Innovative designs in behavioural trials

Clinical trials that compare pharmacological and behavioural treatments require extra attention to design on the part of the investigators. Many of the standard control mechanisms for comparison of active drug to placebo and behavioural therapy to control therapy create problems when the two types of interventions are combined. The most important of these problems is the introduction of non-specific effects introduced by behavioural therapists and physicians that can bias the study. Solutions to these problems require procedures that are common to both types of studies and the introduction of more complex statistical designs to adequately control the proposed comparisons. It may also be necessary to have a robust statistical method to address informative censoring since patients assigned to behavioural therapy may drop out of the study for different reasons than patients who drop out of a pharmacological trial. In this paper we use the design of the Raynaud's Treatment Study to demonstrate methods that can be used to control for non-specific effects and differential drop-out from the study.