Designing evaluations of interventions to change professional practice.

The design and interpretation of experimental studies to examine the effectiveness and efficiency of implementation strategies intended to change professional practice are gaining increasing attention. It has been argued that variations of the simple randomised design are most appropriate in this setting. The principles of block designs and their applicability in this field are reviewed critically, as are the choice of outcomes and approaches to economic analysis. We conclude that more complex designs are not superior per se, and that the design, choice of implementation method and topic, and analysis of such experiments should reflect the circumstances of the study.