At the 50% interim, a 50% reduction (22 deaths in each arm) would have translated to p=.012 two sided, probably not quite enough the declare the trial over at that time. But your argument is one reason I'm wild-guessing a mid-40% reduction at the end.
