Incorporating a Shortened Long-Term Effectiveness End Point Into Catheter Ablation Trials: Statistical Properties and Implications

Evidence regarding the use of a 6-month effectiveness end point for trials of atrial fibrillation ablation is limited. This study aims to evaluate the relationship between ablation effectiveness end point detection and follow-up duration, with the secondary goal of informing the design of clinical trials of paroxysmal atrial fibrillation ablation.

Patient-level data from 4 prospective studies were aggregated. Long-term effectiveness end points were calculated with (1) a 3-month blanking period and 12-month follow-up (current) and (2) a 2-month blanking period and 6-month follow-up (proposed). Sensitivity, specificity, and agreement were calculated for each study under the current and proposed end points. A linear approximation method was used to extrapolate the 6-month survival estimate to 12 months, and the predicted 12-month survival rates were compared with the actual 12-month survival rates. Using the linear approximation, a new performance goal was derived and compared against the 6-month survival of each study.

The comparative analysis between the current and proposed end points demonstrated concordances ranging from 83% to 92% and sensitivities ranging from 57% to 76% across the 4 recent clinical trials. The linear approximation method produced an estimated 12-month survival rate of the 4 pooled studies of 72.5%, compared with the true 12-month survival rate of 71.0% (-1.5% error). When compared against the true 12-month survival rate in each study individually, the estimated 12-month survival rate resulted in errors ranging from -3.6% to 0.3%. A new performance goal of 56.2% is proposed for trials using a 6-month end point.

The proposed end point of a 6-month follow-up duration, inclusive of a 2-month blanking period, demonstrated reasonable levels of agreement with the current 12-month effectiveness end point. Enabling historical comparison, a 12-month survival rate can be reliably extrapolated from a 6-month survival rate using linear approximation.