Doesn't matter if it's centralized or single-payer or whatever.
Every unblinded interim analysis has to be pre-specified, with a hit to final alpha, or it is a meaningless analysis.
Otherwise, why not run the statistical analysis every time someone completes the trial?
For example: 3 patients complete the trial. Oh, look: 2 of the 3 patients improved. Stop the trial! This drug works! We don't need to see the other 400 patients.
That would be an absurd way to run a trial, and an even worse way to approve a drug.
This is why each interim analysis has to be pre-specified, and why there is an alpha hit for each interim analysis.
The easy questions are: how many interim statistical analyses will there be? What events trigger each interim anaysis? What is the alpha spend on each interim analysis?