Five-year outcomes of sirolimus-eluting and paclitaxel-eluting stents in routine clinical practice: are they equally good (or bad)?

Randomized clinical trials of new coronary devices, as with other novel therapies, are generally conducted in a homogeneous and usually low-risk population. There is valid statistical and methodological justification for this approach, including lower variances (resulting in increased statistical power) and increased probability that observed differences are due only to the study therapy. Although such trial designs generally increase the chances of a positive result, they result in a limited ability to generalize the findings to the broader population of patients encountered in routine clinical practice. Furthermore, the quality of evidence required for reasonable assurance of safety and effectiveness, the benchmark for regulatory approval of a new device, is necessarily in equilibrium with the desire to bring novel designs to market and meet the clinical needs of the population for whom the therapy may be beneficial as expeditiously as possible. The resulting limited complexity of the study population and relatively short duration of follow-up after study therapy increase the risk of failing to detect important safety concerns in the premarket period.