We read with interest the report by Guetta et al on the development of high-degree atrioventricular block HDAVB and the frequency of permanent pacemaker (PPM) implantation in a series of 70 patients who underwent transcatheter aortic valve implantation (TAVI) with the Medtronic CoreValve ReValving System (Medtronic, Inc., Minneapolis, Minnesota). The findings presented by the investigators stress the importance of identifying predictors of the development of HDAVB in patients who underwent TAVI. The occurrence of new conduction abnormalities and new PPM implantation are not without clinical implication, as they may have potential detrimental effects on the recovery of left ventricular systolic function and the quality of life. After TAVI with the CoreValve system, new-onset HDAVB has been reported in 14% to 44% of patients, and subsequent PPM implantations have been performed in 18% to 36%. In patients treated with the Edwards Sapien prosthesis (Edwards Lifesciences, Irvine, California), rates were 0% to 12% for these 2 outcomes.

In this cohort of 70 patients, Guetta et al found an equivalent rate (36%) of new HDAVB and subsequent PPM implantation as found in previous studies. They reported that right bundle branch block at baseline and deep valve implantation (>6 mm from the lower edge of the noncoronary cusp to the ventricular end of the prosthesis) were independently associated with the development of HDAVB, as has been shown previously. These findings need to be interpreted with caution for several reasons. First, the investigators used an inappropriate strategy to assess the association between potential predictors and the primary and secondary outcomes. After univariate analysis, they selected variables with p values <0.20 for the multivariate analysis, whereas a p value <0.10 seems more appropriate. In addition, further analysis was performed using a best-subset logistic regression analysis, in which all possible combinations of predictors were fitted to create the best model. Apart from whether it is acceptable to perform multivariate regression analyses in such a small cohort of patients, there are some other issues at hand. The investigators did not state if they restricted the number of variables included in this model. This study included 25 patients with HDAVB, which should lead to a maximum of 2 or 3 variables to be included (1 variable for every event). As a result, the accuracy of the reported point estimates of the outcome and its components may be questioned, because 1 event fewer or more in 1 group could substantially affect the direction of the findings, especially considering the small sample size of the study. When assessing the calculated adjusted odds ratios and corresponding confidence intervals (the investigators’ Table 4), it is apparent that the study was underpowered and that the analysis, as performed, was not justified. Although the frequencies and predictors presented are consistent with published findings, this and other studies lack the power and appropriate statistical design to adequately interpret the point estimate and predictive factors of HDAVB and new PPM implantation after TAVI. We take liberties of addressing these issues because we have missed a critical reflection on this matter in the discussion and conclusion, notwithstanding that the findings of the investigators may be true.