Figure 1 The study selection process according to the PRISMA. Abbreviations: CNKI, China National Knowledge Infrastructure; DC, deceleration capacity; AF, atrial fibrillation.
Data extraction and quality assessment
Two authors (PKZT, LZH) respectively extracted the following information: (1) participant demographics, including age, gender, and sample size; (2) cardiac investigations, including left atrial diameter (LAD) and left ventricular ejection fraction (LVEF); (3) AF types; (4) comorbidities; (5) follow-up duration; (6) details about CA; (7) AF recurrence rates; (8) the pre- and post-ablation data on DC; (9) the ratios based on DC for predicting AF recurrence, including the odds ratio (OR), relative risk (RR), or hazard ratio (HR); and (10) main findings related to DC. Regarding to AF recurrence, atrial tachyarrhythmia (AT) lasting for 30 s during the first 3 months of follow-up was defined as early recurrence, while AT occurring after a blanking period of 3 months was regarded as late recurrence.
The Newcastle-Ottawa Scale (NOS), a specialized tool for assessing the risk of bias in observational studies, was utilized to evaluate the quality of the included studies by two investigators (PKZT, CAY) independently of each other.20 Discrepancies between two reviewers were addressed through consultation with the senior author (ZLH) until a consensus reached.
Statistical analysis
Continuous variables were expressed as mean values, together with standard deviations (SDs) if applicable, and categorical parameters were presented as percentages. According to the follow-up durations of the included articles, comparisons of DC were performed with the following specified time ranges: prior to ablation, within 3 days post-ablation, and after 3 months post-ablation. Weighted means and SDs of DC in patients with and without AF recurrence were calculated and unpaired Student t test was performed in each period, respectively. Weighted mean differences (WMDs) of DC between recurrence and non-recurrence groups, with 95% confidence interval (CI), were calculated when a fixed- or random-effect model was used in the meta-analysis. A fixed-effect model would be adopted to pool WMDs unless heterogeneity estimated with the I 2 index and Q statistic was significant; otherwise, a random-effect model would be utilized (I 2 > 50% or P< .05). ORs based on DC post-ablation for predicting the risk of AF recurrence were transformed logarithmically and the corresponding standard error (SE) was calculated from 95% CI. If HRs or RRs were available only, they would be considered as the best estimate of ORs. Since significant heterogeneity was observed, a random-effect model was adopted to calculate pooled OR. The stability of the outcomes was assessed with sensitivity analyses performed by omitting one study at a time. Regarding to publication bias, funnel plots were presented, but tests for funnel plot asymmetry were not suggested due to a relatively small number of the studies included in the meta-analysis (fewer than 10 studies).21 P < .05 was regarded as statistical significance. The R statistical software (version 4.0, R Foundation for Statistical Computing, Vienna, Austria) was used to perform all the statistical analyses.
Results