The time has come: Embedded implementation research for health care improvement

The field of implementation science has developed in response to slow and inconsistent translation of evidence into practice. Despite utilizing increasingly sophisticated approaches to implementation, including applying a complexity science lens and conducting realist evaluations, challenges remain to getting the kinds of outcomes hoped for by implementation efforts. These include gaining access and buy-in from those implementing the change and accounting for the influence of local context. One emerging approach to address these challenges is embedded implementation research-a collaborative, adaptive approach to improvement. It involves researchers and implementers working together in situ from the outset of, and throughout, an implementation project. Both groups can benefit from the collaboration: it increases the rigor of evaluation, provides opportunities to improve the intervention through direct feedback, and promotes better on-the-ground understanding of the change process. We aimed to examine the potential benefits, and some of the challenges, of increased embeddedness.

We performed a multi-case analysis of implementation research projects that varied by degree of embeddedness.

Embedded implementation research may offer a range of advantages over dichotomized research-practice designs, including better understanding of local context and direct feedback to improve the implementation along the way. We present a model that spans four approaches: dichotomized research-practice, collaborative linking-up, partially-embedded, and deep immersion.

Embedded implementation research approaches hold promise in comparison to traditional dichotomized-research practice designs, where the research is external to the implementation and conducts a summative evaluation. We are only beginning to understand how such partnerships operate in practice and what makes them successful. Our analysis suggests the time has come to consider such approaches.