Six RCTs were included in the review (n=693 patients). Three trials did not report power calculations; the remaining three trials enrolled sufficient patients to fulfil sample size calculations. In trials where crossovers were permitted, 51% of patients assigned to receive non-catheter ablation crossed over and received pulmonary vein isolation. Across four trials that evaluated repeat performance of pulmonary vein isolation, 17% of patients required a repeat procedure. Follow-up was 12 months in all the trials.
There were statistically significant benefits observed with the use of pulmonary vein isolation for maintaining sinus rhythm (OR 9.74, 95% CI 3.98 to 23.87; six trials). There was significant statistical heterogeneity (Cochran Q statistic p<0.001). After the exclusion of one trial that enrolled only patients with persistent atrial fibrillation, higher odds of atrial fibrillation-free survival were found for patients for whom the majority presented with paroxysmal atrial fibrillation (OR 15.78, 95% CI 10.07 to 24.73).
In two trials that reported freedom from atrial fibrillation and anti-arrhythmic medication, 86% of patients randomly assigned to receive pulmonary vein isolation were free from atrial fibrillation and requiring anti-arrhythmic medication at 12 months follow-up.
Pulmonary vein isolation was associated with significantly decreased hospitalisation for cardiac causes (rate ratio 0.15, 95% CI 0.10 to 0.23; three trials).
The rate of major complications with pulmonary vein isolation was 2.6% and included tamponade, symptomatic pulmonary vein stenosis, pericardial effusion, phrenic nerve paralysis, and thromboembolic events. The rate of major complications for the patients who received antiarrhythmic medication was 8%, including pro-arrhythmia with flecainide, thyroid dysfunction secondary to amiodarone, sexual impairment, gastroenterological impairment, corneal micro-deposits, abnormal liver function tests, and sinus node dysfunction caused by amiodarone.
There was no evidence of publication bias from the analysis of the funnel plot.