Background Embedded approaches to implementation research (IR), whereby health system decision-makers participate actively in the research process, are gaining traction as effective approaches to optimise the delivery of health programmes and policies. However, the evidence base on the processes and effectiveness of such collaborative research remains inchoate. Standardised approaches to evaluate these initiatives are needed to identify core elements of ‘embeddedness’, unveil the underlying pathways of change, and assess contribution to evidence uptake in decision-making and overall outcomes of effect. The framework presented in this paper responds to this need, designed to guide the systematic evaluation of embedded IR. Methods This evaluation framework for embedded IR approaches is based on the experience of a joint initiative by the Pan American Health Organization/Alliance for Health Policy and Systems Research, which has supported 19 IR grants in 10 Latin American and Caribbean countries from 2014 to 2017. The conceptualisation of this framework drew on various sources of information, including empirical evidence and conceptual insights from the literature, interviews with content experts, and a prospective evaluation of the 2016 cohort that included semi-structured key informant interviews, document analysis, and a research team survey to examine key aspects of embedded research. Results We developed a widely applicable conceptual framework to guide the evaluation of embedded IR in various contexts. Focused on uncovering how this collaborative research approach influences programme improvement, it outlines expected processes and intermediate outcomes. It also highlights constructs with which to assess ‘embeddedness’ as well as critical contextual factors. The framework is intended to provide a structure by which to systematically examine such embedded research initiatives, proposing three key stages of evidence-informed decision-making – co-production of evidence, engagement with research, and enactment of programme changes. Conclusion Rigorous evaluation of embedded IR is needed to build the evidence on its processes and effectiveness in influencing decision-making. The evaluation framework presented here addresses this gap with consideration of the complexity of such efforts. Its applicability to similar initiatives is bolstered by virtue of being founded on real-world experience; its potential to contribute to a nuanced understanding of embedded IR is significant.

Innovative strategies are needed to improve the delivery of evidence-informed health interventions. Embedded implementation research (EIR) seeks to enhance the generation and use of evidence for programme improvement through four core features: (1) central involvement of programme/policy decision-makers in the research cycle; (2) collaborative research partnerships; (3) positioning research within programme processes and (4) research focused on implementation. This paper examines how these features influence evidence-to-action processes and explores how they are operationalized, their effects and supporting conditions needed. We used a qualitative, comparative case study approach, drawing on document analysis and semi-structured interviews across multiple informant groups, to examine three EIR projects in Bolivia, Colombia and the Dominican Republic. Our findings are presented according to the four core EIR features. The central involvement of decision-makers in EIR was enhanced by decision-maker authority over the programme studied, professional networks and critical reflection. Strong research–practice partnerships were facilitated by commitment, a clear and shared purpose and representation of diverse perspectives. Evidence around positioning research within programme processes was less conclusive; however, as all three cases made significant advances in research use and programme improvement, this feature of EIR may be less critical than others, depending on specific circumstances. Finally, a research focus on implementation demanded proactive engagement by decision-makers in conceptualizing the research and identifying opportunities for direct action by decision-makers. As the EIR approach is a novel approach in these low-resource settings, key supports are needed to build capacity of health sector stakeholders and create an enabling environment through system-level strategies. Key implications for such supports include: promoting EIR and creating incentives for decision-makers to engage in it, establishing structures or mechanisms to facilitate decision-maker involvement, allocating funds for EIR, and developing guidance for EIR practitioners.

Background To achieve global health targets, innovative approaches are needed to strengthen the implementation of efficacious interventions. New approaches in implementation research that bring together health system decision-makers alongside researchers to collaboratively design, produce and apply research evidence are gaining traction. Embedded implementation research (EIR) approaches led by decision-maker principal investigators (DM PIs) appear promising in this regard. Our aim is to describe the strategies study teams employ in the post-research phase of EIR to promote evidence-informed programme or policy improvement. Methods We conducted a prospective, comparative case study of an EIR initiative in Bolivia, Colombia and Dominican Republic. Guided by a conceptual framework on EIR, we used semi-structured key informant interviews (n = 51) and document reviews (n = 20) to examine three decision-maker-led study teams (“cases”). Focusing on three processes (communication/dissemination, stakeholder engagement with evidence, integrating evidence in decision-making) and the main outcome (enacting improvements), we used thematic analysis to identify associated strategies and enabling or hindering factors. Results Across cases, we observed diverse strategies, shaped substantially by whether the DM PI was positioned to lead the response to study findings within their sphere of work. We found two primary change pathways: (1) DM PIs implement remedial measures directly, and (2) DM PIs seek to influence other stakeholders to respond to study findings. Throughout the post-research phase, EIR teams adapted research use strategies based on the evolving context. Conclusions EIR led by well-positioned DM PIs can facilitate impactful research translation efforts. We draw lessons around the importance of (1) understanding DM PI positionality, (2) ongoing assessment of the evolving context and stakeholders and (3) iterative adaptation to dynamic, uncertain circumstances. Findings may guide EIR practitioners in planning and conducting fit-for-purpose and context-sensitive strategies to advance the use of evidence for programme improvement.