Evaluating change in professional behaviour: issues in design and analysis

Implementing the findings of research to change the behaviour of health care professionals has become an increasingly prominent issue. However, designing valid studies to evaluate different methods of achieving changes requires considerable care and there are a number of pitfalls evident from published previous work. The various steps in the development of an implementation method and issues arising are explored in this text. Aspects include conceptualisation, essential background work, a structured development process, the relative merits of randomised and non-equivalent group designs, the unit of analysis, the role of multi-level models, block designs, economic analysis, and the content or message to be disseminated. An ongoing, large, randomised trial of educational outreach visits by trained pharmacists is used to illustrate some of the issues.