Statistical synthesis of contextual knowledge to increase the effectiveness of theory-based behaviour change interventions.

Tailored implementation strategies targeting health professionals' adoption of evidence-based recommendations are currently being developed. Research has focused on how to select an appropriate theoretical base, how to use that theoretical base to explore the local context, and how to translate theoretical constructs associated with the key factors found to influence innovation adoption into feasible and tailored implementation strategies. The reasons why an intervention is thought not to have worked are often cited as being: inappropriate choice of theoretical base; unsystematic development of the implementation strategies; and a poor evidence base to guide the process. One area of implementation research that is commonly overlooked is how to synthesize the data collected in a local context in order to identify what factors to target with the implementation strategies. This is suggested to be a critical process in the development of a theory-based intervention. The potential of multilevel modelling techniques to synthesize data collected at different hierarchical levels, for example, individual attitudes and team level variables, is discussed. Future research is needed to explore further the potential of multilevel modelling for synthesizing contextual data in implementation studies, as well as techniques for synthesizing qualitative and quantitative data.