629–634. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Jayaraj R, Kumarasamy C, Madhav MR. Practical approaches to interpretation of findings from a systematic review and network meta-analysis on efficacy and resistance of different artemisinin-based combination therapies. Parasitol Int 2019;73:101949. [DOI] [PubMed] [Google Scholar]
- 5. Kalinowski P, Fidler F. Interpreting significance: the differences between statistical significance, effect size, and practical importance. Newborn Infant Nurs Rev 2010;10:50–54. [Google Scholar]