Clinical trial design: the neglected problem of asymmetrical transfer in cross-over trials

The paper illustrates drawbacks of the popular cross-over design in clinical trials in psychiatry where treatment effects become distorted by, and confounded with, the effects of their order of administration. It is argued that research workers should analyse data from cross-over trials for asymmetries due to order of treatment and seriously doubt the reliability of data if such effects are found. The problem is avoided by the use of separate-groups designs.