Reply

Abstract

matching) that our colleagues used to perform their post-hoc analysis. But in our study we conducted two completely different risk-adjusted analyses, one of which was entirely ignored by them. Besides the propensity score matching (n ¼ 1,040), we performed multivariate logistic regression to identify those predictors of mediastinitis (as shown in Table 2 of the article) using the entire population (n ¼ 3,118). In this analysis, the use of BITA was shown not be a risk factor for mediastinitis (odds ratio 1.64, 95% confidence interval [CI] 0.8 to 3.4, p ¼ 0.191), further supporting our finding in the propensity-matched analysis. The calculated statistical power with use of a two-sample proportion comparison for two-sided equality was 71% for the entire population (ie, there was a 29% of probability of incorrectly concluding that there was no adverse effect [mediastinitis] when actually BITA were used for myocardial revascularization).In this effort to clarify the relationship between BITA use and mediastinitis, the long-term benefit of BITA should also be included. In this respect, our experience published last year in this journal showed a 10% benefit with the use of BITA regarding long-term mortality (hazard ratio 0.71; 95% CI 0.58 to 0.87; p < 0.001) [4]. Finally, considering that mediastinitis is, indeed, a life-threatening adverse event, any increase in the incidence of mediastinitis in the BITA group must be weighed against its long-term benefit, and this is what is truly clinically important for cardiovascular surgeons.