DolphinM, it's now about software.
The data processing aspects of this type stuff are trivial. The issues will be from the human side of gathering and interpreting the raw data to generate the numbers that need be processed. In this case, doctors have to read diarries and decide the level of the patients reactions. There are many places for errors to occur before you have a nice, clean, locked dataset to be analysed.
You are correct though in that once you have this pristene data, you should be able to unblind and annalyse w/o difficulty.
What should be clear is that the problem (if it exists) had to occure before locking and unblinding, else if would easily have been fixed. It just can't take that long.
"so its better off being at one place, being respectable keeping the foul up quiet so both parties can try to figure this mess out."
But, the labs that screwed it up have seen the unbinded data, and therefore any "figure out the mess" performed by them that goes back to raw data will carry no weight with the FDA.
So where does that leave them?
The best hope is that only one trial was tainted. If so they go for a single trial aproval with supportative data from the screwed up trial and the earlier P3. The FDA allows more leaway in supportive data than in pivital trials, so here they might get away with explaining a screwup.
The only other chance I see (if the raw data really indicated efficacy, but was screwed up in collection/entry) is to get a new unblinded analysis done. To do this, you would have to go with another indpendent lab, else the FDA would really raise an eyebrow.
