Shifting Implementation Science Theory to Empower Primary Care Practices

We appreciate the rethinking of implementation science that Miller et al. call for in the May/June 2019 issue of Annals of Family Medicine. The authors argue that "[C]urrent implementation science-based primary care interventions risk promoting scientific imperialism by predetermining the evidence, setting the research agenda, funding ideologically driven policies, and determining engagement and communication strategies" (p251). They describe implementation science as a top-down, paternalistic endeavor.

We agree that the research-to-practice gap is too often defined by researchers or health care-related authorities. This top-down approach can be contrasted with quality improvement and its more scientific application, improvement science, in which problems are identified in local practice settings in a bottom-up approach. We agree that bottom-up approaches are necessary because the complexities of practice settings preclude universal solutions. However, we argue below that top-down approaches are a critical complement to bottom-up approaches.

The authors critique implementation science for lacking external validity: "Unstated, but assumed, is that evidence, developed and tested in a context remote from settings of implementation, is appropriate to that setting". Here, the authors describe two separate but entwined issues: identifying problems and identifying solutions. Regarding top-down approaches to identifying problems, we agree with the authors. Implementation science developed in the wake of the evidence-based movement, which popularized the notion that the best available empirically supported ("evidence-based") practices should be used to improve practice. This movement brought with it the biomedical emphasis on reductionism and causality in which the practice setting was treated as an exogenous determinant of practice improvement. However, we contend that context is endogenous to practice improvement. Thus, a social science perspective is necessary to understand the complexities of practice improvement, including the bottom-up approaches for which the authors advocate.

Regarding identifying solutions, the authors identify a specific source of "scientific imperialism": implementation science theory: "Recent comprehensive reviews of D&I and its theories...suggest an overall approach that disrespects and undervalues primary care as a co-producer of knowledge and inadvertently bullies practices into conforming to goals they did not choose." To address this, the authors propose that practitioners should "find their own solutions." We contend that this essentially encourages practitioners to "reinvent the wheel," with local practitioners identifying solutions to problems that have in fact been experienced by many other practitioners who might also benefit from these solutions. We agree in principle that practitioners have unique insight into the problems and the solutions to best address them, but in practice, each practitioner identifying his/her own solutions is an inefficient use of practitioners' scarce time. Such an approach fails to build towards an integrated body of knowledge of potentially broader relevance than the specific setting.

We argue that - instead of being the source of "scientific imperialism" - implementation science theory has great potential to utilize practitioners' unique insights while avoiding the inefficiency of a purely bottom-up approach. Specifically, theory offers a benchmark against which practitioners could identify potential solutions based on knowledge of problems accumulated from multiple practice environments. Indeed, we contend that theory is most useful when viewed as a living tool, enhanced with evidence from practice. Practice comes before theory - and theory without experience is mere intellectual play, as Immanuel Kant has reminded us. Theories are essentially assumptions made explicit, which is fundamental to understanding how and why solutions worked or not. However, to remain relevant, theories must be questioned, examined, adapted or abandoned; we may hold on to our beliefs and assumptions even if proven incorrect.

Generating "practice-based evidence" to refine theory requires planning, executing and evaluating internal quality improvement initiatives rigorously, which is essentially the ambition of improvement science. This will allow for conclusions that apply beyond individual settings, thus contributing to knowledge accumulation and building an integrated body of knowledge of benefit for more than those involved in a local initiative.

We thank the authors for articulating the relevance of bottom-up processes to problems and solutions in health care, but we contend that implementation theory does not stand in the way of this; rather, theory is a vital tool for local initiatives and science.

Competing interests: None declared