Replies to post #3761 on Dendreon Corp fka DNDNQ

[stuck_holder]

<They can't use longer-term data than the specified three years for 9901/9902A because it wasn't in the protocol.>

i think i know the general gist of the answer to my question, but i'm going to pose it anyways since i'm not sure i really understand the underlying logic...

i understand that the FDA essentially ignores any data not collected as per the agreed upon trial protocol for fairly obvious reasons, (ie there are no controls over data collection not subject to a formal protocol & what may have influenced it, so its value in establishing & measuring the treatment effect is unreliable) however, in the case of a trial designed to test treatments for & on terminally ill patients, wouldn't the length of survival be an important & reasonably objective consideration, even if looked at "after the fact"?

i know this is probably a moot point from the point of view that since the FDA disregards/dismisses it out of hand (at least in terms of evaluating efficacy), but why wouldn't/shouldn't it be legitimately given some weight (depending on the extent of being able to measure & actually measuring potential contributing factors) to the efficacy question? is it just too subject to "manipulation" (ie. cherry picking) or alternatively not susceptible enough to rigorous and rigid analysis that statisticians seem to defend to the death?