ISCAM has a history of this. Misquoting studies and lieing about them. When caught on this he/she will not respond or at least will not respond to me. ISCAM is a complete waste of time. ISCAM is JNJ through and through.
I'm not sure that I appreciate the point you are trying to make with respect to the German study.
You could criticize the Taxus vs Cypher comparison for not having been the hypothesis that was being tested (you need to specify the question you are trying to answer BEFORE you begin the study) since you can often find at least one subgroup that performs differently between the different treatment arms. e.g. "little old ladys stented on a Tuesday had much more restenosis if they were given a Cypher". But I don't see the point you are trying to make by using your program to crunch the various powers.
Calculating power would be useful in planning a study (e.g. deciding on how many patients to enroll) to get a sense of how likely you are to correctly conclude that there is a difference when one truly exists. It is also useful after the study is completed if your conclusion is that no difference exists - because it tells you how confident you could have been about finding a difference if one had existed.
But having obtained data that leads you to conclude a difference DOES exist (i.e. using the German data to say that Cypher performed better) it is the p-value which describes the accuracy of your statement (the chance that you would falsely conclude that there is a difference when there is none), not the power.
If you embark on a low powered study (perhaps because you can't afford to enroll more patients) to prove a particular point, and you find that a difference can be shown with statistical significance, it makes you lucky - not less likely to be right.
Masterjmc007... welcome... and thanks for the sample size / power estimates.
In fairness to the article touting Cypher, it couches its conclusions w/ phrases like "may be" and "possibly" better than Taxus. Still, I agree w/ your point that the data are weak (at best). I'd expect the reviewers at JAMA to water down the conclusions still more.
As a general comment, many clinicians feel the only meaningful comparison is a 1:1 randomized study of TLR. "Late loss" is of little value to them (as restenosis requiring re-intervention is what drives them). Our JNJ friends on this board will likely disagree...
Totally wrong. 8 month angiographic and 9 month clinical data will be presented. Knowledgable rumor on the street is that 8 month results show Taxus is blown away by angiographic results with Cypher (i.e., Cypher definitively better than Taxus).
I believe (please correct me if I am off to much) 32% of the patients enrolled are diabetic. Anyone, except for the Primary Investigator, very senior management at Cordis, and the Director of the Core lab, is guessing on the results of this study. However many rumors. what I find interesting is ICAM isone of the few that has stated that this study is going to show superior results for Cypher. I have heard, and again it is just a rumor and probably based upon as much information as ICAM's, that the reason JNJ delyed the results is that wanted to get 15 month results for the diabetic arm because the 8 month results were a disaster for the Cypher, showing strong statistical advantage for Taxus. Cordis by doing the 15 month analysis will have a hard time not presenting that data, and it is suppossed to show the delta even getting worse for the Cypher! Again this is just a rumor. If true however, this would be very significant!
Found this on the web. Study should nicely highlight cypher's advantage( with a lot of long lessions.)
MIAMI, Feb 18, 2004 /PRNewswire-FirstCall via COMTEX/ -- Cordis Corporation today announced completion of patient enrollment in the first, randomized head-to-head coronary stent trial comparing its new CYPHER� Sirolimus-eluting Coronary Stent with the Taxus* Express(2)* Paclitaxel-eluting Coronary Stent. Sponsored by Cordis Corporation, a Johnson & Johnson company, the REALITY Trial is a prospective, randomized study involving 1,386 patients at 90 hospital centers in Europe, Asia and Latin America.
Investigators describe the REALITY Trial as expressly designed to compare performance differences between the two drug-eluting stents, particularly in high-risk patient group subsets, such as diabetics and patients with long lesions and small-diameter vessels. Diabetics represent 32% of the study's population.
"The REALITY trial is significant for two reasons," said Principal Investigator, Marie-Claude Morice, M.D., Head of Interventional Cardiology at the Institut Hospitalier Jaques Cartier, Massy, France. "Not only is it the first direct comparison of two drug-eluting coronary stents, but it is also the first drug-eluting stent trial to include a high percentage of diabetic patients. Until now, the subsets of diabetic patients in these trials have been very small, making it difficult to draw meaningful conclusions on the efficacy of one drug-eluting stent versus another in treating diabetic patients. The REALITY trial offers us our first opportunity to get a true picture of the relative performance of these products in a very challenging population."
According to Dennis Donohoe, M.D., Cordis vice president of Regulatory and Clinical Affairs, the introduction of drug-eluting stents has "changed the paradigm for evaluating coronary artery stents."
"In the world of drug-eluting stents, it's about which product can achieve a safe, more durable result in the widest range of patients. Cordis has clinical trial data out to four years documenting the sustained performance and safety of the CYPHER� Stent. There is no comparable body of long-term data available on the Taxus Stent."
Dr. Donohoe described sirolimus and paclitaxel as "two very different drugs with two very different mechanisms of action."
Inclusion criteria included a primary lesion of at least 15 mm in length in vessels of 2.25 to 3.0 mm in diameter, with a maximum of two treated lesions per patient. The trial endpoint is 8-month angiographic assessment of binary restenosis (reblockage).
beanyboy - As others have just posted, a small study can certainly be statistically significant if the difference between the two groups is large enough. It also makes intuitive sense that treating in-stent restenosis would offer a situation where less late loss could provide a substantial advantage. The same is true for complex lesions (where there also appears to be a Cypher advantage).
icam - do you know anything about the patients enrolled in REALITY? Was there any effort to enroll more challenging lesions to highlight any difference between the two stents?
You obviously do not understand statistics. If it takes thousands of cases to detect a difference then the difference between the devices is probably small. If the difference (superiority of one over the other) is profound (like Taxus vs Cypher) you do not need many cases to find a statistically significant difference between the two groups. In this case a striking statistical superiority was seen for TLR for Cypher compared to Taxus (8% vs. 19% with p < 0.02!) with only 200 cases. This means that the results are even more profound in demonstrating not a small difference, but a marked superiority of Cypher. I suspect you have never studied or understood statistical analysis; otherwise you would understand the implications of the findings. This study is also corroborated by the long lesion data from ASAN medical, and T-SEARCH vs. R-ESEARCH studies, which both showed marked superiority of Cypher over Taxus. REALITY Trial to be presented in March, will be one of final nails in Taxus coffin to further validate these findings.