Thank You Nuke661. This, combined with the non-answer/avoidance of my question, confirms the conclusion that the "out of bounds" remark has no underlying facts or evidence to support it, but is based solely on a subjective "feeling" at best.
. If patients have been in the trial for say 6 months, then there is no data for months 7 and above. This will require the patient be censored at month 7, these are EOT, end of trial, censors.
. If the patient status is unknown (which would have to be because they can not be found), then they will be censored. These are the LTFU, lost to follow up censors.
There is no determination of anything else. If a patient is run over by a truck on his way for his first treatment, he counts as dead, not censored.
There would be no need of an independent review board to decide on these.
The number of LTFU censors early in the trial seams to be on the high side compared to other OS studies I have seen, but I will not make a claim that they are off the charts. The cause though simply does not matter, one would rationally expect the LTFU patients to be doing worse than the others, so there would be a bias in the stats.
Since censors include both EOT (which are normal and can be quite high) and LTFU censors, the number means little. I wuld not expect any FDA guidance on this. They might have some comments on LTFU if you search around for that, but even then it will probably be general as this will vary. The FDA would certainly accept a dataset with censors such as this, but might look into the early LTFUs. Might be an interesting search going through FDA breifing docs for ODAC and see if the subject ever comes up.