Our objectives in the present synthesis were to understand how PARIHS has been used in implementation studies, how it has been operationalized, and the strengths and limitations of PARIHS and its supporting literature. We found a reasonably large published literature (33 published papers, 18 of which were empirical), but this is a body of findings that reflects many of the current limitations of the broader implementation science literature. These limitations provide great opportunities for improvement, notably three.
First, PARIHS was largely used and operationalized as an organizing device or heuristic, usually post hoc
. However, PARIHS developers intended the framework to be used to assess evidence and context prior to implementation and then using these findings to guide facilitation of implementation. To move the framework forward, we need empirical studies that use PARIHS to prospectively design or comprehensively evaluate implementation activities. Researchers should explain the degree to which intervention design decisions and change strategies are based on PARIHS. The lack of prospective implementation studies is not unique to PARIHS; all but a fraction of published implementation studies fail to explicitly use any theory at all [43
], so researchers do not appear to be conducting prospective implementation studies based on any conceptual frameworks; a similar lack of theoretical foundation is reported among studies of organizational factors linked to patient safety [45
]. Our findings echo those of Kajermo and colleagues in a recent literature synthesis on use of the BARRIER scale, which is intended to prospectively identify barriers to research use by nurses [46
]. Based on a paucity of prospective studies, they concluded that no further descriptive studies should be done, and that only prospective studies would move the science forward. We extend the same call for studies using PARIHS.
Second, though a strength of the empirical literature was that some studies showed empirical support for PARIHS, this finding needs to be interpreted in light of the overall study designs, which were retrospective case reports or cross-sectional analyses, and often lacked key methodological details. Furthermore, authors rarely contrasted findings to previous studies; the citation of prior work using PARIHS occurred almost exclusively in the introduction to set the stage for the study or conceptual rationale of the study. This too, may in part, be a function of the current development of the implementation science literature, and the natural evolution of standards and expectations about what details researchers most need to report. It may be time for something akin to CONSORT [47
] or MOOSE [48
] guidelines for reporting results of implementation intervention studies or implementation project evaluations. While implementation science may not be amenable to the same manner of checklists that have been applied to randomized trials and meta-analyses, there are key elements that could be described in sufficient specificity to provide guidance to both journal editors and researchers. These might include an explanation or rationale for mapping study findings to the constructs of the conceptual framework being used; a rationale for excluding certain elements; details about operationalization of constructs, including coding definitions for qualitative analyses; and discussion of the criteria authors use to draw conclusions about relationships between determinants and implementation outcomes. This might help address a key criticism of efforts to promote more theory-based implementation research, namely that translation of theory into intervention design is too subjective and opaque [49
Finally, there are opportunities to improve the conceptual clarity of the framework itself, including refining conceptual definitions to more clearly draw distinctions among related sub-elements, such as receptive context, leadership, and culture. This will help provide for more rigorous studies by making it easier for users to map measures back to PARIHS consistently, derive testable hypotheses using the framework, and design more effective implementation strategies. We have drafted an implementation guide, being published separately, which discusses in more detail recommendations for those using PARIHS in task-oriented implementation projects and research, or seeking to refine the framework. Below, we briefly discuss three specific opportunities to refine the PARIHS framework.
First, PARIHS acknowledges the dynamic relationships among elements and sub-elements in the framework and the often unpredictable nature of implementation. However, dynamic implies that elements/sub-elements interact or act as modifiers or contingencies, such that the effects of one is dependent on others [50
]. As a result, the same implementation intervention may have wildly different effects in different settings [51
]. PARIHS would be strengthened even more by beginning to describe how those dynamics might emerge and provide examples that could eventually help identify more generalizable patterns. Identifying and describing all potential interactions is clearly impossible, but currently, PARIHS elements are described on a continuum, low to high, that strongly implies linear relationships, which are inconsistent both with the broader concept of PARIHS as a dynamic model and with available evidence. For example, we have prospective studies that find senior leadership support changes dramatically over time, with senior leaders shifting among roles ranging from institutional mentors for the change to critics of it [52
]; and that senior leadership support is not always a strong driver and certainly not always a necessary condition for implementation [53
]. It may be possible to identify generalizable contextual interactions, such as senior leadership support being necessary for EBPs that involve coordination across departments or services, require large capital investments or lack strong professional endorsement.
In part, the lack of specifics about interactions among elements may arise from PARIHS straddling the line between a higher order planned action (or prescriptive) theory (PAT) for use by change agents to guide their implementation strategy, and a classical (or descriptive/explanatory) model meant to describe or explain how change occurs. The core concept articles explicitly propose that PARIHS be used to guide implementation by assessing evidence and context in order to inform facilitation, strongly positioning PARIHS as a prescriptive model, albeit not with the detail of a PAT as described by Graham and Tetroe [53
Second, we also noted that a more explicit definition for 'successful implementation' is needed. This again is both a key strength of the framework and an opportunity to strengthen it. A clear definition of successful implementation is critical for moving implementation science literature forward, and we may do well to draw on the literatures of other disciplines. For example, researchers in education [55
] and health promotion [56
] have written specifically about criteria for determining when new programs are fully implemented. Likewise, scholars in management have written about conceptual considerations for defining effective implementation of new practices such as IT systems [57
] and banking practices [58
], including distinguishing implementation from 'compliant' use that is either incomplete or likely to degrade.
Conceptually, successful implementation might comprise three distinct aspects, identified as part of our aforementioned implementation Guide. All represent seemingly necessary conditions for concluding that a project has achieved successful implementation: realization of the implementation plan or strategy; achievement and maintenance of the targeted EBP; and achievement and maintenance of end-point patient or organizational outcomes. These three components reflect a logic model linking an implementation strategy to ultimate outcomes. This definition of successful implementation affords an understanding of when and how an implementation program has delivered the benefits as hypothesized. To accomplish that, we need to assess whether the implementation strategy occurred as planned, whether the EBP was established as needed, and whether desired outcomes followed.
Third, other conceptual models should be drawn on and compared to better elaborate the core PARIHS elements or to better position work using PARIHS in the broader literature. The PARIHS core concept papers make it clear that the developers envision PARIHS being used in combination with other conceptual frameworks. Findings in some of the studies suggest the value of making additional attributes of the evidence-based change more explicit such as those identified in Rogers' Diffusion of Innovation framework [34
]. For example, Rogers' innovation attribute of the observability of a new practice (i.e.
, the extent to which its use by an individual is readily perceived by others in their social network) [2
] does not appear to have an analogue in PARIHS. These types of comparisons and extensions would help build cumulative knowledge and inform refinements to the framework.
The PARIHS authors continue to revisit and refine the framework, recognize its limitations, and call for further research [7
]. We consider a critical strength of any framework. Researchers [60
] and practitioners [61
] continue to use PARIHS and we expect more rigorous studies will be published. Already in the period since we completed our literature search, we are aware of at least five new publications citing PARIHS including two articles presenting results of validations of survey instruments based on the framework [62
]. Also, several prospective research studies based on the framework are in progress by both the PARIHS team (http://www.parihs.org
) and other research teams, including one conducting research in Vietnam and several conducting research in the Veterans Health Administration QUERI program within the US.