2.1 Search Strategy
The search was performed using the PubMed, Cochrane and Embase databases with the start date prior to May 30, 2020. To identify and retrieve all potentially relevant articles, the search was performed utilizing the following expression: [(‘epicardial adipose tissue’ OR ‘epicardial fat’ OR ‘pericardial adipose tissue’ OR‘pericardial fat’ OR ‘paracardial adipose tissue’ OR ‘paracardial fat’) AND (‘atrial fibrillation’ OR Arrhythmias, Cardiac’)]. Additionally, hand searches of the reference list of included studies were performed. We conducted this meta-analysis in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. The systematic review registered in PROSPERO (ID: 219400). As per the different measuring methods of EAT, eligible studies fall into three sub-groups, including total EAT volume, EAT thickness and LA-EAT volume.
2.2 Eligibility criteria
Case reports and review articles were excluded. Only English publications in peer-reviewed journals were eligible for this meta-analysis if they: (1) reported EAT thickness or volume with at least one of the following main confounders: age, gender, hypertension and Body Mass Index (BMI); (2) reported EAT parameters in two groups of patients with and without AF occurrence after cardiac surgeries, and 3) measured EAT by CT, magnetic resonance imaging or echocardiography. The literature search was independently reviewed by two assessors to ensure full compliance with the study inclusion criteria. If no agreement could be reached, it would be decided by a third senior author.
2.3 Study quality assessment
The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement 23 was used to evaluate the quality of the included studies. STROBE consists of 22 items assessing quality of the information reported in the abstract, including introduction, study design and setting, statistical evaluation, results, discussion and findings. The point of STROBE judgment for all included studies was prepared based on rating scale ranging from 0 to 22 with higher score resulting in high quality. The STROBE score was independently evaluated by two investigators. And a third senior author was consulted when there were discrepancies. In the case of conference abstracts, which are without full-text, we used another 12-item STROBE checklist. All quantity scores were included in Table 1.
2.4 Statistical methods
The Cochrane Q test and I squared index were used to evaluate heterogeneity of individual studies. I2> 50% suggested high heterogeneity. In the case of high heterogeneity, a random effect model was adopted. Considering the small size of samples, we used sensitivity analysis to explore probable sources of heterogeneity. The possibility of publication bias was evaluated through the Funnel plots and Egger’s regression test. Statistical analysis was performed using STATA package vers.16 (StataCorp, College Station, TX, USA). P<0.05 was considered statistically significant.
3. RESULTS