Dean L. Fixsen, Karen A. Blase, Sandra F. Naoom, Frances Wallace
5 min
Despite decades of research aimed at developing evidence-based programs in fields like child welfare, mental health, and education, the actual delivery of these services remains inconsistent and often ineffective. The authors argue that the primary reason for this failure is not a lack of scientific knowledge, but a lack of effective implementation. While manufactured goods can have quality built into the product itself, human services rely on the practitioner as the intervention. Therefore, the challenge is to build scientific quality into the daily performance of millions of individual practitioners.
Historically, the process of moving research into practice has been viewed as a passive, "let it happen" approach involving the simple dissemination of information. The authors contend that this is insufficient. Instead, they advocate for an active, "make it happen" approach. This requires a dedicated implementation infrastructure that treats the process of adoption as a distinct scientific endeavor, moving beyond mere awareness to active support and organizational change.
To achieve high-fidelity implementation, the authors identify seven interactive "drivers" that must be integrated into the organization:
These components are compensatory; a weakness in one area, such as limited training, can often be mitigated by strengths in others, such as intensive coaching. The authors emphasize that these components are not new individually, but their integration into a cohesive, intentional implementation strategy is the missing link in the science-to-service chain.
The failure of better science to readily produce better services has led to increasing interest in the science and practice of implementation. The results of recent reviews of implementation literature and best practices are summarized in this article. Two frameworks related to implementation stages and core implementation components are described and presented as critical links in the science to service chain. It is posited that careful attention to these frameworks can more rapidly advance research and practice in this complex and fascinating area.
Sam: That explains the drift problem. Programs that look effective in year one often look quite different by year three, not because the model changed on paper, but because the implementation infrastructure quietly eroded.
Alex: Exactly. And the authors are direct about the implication: an effective program implemented poorly produces no benefit. That's not a rhetorical point — it's the central empirical claim the review is built around. Which is why they argue implementation has to be treated as a rigorous, multi-year, recursive process, not a one-time rollout event. [[RP_SECTION:critique-of-implementation-models|Critique of Implementation Models]]
Sam: Where does a careful reader push back on that framing?
Alex: A few places. First, the seven-driver model is synthesized from the literature rather than derived from a controlled trial, so the causal weight assigned to any individual component is harder to establish than the framework implies. You can observe that programs with stronger implementation infrastructure tend to produce better outcomes — but isolating which driver is doing the work is genuinely difficult.
Sam: And the compensatory logic — the idea that a strong coaching system can offset weak initial selection — that's more theoretical than empirically tested?
Alex: Largely, yes. The review makes a compelling conceptual case, but the field doesn't yet have the factorial designs that would let you quantify those tradeoffs rigorously. What the paper does well is reframe the question: instead of asking "does this intervention work?", it pushes researchers and practitioners to ask "under what implementation conditions does this intervention work, and what does it cost to maintain those conditions?"
Sam: That's a meaningful shift in how you'd design a dissemination study. You'd need to instrument the implementation process itself, not just measure outcomes at the end. [[RP_SECTION:future-research-directions|Future Research Directions]]
Alex: Right — and that's where a lot of current implementation science is heading. Measuring fidelity, tracking driver components over time, treating implementation as the independent variable rather than an assumed constant. The Fixsen review is essentially the theoretical scaffolding that makes that research agenda coherent.
Sam: So the practical takeaway for someone designing a scale-up study is: don't treat implementation as background noise. Treat it as a variable you have to measure and actively manage, or your outcome data won't be interpretable.
Alex: That's it exactly. The intervention and its implementation are not separable. If you don't engineer the conditions under which practitioners can actually deliver the program with fidelity — selection, coaching, data feedback, administrative support — you're not testing the intervention. You're testing whatever happens to emerge when you hand people a manual and walk away.
Sam: Which is a much harder research problem, but probably the right one to be solving.
Alex: And the one the field has been deferring for too long. Thanks for listening to ResearchPod.