Strategies are central to the National Institutes of Health’s (NIH’s) definition of implementation research as “the study of strategies to integrate evidence-based interventions into specific settings” [1]. Multiple papers have been written proposing frameworks and lists to describe the strategies used in implementation research and practice [2, 3]. In a recent review, Lokker et al. [4] identified 23 models, frameworks, and taxonomies for classifying the type of strategies used “to promote and integrate evidence into practice in healthcare.” These strategies increasingly are being classified under the single term “implementation strategies” [2]. We contend that classifying all strategies under a single term may constrain efforts to synthesize findings across tests of strategy effectiveness and limit advancement of the full range of strategies of importance to implementation science and practice.

The ability to synthesize findings across studies is essential to building the evidence base for what, how, and when implementation strategies work to improve which implementation outcomes. Despite the valuable work that has been done to name and define strategies, efforts to synthesize findings across studies continue to be challenging [4]. Names and definitions do not adequately differentiate who enacted the strategies or for what purpose, resulting in findings that are not “readily comparable” or amendable to the synthesis needed to build the evidence base for their use [5].

We illustrate the problem with an example from our research. Two authors (JL, CR) are collaborating on a study testing “implementation strategies” that practice facilitators (e.g., coaches) provide to primary care clinic staff to build the staffs’ capacity to use “implementation strategies” to integrate cancer screening evidence-based interventions (EBIs) into their practice setting. The broad use of the term “implementation strategy” creates confusion for our team because it does not distinguish what the practice facilitators are doing to support implementation from what primary care staffs are doing to integrate EBIs into routine practice. We contend that these distinctions are important when selecting and reporting strategies for use in research and practice because the evidence base and, to a large extent, the underlying theory, differ for strategies used to support implementation (e.g., [6, 7]) as compared to strategies used to integrate screening EBIs into practice (e.g., [8, 9]).

The use of the single term “implementation strategy” also increases the risk that some of the strategies essential to implementation science will be overlooked. Those using the term “implementation strategy” tend to define it in relation to the implementation of a specific intervention or guideline. Powell et al. [2], for example, define “implementation strategies” as “methods or techniques used to enhance the adoption, implementation, and sustainability of a clinical program or practice.” Similarly, Mazza et al. [3] define an “implementation strategy” as a “purposeful procedure to achieve clinical practice compliance with a guideline recommendation.” These definitions suggest that implementation begins with a predefined “program or practice” or “guideline” and have the potential to perpetuate the field’s historic focus on a “push” model of implementation that begins with the identification of a specific evidence-based intervention (EBI) that is then “pushed” into practice [10]. Many scholars have called for more research that applies a “pull” model, with the goal of building practice-level capacity to select, adapt, and implement the EBIs they need to address locally identified needs [10, 11]. We contend that to achieve this goal, implementation science needs to invest more in strategies that build practice-level capacity to prioritize areas in need of improvement and to select (i.e., pull) from the menus of available EBIs. The creation, promotion, and distribution of these EBI menus require dissemination strategies—yet another set of strategies of importance to implementation.

To address these challenges (among others), Proctor et al. [5] recommended that authors not only name and define their implementation strategies but also describe how they operationalized the strategy in their study. Specifically, they recommended specifying who enacted the strategy (i.e., the actor), the level and determinants that were targeted (action targets), and the intended implementation outcomes. Clearly delineating how strategies are operationalized is critical to reporting findings in ways that are amenable to synthesis. Using a more consistent approach to operationalizing strategies also has potential to foreground the full range of strategies used in implementation science.

Building on the recommendations of Proctor et al. [5], we propose a system for classifying implementation strategies based on the strategies’ actor and action targets. (Table 1 provides an overview of definitions for these and other terms used in this paper.) Structuring strategies into classes will aid in aligning strategies with relevant theories (i.e., those that pertain to related determinants), synthesizing findings across studies, and identifying potential gaps in current strategy listings. Organizing strategies into classes also will assist users in locating the strategies that best match their needs. We are not suggesting a new taxonomy but rather a system for classifying implementation strategies.Table 1

Key terms and their definitions

As further described below, identifying actors and action targets resulted in five conceptually distinct classes of strategies: dissemination, implementation process, integration, capacity-building, and scale-up (See Fig. 1 and Table 1). In our descriptions of each class, we identify the ISF system actors that enacted the strategy (actors), levels and determinants targeted (action targets), and outcomes used to assess strategy effectiveness. We illustrate how each class would apply to strategies to improve colorectal cancer (CRC) screening rates in Federally Qualified Health Centers (FQHCs). Because EBI adoption and implementation is the central goal of implementation science and practice, we begin with our definition of EBIs.Fig. 1

Classes of implementation strategies organized within the Interactive System Framework [12]. The bi-directional arrows represent the importance of communication across levels

Existing taxonomies provide both names and definitions for their lists of implementation strategies. Building on Proctor et al.’s [5] reporting guidelines, the proposed classification system adds greater conceptual clarity to these definitions by classifying strategies according to who enacts them (actor) and the levels and determinants they target (action targets). These distinctions will aid communication among those engaged in both implementation science and practice, who will now be able to identify their need for or use of implementation strategies as aligning with one (or more) of the five classes. Placing strategies into five broad classes will also assist those who are seeking implementation strategies, whether for research or practice. Accordingly, Table 2 provides references for lists of strategies that are applicable to each class. Finally, the classifications will assist systematic reviewers in aggregating findings across implementation strategies by providing an additional scheme for identifying studies with comparable findings.Table 2

Five classifications for implementation strategies

Consistent with the intent of Proctor et al.’s guidelines, this classification system’s focus on targets of action (levels and determinants) has potential to aid in selecting and strategically combining multiple strategies [45, 46]. For example, by applying intervention mapping methods, implementation strategies could be selected to match determinants across the CFIR levels of the individual, inner setting, and outer setting [47]. By identifying the targeted determinants and measuring strategies’ effects on those determinants, implementation researchers can begin to disentangle the contributions of discrete strategies when multiple strategies are used in combination. They also could advance understanding of the mechanisms through which implementation strategies influence EBI adoption, implementation, scale-up, and sustainment by facilitating the alignment of strategies to theories that explain how they work [46].

This classification is intended to be broadly applicable to EBIs that target changes in individuals, populations, environments, and policy. Although our illustration applied the classification to a clinical EBIs, we easily could have illustrated their applicability to implementation strategies related to policy EBIs (e.g., to regulate tobacco marketing) or to environmental EBIs (e.g., to increase access to healthy foods and beverages).

The next steps for developing this framework might include mapping it to existing taxonomies, such as those developed by Powell et al. [3] and Mazza et al. [4]. Subsequently, a review of the literature and/or Delphi approach might be applied to generate lists of strategies for classes that may be underrepresented in existing taxonomies (e.g., scale-up strategies or dissemination strategies). Further work might also explore the need for additional classes. For example, are different strategies used to de-escalate or to sustain an EBI within a specific setting beyond those that would be listed as integration strategies? Additionally, there may be a need to identify and refine taxonomies of outcomes (like those identified by Proctor et al. [37]) that are relevant for each of these classes (e.g., dissemination, scale-up, etc.).

This proposed classification system reflects the ongoing efforts of implementation scientists to develop clearer and more meaningful ways of communicating about the strategies that are crucial to moving EBIs into real-world practice. Structuring strategies into classes has potential to add and facilitate clearer reporting of implementation research findings, alignment of strategies with relevant theories, synthesis of findings across studies, and identification of potential gaps in current strategy listings. Organizing strategies into classes also will assist users in locating the strategies that best match their needs.