Clinical trials design lessons from the CATIE study.

The Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) study was funded by the National Institute of Mental Health to compare the effectiveness of drugs for schizophrenia. The focus here is not on its conclusions but on the knotty issues of design and methods, in order to support appropriate clinical interpretation of the conclusions, and on using the CATIE experience to indicate directions for improvement of future clinical trials. While many of the CATIE design and implementation decisions are excellent and serve as models for future research, other decisions resulted in a study with a large study group but inadequate power. Multiple treatment interventions, unbalanced randomization within and across clinical sites, and multiple secondary outcomes are among the issues that require even more serious consideration in future large multisite clinical trials. Moreover, it is crucial to clarify whether the intent of a study is to establish superiority of some treatments or to establish equivalence, for the appropriate designs and analyses differ in these situations. If the study is designed, as was CATIE, to demonstrate some treatments' superiority, statistically nonsignificant results should not be misinterpreted as evidence of "equivalence." For establishing either superiority or equivalence, future treatment comparisons might better be designed with fewer sites, more subjects per site, fewer treatments, and fewer outcomes, in order to have the power for definitively establishing superiority or equivalence at a lower cost.