I agree with your premise whole heartedly. It has been argued that the 331 may not be representative of soc population as whole. Removing either ends (longer and shorter lived) because of pseudo progessors being removed. I just want to make sure you are using the right %s. Also I came up with 9 maybe 10 per month in the last 12 months enrolled, 3 the last month, small difference. Do you have another more similar population to use for survival %s, like ict 107 control or others that accounts for removed psuedos? NWBO has more than the 331 to look at, although some are blinded(331) and some are not, they know the start and end dates of each individual patient. AVII's stats also assumes hazard event rates that are proportional without delayed separation and this underpowers his results, doesnt account for long/fat/flat tail. A soft landing if you will. In example if a placebo and treatment arm both had an mos of 18 months yet separation occured after and twice as many people were alive at 3 years in the treatment arm than this would be meaningful to patients. Stopping a trial early wouldn't capture this. Could it be they are looking at capturing landmark survival %s. If you almost double your chance of living at 2 and 3 years, I beleive this magnitude of difference would be stat sig and very simple to explain to the FDA, doctors, patients, investors, the st. Also the benefit in survival %s would presumably increase at 4, 5 and 6 years(more essentially "cured" pts). But we will never make it that far for all 331. They did say a slowing event rate.
