The primary endpoint per protocol accoridng to the company was set at 95% CI lower limit of 30%.
The trial as called early ended witha LL 5.3%, missing the pre-specified by 24.7%.
With the added alpha penatly (for an early unblind) , the trial as called missed by more than 30%.
Had the trial run to latest stated full enrollment (8616pts), and all else being equal, then the result would have been LL 17.9%, missing the primary endpoint 12.1%.
But all else would not have been equal. The data is far from uniform. Not only did they 'trim the trial, not the fat' but trial execution was only kinda a global trial. It was more of a global travelling trial, one country one season, another country another season.
The interim analysis was a 53% point efficacy. The added data after that, mostly USA, was negative efficacy data. Adding more negative efficacy data does not save this thing. So they trimmed the trial, and also during execution ... they trimmed the implementation (true multi-national multi-season being more expensive than the travelling global road show thing they did).
They stated the follow on data had ARs comparable to the interim data. It Did Not.
Never been fond of a non-scientist running things at this stage, and never been impressed by Stan. But tell me this: Where the hell was Young?