CATIE Trial Redux

So, among several other things, here are some pretty problematic portions of the CATIE trial:

1. The statistical power for results involving ziprasidone is only 58%. Any comparison involving ziprasidone has just about a 60% chance of being correct.

2. The mean dose for olanzapine is the only one of any test medication that is larger than the FDA recommended maximum. Might this be the reason it came out ahead of other trial medications? All the others were within the recommended treatment dose ranges.

3. One conclusion states that there is no difference in EPS between trial medications. The authors fail to remind their readers that patients with a history of TD were never randomized into the perphenazine group.

Phase II of the trial doesn't include perphenazine because the study designers didn't expect the drug to be as efficacious as it was. This is a shame because one dose of that medicine costs pennies compared with single doses of the other medicines. They must have recognized their mistake, because the study designers have chosen to include perphenazine again in Phase III, as a third-line medicine if two antipsychotics (Phases I and II) fail.