1. Since they are using the smaller per protocol completers (PP)rather than the expected ITT population I think n should be 161 for placebo and n=301 for treated. 2. The PP cohort is a subset of the whole subset --- the means on 16 are calculated from the entire n=170 and n=338 population while the means on 21 are from the smaller n=161 and n=301 PP cohort
Big question is did the data - either combined, 30 mg or 50 mg hit the endpoint using the ITT or mITT cohorts