Yep. Differences are that they are using n = 51, while I used n = 75. Also, they are using normal approximation to binomial and I am using logit. These differences will make for slight differences in the results (maybe pretty good-sized differences on very small or very large mortality rates) but nothing that matters very much...we don't know what model they'll use.

The differences that do matter are 1) 1-sample vs. 2-sample tests, which has a big influence on results (and they will definitely be using a 2-sample test), and 2) one-tail or two-tail test (and they will most likely be using a one-tail test). Also, sample size matters. Hopefully we are way above 50 by the time they cut the trial off, but we might not be.