Data extraction and quality assessment
The retrieved articles were initially reviewed by two independent reviewers (SJ and AD). They screened titles and abstracts and removed the duplicates using the EndNote X9 software. The extracted data were further verified by the reviewers. The third investigator (MM) was then consulted to address any discrepancies concerning the evaluation of studies. The study design, baseline characteristics, and various outcomes were extracted. For the quality assessment of the included randomized control trials (RCTs) (13, 14), the Revised Cochrane Risk of Bias tool (ROB-2) was used. The modified Newcastle Ottawa scale was used for quality assessment of the remaining non-randomized trial (15).
Interventions
Operators performed PVI under fluoroscopic guidance by advancing an inflated cryoballoon (CB) catheter (Medtronic Inc., Minneapolis, MN) to each pulmonary vein antrum and freezing the tissue.
PWI was performed by delivery of at least two cryo-balloon freezes applied to each quadrant of the left atrial posterior wall (LAPW) under fluoroscopic guidance. This was accomplished by clocking/counter clocking of the sheath and balloon catheter after positioning the cryo-balloon in the individual pulmonary veins supported by pushing in the Achieve catheter (Medtronic Inc., Minneapolis, MN) to apply pressure to the LAPW. The cryo-balloon position was monitored on intracardiac ultrasound. Post ablation 3-dimensional (3D) voltage maps were created after each procedure.
Study definitions and end points
The primary outcome of interest was the recurrence of AF after the 90-day blanking period. Recurrence of any type of atrial tachyarrhythmia was defined as >30 seconds on any cardiac rhythm recording following the specified blanking period after the index CBA procedure.
Secondary outcomes included recurrence of all atrial arrhythmias (AF, atrial flutter, and atrial tachycardia), total ablation time, and adverse events.
Statistical analysis
We used the Review Manager (RevMan) computer program, version 5.4, to perform statistical analysis. A random effects model with Mantel-Haenszel weighting was then used to analyze our primary and secondary endpoints. The outcomes were reported as risk ratios (RR) and mean difference (MD) with a 95% confidence interval (CI). For assessment of study heterogeneity, the Higgins-I-squared (I2) model was used with values < 25%, 25–50%, 50–75%, and > 75% corresponding to no, low, moderate, and high degrees of heterogeneity, respectively. (16) A p-value of <0.05 was considered statistically significant. The publication bias was depicted graphically using funnel plots