arbitrary as correctly stated by Moreno et al., in their current letter, came from previous suggestions and analyses done by experts (as we stated in our methods section) and it was a predefined form of subgroup analysis to assess whether the main result was also maintained in the subgroups, or there were different trends, as previously explained. Indeed, we also performed other subgroup analyses, but they were all burdened by significant heterogeneity. Moreover, the references 4 and 5 used by

more »

... oreno et al. in their current letter refer to two studies performed in coronary vessels with a large range of vessel size and not only small ones, thus the cut-off of 35% after angioplasty as residual stenosis indicating suboptimal result may not correctly apply to small vessels. Finally, in our opinion, the intention-to-treat analysis (including cross-over from balloon angioplasty to stent, considered in each trial as bail-out and not provisional) was the best approach as we did not want to simply compare stenting with balloon, but we would like to understand why there were different results in trials with similar design and whether the possible benefit of routine stenting was real or could be balanced by a strategy of provisional stenting. The real value of a meta-analysis relies on the correct selection of data and also on the right statistical methods used. 6, 7 The best approach to reconcile Moreno's and our work would be an individual patient data meta-analysis, scientifically intriguing, but logistically and economically demanding. 7