The challenge of transferring an implementation strategy from academia to the field A process evaluation of local quality improvement collaboratives in Dutch primary care using the normalization process theory

Rationale, aims and objectives Aquality improvement strategy consisting of comparative feedback and peer review embedded in available local quality improvement collaboratives proved to be effective in changing the test-ordering behaviour of general practitioners. However, implementing this strategy was problematic. We aimed for large-scale implementation of an adapted strategy covering both test ordering and prescribing performance. Because we failed to achieve large-scale implementation, the aim of this study was to describe and analyse the challenges of the transferring process. Methods In a qualitative study 19 regional health officers, pharmacists, laboratory specialists and general practitioners were interviewed within 6 months after the transfer period. The interviews were audiotaped, transcribed and independently coded by two of the authors. The codes were matched to the dimensions of the normalization process theory. Results The general idea of the strategy was widely supported, but generating the feedback was more complex than expected and the need for external support after transfer of the strategy remained high because participants did not assume responsibility for the work and the distribution of resources that came with it. Conclusion Evidence on effectiveness, a national infrastructure for these collaboratives and a general positive attitude were not sufficient for normalization. Thinking about managing large databases, responsibility for tasks and distribution of resources should start as early as possible when planning complex quality improvement strategies. Merely exploring the barriers and facilitators experienced in a preceding trial is not sufficient. Although multifaceted implementation strategies to change professional behaviour are attractive, their inherent complexity is also a pitfall for large-scale implementation.