Instead of weakening, the crossover strengthens the case that 9902a supports 9901. The fact that there were more patients who crossed over in 9902a than in 9901, helps explain why the result is not as statistically significant as anticipated, in addition to 9902a enrolled disportionaately worse prognostic to the placebo arm, and the small sample size.
<But the results deal more with whether early Provenge is better than later Provenge (as a result of the crossovers).>
Don't agree with statement, the statistical hypothesis testing is made intact, because if the null hypothesis is true, that provenge has no effect, then you will not see any difference in survival. But this hypothesis was rejected.
The reason this kind of built-in crossover design was not done more often was that aside from ethics, it weakens the power to reject the null hypothesis.
If one could show patients are worse off by a delay in administration of provenge
IF delaying administration , whenever there is a built-in crossover, it becomes a test
News 
Market Data 
Discover 
AI
