CATIE Schizophrenia, Statistical Methods for The Primary Analysis, In The Form of an Updated Methods Section From The Protocol 

The primary objective of the CATIE trial was updated during the course of the trial and prior to the completion of data collection. The original primary hypothesis, specified in 1999, was that the three atypical antipsychotics, pooled together, would be superior to perphenazine, the conventional antipsychotic treatment. The second step of the primary analysis was to compare the three atypical antipsychotics to each other, given that they were together found to be superior to perphenazine.

Analyses presented in 2003, particularly a meta analysis of 124 randomized controlled trials that examined efficacy (Davis et al 2003), suggested that the atypical antipsychotics are not a homogeneous group, and found that not all of them were substantially different from conventionals such as perphenazine. Therefore, in Spring 2004, the primary hypothesis of the CATIE trial was modified to directly compare all four treatment groups. The updated primary hypothesis is that there are some differences between the four treatment groups. The second step of the primary analysis, comparing the three atypicals, remains the same, and is to be investigated assuming the primary analysis finds an overall difference between the four treatments.

Excerpt