Conceptual and relational advances of the PARIHS and i-PARIHS frameworks over the last decade: a critical interpretive synthesis

Abstract

Background

The number of research publications reporting the use of the Promoting Action on Research Implementation in Health Services (PARIHS) framework and the integrated PARIHS (i-PARIHS) framework has grown steadily. We asked how the last decade of implementation research, predicated on the (i-)PARIHS framework (referring to the PARIHS or i-PARIHS framework), has contributed to our understanding of the conceptualizations of, relationships between, and dynamics among the core framework elements/sub-elements. Building on the Helfrich et al. (2010) review of research on the PARIHS framework, we undertook a critical interpretive synthesis to: (1) identify conceptual and relational advances in the (i-)PARIHS framework and (2) identify conceptual and relational aspects of the (i-)PARIHS framework that warrant further work.

Methods

We performed a systematic search in PubMed/PubMed Central, Ovid MEDLINE, CINAHL, JSTOR, SCOPUS, Web of Science, and PsycInfo. Articles were eligible for synthesis if they (a) were peer-reviewed articles, written in English, and published between January 2009 and December 2021, (b) applied the (i-)PARIHS framework explicitly to guide implementation research, and (c) made conceptual (expanding the conceptualization of core elements) and/or relational contributions (elaborating relationships among elements/sub-elements, or theorizing the relationships using empirical data). We used a critical interpretive synthesis approach to synthesize conceptual-relational advances of the (i-)PARIHS framework.

Results

Thirty-seven articles were eligible for synthesis. Twenty-four offered conceptual contributions, and 18 offered relational contributions (5 articles contributed in both ways). We found conceptual expansion of all core (i-)PARIHS elements, with most emphasis on context (particularly outer context and leadership), facilitation, and implementation success. Articles also gave insights into the complex relationships and relational dynamism among these elements, characterized as contingent, interactive, multilevel, and temporal effects.

Conclusions

We observed developmental advances of the (i-)PARIHS framework and proposed several directions to further advance the framework. Conceptualization of (i-)PARIHS elements (particularly evidence/innovation and recipients) need to be further developed by specifying conceptual and operational definitions of underlying sub-elements. Relationships among (i-)PARIHS elements/sub-elements need to be further elaborated through empirical studies that consider situational contingencies and causal complexities. This will require examining necessity and sufficiency of (i-)PARIHS elements/sub-elements in relation to implementation outcomes, interactions among elements, and mechanism-based explanations.

Contributions to the literature

• By summarizing the last decade of literature, this review reveals an increasing number of studies that explicitly applied the Promoting Action on Research Implementation in Health Services (PARIHS) or i-PARIHS framework to guide implementation research.

• Although most of the studies only confirmed established aspects of the framework, we identify some studies that offer conceptual and relational advances by expanding the conceptualization of core elements of the framework and empirically testing complex relationships among elements.

• Findings also suggest theoretical and methodological considerations for researchers to advance the framework in terms of the conceptualizations of and relationships among elements of the framework.

Introduction

The Promoting Action on Research Implementation in Health Services (PARIHS) framework was first introduced in 1998. Since then, it has been used extensively by implementation science researchers and practitioners to predict or explain successful implementation of evidence-based healthcare initiatives [1,2,3]. Authors of the original PARIHS framework contended that successful implementation of evidence-based practices was a function of evidence [4], context [5], and facilitation [6]. Accordingly, successful implementation is more likely when (a) robust evidence matches practitioner, patient, and local experiences; (b) context is receptive to change with sympathetic cultures, effective leadership, and appropriate evaluative systems; and (c) skilled facilitators provide appropriate facilitation of change [7]. While it is applied extensively, the PARIHS framework has been criticized for challenges related to applying it [8,9,10]. In their review of the PARIHS literature, Helfrich et al. [9] noted that (a) the PARIHS elements/sub-elements lack conceptual clarity, (b) definitions of the key outcome successful implementation vary, and (c) those who developed and applied the PARIHS framework rarely examine relationships among elements/sub-elements. Another critique was that PARIHS was largely used as an organizing or heuristic device, and often retrospectively rather than prospectively [8,9,10]. This limits the framework’s potential to guide diagnostic analysis of evidence and context, and to inform design of facilitation/implementation strategies.

Some of the framework’s original authors published the integrated PARIHS framework (i-PARIHS) in 2016 to address concerns over conceptual clarity and describe the framework’s theoretical antecedents [11]. The main changes to the framework were (a) including intended recipients of implementation as a new core concept; (b) enhancing the description of evidence (now identified as innovation, given that new knowledge is usually introduced to generate change or improvement); (c) elaborating the definition of context to differentiate inner and outer context; and (d) refining the concept of facilitation by dividing facilitator roles into beginner, experienced, or expert facilitators and acknowledging internal or external facilitators [11, 12]. The i-PARIHS framework also sets forth interconnections of the core elements by “positioning facilitation as the active ingredient of implementation, assessing and aligning the innovation to be implemented with the intended recipients in their local, organizational, and wider system context.” [11]. It is not clear if these adaptations have supported more effective application of the framework in implementation research and practices.

The number of research publications reporting the use of the (i-)PARIHS framework (referring to the PARIHS or i-PARIHS framework), has grown steadily since the Helfrich et al. [9] review in 2010, according to citation analysis [3]. Building on the Helfrich et al. [9] review, we undertook a critical interpretive synthesis. We asked how the last decade of implementation research, predicated on the (i-)PARIHS framework, has contributed to our understanding of the conceptualizations of, relationships between, and dynamics among the main framework elements/sub-elements. Specifically, the aims of this synthesis were to (1) identify conceptual and relational advances in the (i-)PARIHS framework and (2) identify conceptual and relational aspects of the (i-)PARIHS framework that warrant further work.

Methods

We used a critical interpretive synthesis approach because our objectives were descriptive and interpretive rather than aggregative as intended by conventional systematic reviews [13]. This approach was also that applied in Helfrich et al. [9]’s review, from which ours built. The focus of the current review is to answer questions: How were (i-)PARIHS elements/sub-elements conceptualized? And, Were conceptual-relational understandings of the framework advanced? A critical interpretive synthesis approach allowed us to critically reflect on evidence related to conceptualizing and theorizing the (i-)PARIHS framework drawn from a diverse body of literature (including both qualitative and quantitative research) and to produce a descriptive and interpretive account of the final findings.

This review did not involve an appraisal of the methodological quality of included studies. While conventional systematic reviews evaluate against methodological standards designed for specific study designs, a critical interpretive synthesis locates evidence throughout a research article from the study introduction to discussions of research findings and is conceptual in synthesis process and output [13]. A “hierarchy of evidence” approach to evaluating strength of evidence and methodological quality appraisal tools to examine a range of study validity (e.g., internal validity, external validity, statistical validity), as per that employed in conventional systematic reviews, is not applicable for the current review. Our decisions about including articles for the final synthesis prioritized conceptual relevance. The decision-making processes involved recurrent team discussions and reflection, and consensus-seeking processes (as described in the subsequent Methods subsections).

Search for relevant literature

We performed a systematic search in databases of multiple disciplines, including PubMed/PubMed Central, Ovid MEDLINE, CINAHL, JSTOR, SCOPUS, Web of Science, and PsycInfo. We used a key term search strategy, with key terms including ‘PARIHS,’ ‘promoting action on research implementation in health services,’ and ‘i-PARIHS.’ We restricted the search to English-language publications from January 1, 2009 to December 31, 2021. This work built on Helfrich and colleagues’ 2010 review of literature published before March 2009, so we chose January 2009 as the start date of publication for our search. A health sciences librarian who specialized in knowledge translation and implementation sciences assisted in developing search strategies for each database. Additional file 1: Supplementary file 1 shows search strategies used in each database.

Article selection

We used predefined inclusion criteria: (a) peer-reviewed articles with full text available, written in English, and published between January 1, 2009 and December 31, 2021; (b) articles making explicit reference to the (i-)PARIHS framework; (c) empirical articles using the (i-)PARIHS framework to inform the development of implementation interventions, measurement or data collection, data analysis, or interpretation or evaluation of study findings; or (d) conceptual articles contributing to conceptual advances of the framework. Articles were excluded if they: (a) did not mention the (i-)PARIHS framework in the text or (b) mentioned or cited the (i-)PARIHS framework but the use of the framework was cursory or peripheral to the focus of the article.

One reviewer (YD) first screened titles and abstracts to remove articles that were not peer-reviewed (e.g., dissertations, theses, or book chapters) and articles with no full text available. Two reviewers (YD and AI) performed dual, independent full-text screening. For each article, they documented how the (i-)PARIHS framework were used and assigned the article to include, exclude, or not sure categories based on our predefined inclusion and exclusion criteria. Screening results from the two reviewers were compared, and articles with conflicting decisions or assigned not sure by both reviewers were screened by a third reviewer (JSP or JW). Articles marked for inclusion by at least two reviewers were included for the next step of data extraction.

Data extraction

We extracted both general information about each article and information specific to application of the (i-)PARIHS framework (see Table 1). Iteratively, we developed a standardized data extraction tool using Microsoft Excel, with detailed instructions on extracting each type of information. Eligible articles were distributed among six reviewers (YD, AI, JSP, JW, YS, WB) for independent data extraction. Before independent coding, all reviewers participated in extensive discussions on the extraction form and fields to ensure a consistent understanding of the data extraction instructions. As a group, reviewers piloted the data extraction form on 20 randomly selected articles. Here, each reviewer independently extracted data from the same set of articles, then the reviewers met as a group to discuss the pilot data extraction. These discussions led to improvements in the clarity of data extraction instructions.

Table 1 Data extraction template

Analysis

Analysis consisted of coding and interpretive synthesis steps. Each reviewer (YD, AI, JP, JW, YS, WB) coded a sub-set of articles based on the data extraction results, using predefined codes drawing on the (i-)PARIHS framework. Each article was reviewed and independently coded by two reviewers. The codes specified whether the article made conceptual and/or relational contributions to the (i-)PARIHS framework and which elements/sub-elements were involved. Reviewers also gave narrative accounts of the contributions. Articles were considered to make conceptual and/or relational contributions if they: (i) expanded the conceptualization of core (i-)PARIHS elements, (ii) elaborated relationships among (i-)PARIHS elements/sub-elements, or (iii) theorized the relationships using empirical data. Articles were excluded if they only confirmed established aspects of the (i-)PARIHS framework. We excluded articles by the PARIHS developers that focused on revision of the original PARIHS framework. Two reviewers (YD and AI) reviewed the coding results across the reviewer pairs and documented their agreement or disagreement. Discrepancies were discussed in meetings of all reviewers until consensus was reached.

For our final interpretive synthesis, we only included articles that met our criteria for conceptual and/or relational advances. Two reviewers (YD and WB) independently conducted in-depth analysis of conceptual or relational contributions of articles based on the narrative accounts of contributions provided by reviewers in the coding process. YD and WB documented the results thematically according to types of contributions (conceptual, relational) and element/sub-elements of the (i-)PARIHS framework involved. The results were discussed with all other reviewers at recurrent team meetings. The final output, then, incorporated the outcomes of the two independent analyses and inputs/comments from other reviewers.

Results

Figure 1 shows our article selection process. We identified 280 articles that met the eligibility criteria for data extraction. Of those, 37 articles met the criteria for conceptual and/or relational contributions to the (i-)PARIHS framework. We included these 37 articles in the final interpretive synthesis (Table 2). Of these, 29 were empirical studies (23 qualitative studies, 3 quantitative studies, 3 mixed-methods studies), 4 were literature reviews, and 4 were theory/methodological articles. Except for one study conducted in the construction industry, all empirical studies featuring the (i-)PARIHS were conducted in healthcare settings including acute care/hospitals, long-term care, primary care/community care, and US Veterans Health Administration. Twenty-four offered conceptual contributions, and 18 offered relationships among core elements and/or sub-elements (5 articles contributed in both ways).

Fig. 1

Flow diagram of article selection

Table 2 Overview of articles included in the interpretive synthesis (n = 37)

Conceptual advances of (i-)PARIHS elements and sub-elements

Table 3 summarizes conceptual contributions to (i-)PARIHS elements/sub-elements from 24 articles reviewed.

Table 3 Original and advanced conceptualizations of (i-)PARIHS elements and sub-elements

Twelve articles advanced the conceptualization of context

Three articles proposed new context typologies with particular applicability in certain implementation situations, including implementation of complex interventions [14], implementation of public health interventions [15], and evidence application by nursing professionals [16].

Six empirical articles proposed conceptualizations of outer/external context [17,18,19,20,21,22]. Emphases on sub-elements of outer/external context depended on implementation contexts, such as external resources (abundance or scarcity) in low-income healthcare settings [22]; capacity of researchers, clinicians, and other stakeholders for integrated knowledge translation programs [20]; societal culture for implementing public health interventions [18]; credibility of external entities responsible for developing or promoting clinical guidelines to implement a specific clinical guideline [17, 19]; and social networks and communication pathways with external entities in sectors such as the construction industry, where lines of information-sharing were fragmented and multidirectional [21].

Three empirical articles [23,24,25] expanded the conceptualization of leadership (one sub-element of context) beyond the low–high continuum in the PARIHS framework. One article suggested conceptualizing leadership as a versatile, dynamic construct that operates in multiple ways and on multiple levels in an organization and evolves over implementation intervals [23]. One article detailed sub-elements under each general element of leadership (teamwork organizational structure, transformational leadership) in the PARIHS framework [25]. One article suggested “balance between leadership and management to maximize their influence on the implementation process” as a new element of leadership [24].

Ten articles advanced the conceptualization of facilitation

Articles extended the conceptualization of facilitation as roles, processes, or strategies that facilitate the achievement of implementation goals [19, 26,27,28,29,30,31,32,33,34]. Among articles focusing on facilitation processes or strategies, three suggested facilitation as a deliberate implementation intervention targeting evidence–context fit [19, 26, 29]. Two articles discussed holistic-oriented facilitation through creating a psychologically safe space [30] or establishing capacity-building mentoring relationships [27]. One article described facilitation as four categories of strategies—planning, restructuring, education, and quality management—with specific facilitative actions under each category [33].

Among articles focusing on facilitation roles, one elaborated the conceptualization of facilitation by distinguishing between external and internal facilitation roles and the dynamics between them [28]. One article compared active and passive facilitation roles of national agencies to promote innovation uptake [34]. One article described the role transfer process from expert to novice facilitators [32]. One article suggested two new facilitation sub-elements, facilitation intensity and facilitator resilience, conceptualized based on the perceptions and experiences of facilitators themselves [31].

Five articles explicitly discussed the conceptualization of implementation success

Three articles focused on outcomes in specific implementation situations such as implementing a particular targeted evidence-based practice (EBP) [26, 35] and implementing holistic-oriented facilitation [36]. Successful implementation of an EBP is suggested to be reflected in three aspects: the implementation plan and its realization, EBP innovation uptake, and the achievement of patient and/or organizational outcomes [26, 35]. When implementing holistic-oriented facilitation, implementation outcomes encompass both instrumental outcomes (improvement in system performance and patient outcomes), and conceptual outcomes (professional development and improvement in organizational context).

Two articles addressed broader knowledge translation outcomes [20, 37]. For example, outcomes in an integrated knowledge translation project were suggested to include health service outcomes (e.g., accomplished professional goals and use of research), social outcomes (e.g., collaborations between researchers and research users), and research outcomes (e.g., enhanced quality/relevance of research) [20]. One article expanded the conceptualization of knowledge use, categorizing it into conceptual, instrumental, and strategic use of knowledge [37].

Finally, one article expanded the conceptualization of evidence and suggested EBP characteristics as a new sub-element of evidence when applying the PARIHS framework in implementing programs that target a specific EBP [26]. Two studies discussed the conceptualization of recipients [33, 35], suggesting individual factors as a new element [35] or “professional expectations and requirements” as a new sub-element of recipients [33].

Advancing understanding of relationships among (i-)PARIHS elements

Table 4 describes contributions to relational aspects of the (i-)PARIHS framework. Four articles provided insights into dyadic relationships (relationships between two variables) of facilitation with other (i-)PARIHS elements. Three of those articles discussed the dyadic relationship of facilitation with evidence [38], with context [38,39,40], or with recipients [40]. One article discussed temporal changes in other elements in response to facilitation [41].

Table 4 Relational contributions

Eleven articles on individual empirical studies extended beyond dyadic considerations to more complex relationships linking core (i-)PARIHS elements (evidence, context, facilitation, recipients, and their sub-elements) to implementation outcomes [20, 35, 36, 40, 42,43,44,45,46,47,48]. These articles addressed three broad relational questions:

• (i) What is the relative importance of the core (i-)PARIHS elements/sub-elements in affecting implementation outcomes? The articles identified that evidence is necessary but not sufficient [35, 42], leadership is necessary and sufficient [20, 40], and impacts of facilitation and context on implementation outcomes depend on the phase of implementation [43];

• (ii) How do (i-)PARIHS elements/sub-elements interact with one another in affecting implementation outcomes? Articles identified that facilitation’s contribution to implementation outcomes or research use is contingent on context [45, 47, 48], that interactions among context sub-elements can produce synergy or compensation [42], and that facilitation and context are complementary [44].

• (iii) What types of facilitation do or do not work, for whom, how, why, and in what circumstances, in improving implementation outcomes? Articles identified that facilitation mechanisms can be usefully grounded in organizational learning frameworks [36] and that context-mechanism-outcome configurations are feasibly grounded in realist approaches [46].

Using research synthesis approaches, three articles integrated (i-)PARIHS with complex system perspectives and broader theoretical models in implementation science to describe the dynamism among (i-)PARIHS elements, particularly in the context of implementing complex interventions [15, 49, 50]. These reports identify complexity is an inherent characteristic of each (i-)PARIHS element and also as a component of the dynamic relationships among these elements. The dynamism is reflected by the non-linear, non-static, and adaptive relationships among the core elements.

Discussion

We undertook this review to determine how the conceptualizations of (i-)PARIHS elements/sub-elements, the relationships between them, and dynamics among them have advanced since publication of the critical review by Helfrich and colleagues [9]. We observed an increasing number of studies that applied (i-)PARIHS explicitly to guide implementation research at different stages, from developing implementation interventions to evaluating implementation outcomes. Although much of the (i-)PARIHS literature confirmed what is set out in the (i-)PARIHS framework, we identified some articles that offered conceptual and relational advances. Here we discuss contributions added by those articles to the (i-)PARIHS framework. We also offer directions to further advance the frameworks.

Steps toward conceptual clarity and comprehensiveness

The articles included in this review addressed conceptual expansion of all the core (i-)PARIHS elements with most emphasis placed on context (particularly outer context and leadership), facilitation, and implementation successes. Some conceptual limitations noted for the original PARIHS framework were resolved with the revised i-PARIHS framework [11, 12], but we highlight persisting concerns about conceptualization of some concepts that need to be further developed.

Although the concepts of evidence/innovations and recipients were not discussed frequently, we see the introduction of ‘implementable’ characteristics of evidence in the article by Stetler et al. [26] as promising, along with Rycroft-Malone et al.’s [35] suggestion of including individual recipients’ characteristics as a new element. Findings from these two articles offered implications for developing the revised i-PARIHS framework where evidence is conceptualized as a new concept innovation and recipients is added as a stand-alone element [26, 35]. Explicit definitions of and differentiation between various implementable characteristics of evidence/innovations have rendered this concept more measurable. Inclusion of recipients as a core element has improved the explanatory power of the framework. Nevertheless, we note that these two concepts received less attention than other (i-)PARIHS concepts in diagnostic assessments or process evaluations in implementation studies, and these concepts’ operational definitions remain to be developed.

The most obvious advance in conceptual development relates to the concept of context. A number of articles usefully distinguish between the influences of inner and outer context and consider the importance of outer context to implementation [14, 15, 17,18,19,20,21,22]. This work is commendable and necessary. However, the considerable variation in how outer context was conceptualized (ranging from the influences of specific extra-organizational entities such as guideline-producing organizations to various levels of ‘contextual trust’ ascribed to broad political and economic characteristics), and the under-developed evaluative approaches of outer context, make it difficult at this stage to determine its effects on implementation.

The conceptualization of leadership (one sub-element of context) is discussed widely. Several articles suggest conceptualizing leadership as a multifaced construct beyond a low–high continuum [14,15,16, 23]. Despite the conceptual insight, we note a lack of empirical studies that explored typologies of leadership to address the issues associated with the low–high-continuum approach. Another insight into conceptualization of leadership is the suggestion of describing leadership as perceived by clinical leaders themselves (such as care managers). This could uncover an often-overlooked aspect of leadership—the balance between management and leadership and the dual roles of a clinical leader in management tasks and in leading changes [24, 25]. This has not been studied widely. We suggest future research to further examine leadership from the lens of clinical leaders and differentiate between leadership and management. Moreover, other sub-elements of context at the local (micro) or organizational (meso) level, such as culture, learning environment, and evaluation and feedback process, need to be further studied. These sub-elements of context are central to implantation research and yet to be explicitly conceptualized and operationalized.

Expansion of the concepts of facilitation and implementation success are major contributions to the literature over the past decade. We found it helpful to distinguish between different types of implementation contexts when defining facilitation and implementation success, as was attempted in several articles we reviewed. These articles examined two types of implementation contexts. One implemented a targeted EPB or clinical intervention through explicit task-oriented facilitation strategies that aimed for specific clinical outcomes [19, 26, 29, 35]. The other focused on holistic-oriented facilitation that aimed to generally enhance organizational context and staff’s attitudes and behaviors [27, 30, 36]. Facilitation and implementation success were conceptualized differently: narrow and project-specific in the first context versus broad and organization-wide in the second.

The conceptualization of facilitation has expanded beyond internal facilitation roles and processes to address the dynamics between internal and external facilitation [28] and the process of transferring facilitation skills from expert to novice facilitators [32]. This research area needs more attention, given the growing application of complex innovations that involve multiple players from internal and external organizations and facilitators with different levels of facilitation skills. Another important research agenda is to understand the concept of facilitation from the perspectives, perceptions, and experiences of facilitators themselves, which will provide a different lens to examine facilitation and its relationship to implementation outcomes [31].

Similarly, the concept of implementation success has expanded to include conceptualizations at different levels of analysis: broadly at system or organization levels and more narrowly at the level of work units or projects [20, 37]. The literature advocates for comprehensive measures of outcomes at different stages of implementation rather than merely outcome measures at the final stage. Theory-driven formative or process evaluations were frequently recommended [20, 26, 35,36,37]. We suggest that classifying implementation success to incorporate levels of analysis and implementation stages would facilitate conceptualization of the (broad) concept of implementation success.

Relational insights

Several articles explicitly attempted to improve our understanding of the complex relationships among (i-)PARIHS elements/sub-elements. Nonetheless, more empirical research is needed to theorize the relationships through actively using empirical data and carefully clarifying the range of specific instances to which findings apply. As suggested by Kislov et al. in 2019 [51], knowledge will advance with theoretical informative research (in addition to mere theoretical informed research) based on theorizing practices, instead of using the (i-)PARIHS framework as heuristic tools. Here we discuss theoretical and methodological considerations for researchers to further theorize relationships between (i-)PARIHS elements/sub-elements.

Articles articulating dyadic relationships of facilitation with either of the other three elements (evidence/innovation, context, recipients) shed some light on the role of facilitation as the active ingredient in integrating other elements through bidirectional (rather than unidirectional) relationships unfolding at different levels (macro, meso, and micro) and at different stages (temporal dynamic) [38,39,40]. However, research was lacking that thoroughly examined relationships of facilitation with all other elements. In addition, a lack of longitudinal quantitative designs hindered a systematic understanding of the temporal dynamics among the (i-)PARIHS elements and how they correspond to different stages of an implementation process. More investigation at the sub-element level is also needed to understand more granular relationships, such as how links of facilitation to leadership (one sub-element of context) might differ from links of facilitation to evaluation (another sub-element of context).

Some articles [20, 35, 36, 40, 42,43,44,45,46,47,48] extended beyond dyadic relationships to more complex relationships that linked implementation outcomes with (i-)PARIHS core elements (evidence, context, facilitation, recipients, and their sub-elements). These articles provided three important implications for future research to establish the relational propositions. First, it continues to be important for future studies to systematically examine the necessity and sufficiency of (i-)PARIHS elements/sub-elements in relation to implementation outcomes. More empirical research is needed to verify the findings synthesized from these articles on the necessity and sufficiency of some elements or sub-elements. For example, evidence (specifically belief in quality of evidence) was viewed as a necessary but not sufficient factor [35, 42], and leadership (one sub-element of context) was found as a necessary and sufficient determinant of positive implementation outcomes [20, 40]. The primacy of other elements/sub-elements also needs investigation. Studies we reviewed that examined necessary and sufficient factors for implementation success all applied qualitative methods, but we recommend future quantitative research to systematically examine primary ‘difference makers’ in an implementation process. Advances in quantitative methods such as configurational comparative analysis have enabled researchers to quantify the causal complexity in implementation research [52].

Second, an important research agenda is to clarify the interactive relationships among (i-)PARIHS elements/sub-elements and the effects of these interactions on implementation outcomes. Interactions among elements and sub-elements may exert complementary, synergistic, or buffering effects on implementation outcomes [42, 44, 45, 47, 48]. Some quantitative studies we reviewed offer good examples of using basic and advanced statistical interaction analyses to quantify the interactive complexities among (i-)PARIHS elements/sub-elements [45, 47]. However, much more empirical research is still needed to uncover underlying casual structures that explain why the interactions occur. More advanced quantitative methods such as structural equation modeling techniques might be used to pursue advanced relational studies [53].

Third, research focusing on mechanism-based explanations has enormous scientific and practical value, given the increasing complexity and variability of facilitation/implementation interventions that unfold in diverse and changing contexts. Two articles [36, 46] illustrate that facilitation influences implementation outcomes through more than one mechanism. Complexities embedded in facilitation mechanisms call for integration of (i-)PARIHS with other theoretical and methodological perspectives and tools to provide mechanism-based explanations. For example, organizational learning frameworks [36] and a realist-informed approach [46] were used in combination with the (i-)PARIHS framework to explain what/how/why/for whom a facilitation intervention works and to develop context–mechanism–outcome configurations (referred to as realist-informed contingent explanations). Only a few studies [36, 46] offer mechanism-based explanations. This form of research requires in-depth, extensive process evaluations and often includes modeling techniques such as structural equation modeling and configurational comparative analysis to draw quantitative conclusions. Nonetheless, process evaluations and mechanism-based explanations offer great promise. Outcome measures alone may lead to false negative conclusions on the effectiveness of facilitation/implementation interventions and effects of clinical interventions or evidence-based practice being implemented.

Some articles integrated the (i-)PARIHS framework with complex system perspectives and broader theoretical knowledge in implementation science and other disciplines [15, 49, 50]. These articles strengthened our understanding of the dynamism among (i-)PARIHS elements that unfold at multiple levels of context (macro, meso, micro) and at different stages of an implementation process [15, 49, 50]. These articles developed more comprehensive models that varied in levels of abstraction and generalizability, from a logic model/program theory [49] to a conceptual model that synthesized and consolidated existing theoretical and empirical findings [15, 50]. These integrated models might support developing mid-range theories, which are seen as critical for implementation science because they generate testable propositions and provide practically adequate explanations applicable to a specified set of contexts [51]. However, the utility of the integrated models generated in these articles needs validation with empirical data.

Limitations of this review

Some limitations of this critical review must be noted. First, we included only peer-reviewed published research in English. We might have omitted important work applying the (i-)PARIHS framework that was published as gray literature or in different languages. Second, our review focused only on articles and studies applying the (i-)PARIHS framework. Implications of the conceptualizations of and relationships among the core implementation constructs are not generalizable broadly for implementation science. Researchers who seek to elaborate the conceptualization of these the core implementation constructs more generally may also want to draw on other conceptual and theoretical models in implementation science or other disciplines, as suggested by Nielsen [54]. Third, the scope of included articles was wide and the applicability of broad findings to specific healthcare settings is difficult to determine. Finally, although a critical interpretive synthesis approach (which was initially developed to synthesize evidence related to concept and theory development from both qualitative and quantitative research) was better suited for the purpose of this review than other synthesis methods, this method is not without limitations. One limitation of a critical interpretive synthesis that has been noted is the low reproducibility of its protocols and findings as a different team using the same set of articles might produce different synthesis results and different interpretation of findings [13]. We used a team consensus-based decision-making approach and worked to maintain transparency in documenting study steps to mitigate the limitation of low reproducibility.

Conclusions

We reviewed the last decade of implementation research articles predicated on the (i-)PARIHS framework. We used critical interpretive synthesis to identify conceptual and relational advances in the frameworks. We found conceptual expansion of all core (i-)PARIHS elements with most emphasis on context (particularly outer context and one of the key elements of context—leadership), facilitation, and implementation success. Articles also gave insights into the complex relationships and relational dynamism among these elements, characterized as contingent, interactive, multilevel, and temporal effects. We propose several directions to further advance the (i-)PARIHS framework:

• 1. We suggest that conceptualization of core (i-)PARIHS elements (particularly evidence/innovation and recipients) need to be further developed by specifying conceptual and operational definitions of underlying sub-elements.

• 2. Relationships among (i-)PARIHS elements/sub-elements need to be further elaborated through empirical studies that consider the situational contingencies for relationships.

• 3. To provide theoretically and practically meaningful indications, research needs to address relational and causal complexities. This will require examining necessity and sufficiency of (i-)PARIHS elements/sub-elements in relation to implementation outcomes, interactions among elements, and mechanism-based explanations.

• 4. More quantitative methods, such as configurational comparative analysis [52] and structural equation modeling techniques [53], may be used to quantify causal complexities.