How do registries differ from randomized controlled trials (RCTs)? Unlike RCTs, registries are not built on randomization and therefore may at times be subject to case selection. Furthermore, it is often difficult to draw a conclusion from registries since they lack a good control group to compare outcomes. In addition, patients in RCTs are usually much more highly scrutinized in terms of angiographic follow-up. On the other hand, registries usually have much broader inclusion criteria than RCTs, meaning they capture a more realistic picture of the complex, diverse patient population seen in the cath lab on a daily basis. Because of their strict exclusion criteria, most RCTs include only the basic, de novo, vanilla-type lesions. Therefore, it’s very helpful to review this data along with RCTs to gain insight into how devices perform in these patient populations. What are the important elements of a good registry? Good registries require rigorous data collection and follow-up. It is important that records are accurate. The challenge is to get a true follow-up on these patients, especially when angiographic analysis is not included. What valuable information can interventionalists gather from registries that cannot be gathered from RCTs? As I mentioned, registries are much more inclusive than RCTs. Thus, they allow us to evaluate the efficacy and safety of a device in more complex, high-risk subsets of patients and a wider range of clinical situations that reflect everyday practice. What type of information can head-to-head registries provide? What do you need to take into account when evaluating them? Head-to-head registries can provide useful information. With thousands of patients, you can compensate to some extent for the lack of true randomization and selection bias. For example, in the S.T.E.N.T. and REWARD registries, you see that one drug-eluting stent (DES) was chosen more frequently than the other in more difficult subsets of patients. In order to fairly judge whether there are true differences between the stents, you have to adjust for differences in baseline characteristics between the subsets and the complexity of the groups in which each stent was selected. The Research/T-Search registry was smaller but has some value because it was conducted as two consecutive series, one of them using just the Cypher® stent and the other just the TAXUS® stent. In this case, the stent choice was mandated and not left to operator discretion. Can you talk about the importance of obtaining information on deliverability from registries? Registries test the ability to deliver a stent in circumstances and lesions you might not encounter in RCTs calcified lesions, tortuous vessels and so forth. Obviously, deliverability can be judged more easily with a registry than with RCTs, because at the end of the day if the stent has not been delivered, these patients cannot enter an RCT at all. In registries, you will see that a stent was either primarily selected by the operator, or there was a switch because the other stent enabled better deliverability. However, one thing that registries and RCTs do not capture is the effort needed to deliver the stent. Insight into that issue can be indirectly derived by looking at the baseline characteristics and the complexity of the lesions. As I mentioned, investigators in the S.T.E.N.T. and REWARD registries consistently selected one stent over the other for more complex lesions. This reveals either the perception or the situation that led the operator to use one stent versus the other in the more complex patients. Because stent selection occurs at operator discretion versus mandated by protocol, I think you can draw a conclusion as to why one stent is used over another in difficult lesion subsets and patient subgroups, which can make registry data interesting and useful. How has your practice changed with the introduction of DES? What role has RCT and registry data played in those changes? Initial data from RCTs supported the approval and adoption of DES. Within a few months, there was 87% penetration of DES in cath labs. There have been some concerns, such as the issue of sub-acute and late thrombosis, especially in acute myocardial infarction patients. These reports also influence practice in the cath lab. Another interesting issue is the decision of which DES to use. Early on, RCT and registry data convinced some physicians in our lab to use one stent versus the other. But as time has gone by and as more information has been gathered, I think physicians have become more settled and confident in terms of the performance of the stents. Nevertheless, physicians are still paying attention to registry data in this phase of product development, since registries are the only means to study a large number of patients. TAXUS is a trademark of Boston Scientific Corporation or its affiliates. Cypher is a trademark of Cordis Corporation. Sponsored by Boston Scientific.