Critical role of bare-metal stent controls in trials of drug-eluting stents: reply

have a negligible impact on the analysis. Our findings of a more pronounced antirestenotic effect (P, 0.001) in Cypher vs. Taxus have actually raised concerns that most of the relative risk reduction apparent with Cypher was due to the severely inferior performance of Bx Velocity (a well-known thick strut BMS) when compared with the substantially superior performance of Cypher [when compared with Express (a thinner strut BMS) and Taxus, respectively]. Using results reported from Pache et al. (Figure 1A), it can now be shown that the benefits obtained from Cypher vs. Bx Velocity are far greater than those obtained from Cypher vs. thin-strut BMS, thus casting a shadow of doubt on our indirect comparative results (P, 0.0001 for statistical heterogeneity, when comparing the estimates obtained with thin-strut BMS vs. thick-strut BMS). Conversely, data recently reported from several direct head-to-head Cypher vs. Taxus randomized trials and demonstrating a lower risk of restenosis with Cypher vs. Taxus (P, 0.0001) appear largely in agreement with our previous indirect adjusted findings (dating back as early as June 2004), both at graphical inspection and consistency testing with heterogeneity and inconsistency (I) tests (Figure 1B). Nonetheless, some degree of overestimation cannot be dismissed and is likely due to the difference between thinand thick-strut BMS, as correctly pointed out by Pache et al. In conclusion, the authors should be complimented for their independent research effort and for having reminded us that any piece of new clinical knowledge can provide important and relevant insights, both by itself and when combined with appropriate analytical methods in systematic overviews.