The use of the PARIHS framework in implementation research and practice—a citation analysis of the literature

Background The Promoting Action on Research Implementation in Health Services (PARIHS) framework was developed two decades ago and conceptualizes successful implementation (SI) as a function (f) of the evidence (E) nature and type, context (C) quality, and the facilitation (F), [SI = f (E,C,F)]. Despite a growing number of citations of theoretical frameworks including PARIHS, details of how theoretical frameworks are used remains largely unknown. This review aimed to enhance the understanding of the breadth and depth of the use of the PARIHS framework. Methods This citation analysis commenced from four core articles representing the key stages of the framework’s development. The citation search was performed in Web of Science and Scopus. After exclusion, we undertook an initial assessment aimed to identify articles using PARIHS and not only referencing any of the core articles. To assess this, all articles were read in full. Further data extraction included capturing information about where (country/countries and setting/s) PARIHS had been used, as well as categorizing how the framework was applied. Also, strengths and weaknesses, as well as efforts to validate the framework, were explored in detail. Results The citation search yielded 1613 articles. After applying exclusion criteria, 1475 articles were read in full, and the initial assessment yielded a total of 367 articles reported to have used the PARIHS framework. These articles were included for data extraction. The framework had been used in a variety of settings and in both high-, middle-, and low-income countries. With regard to types of use, 32% used PARIHS in planning and delivering an intervention, 50% in data analysis, 55% in the evaluation of study findings, and/or 37% in any other way. Further analysis showed that its actual application was frequently partial and generally not well elaborated. Conclusions In line with previous citation analysis of the use of theoretical frameworks in implementation science, we also found a rather superficial description of the use of PARIHS. Thus, we propose the development and adoption of reporting guidelines on how framework(s) are used in implementation studies, with the expectation that this will enhance the maturity of implementation science.


Introduction
There has been an increased use of theoretical frameworks in the field of implementation science in the last decade, with most developed in the last two decades [1,2]. Tabak et al. identified 61 theoretical models used in dissemination and implementation science [3]. However, while theoretical frameworks are increasingly being cited, more research is needed to understand how they are chosen and applied, and how their use relates to improved implementation outcomes [1,4]. Variously described in the form of theories, frameworks, or models, all strive to provide conceptual clarity on different aspects of implementation practice and research. For consistency, we will refer to these as theoretical frameworks, or simply "frameworks." The Promoting Action on Research Implementation in Health Services (PARIHS) framework is a multidimensional framework which was developed to explicitly challenge the pipeline conceptualization of implementation [5]. The PARIHS framework is a commonly used conceptual framework [1,4] that posits successful implementation (SI) as a function (f) of the nature and type of evidence (E) (including research, clinical experience, patient experience, and local information), the qualities of the context (C) of implementation (including culture, leadership, and evaluation), and the way the implementation process is facilitated (F) (internal and/or external person acting as a facilitator to enable the process of implementation); SI = f(E,C,F). The framework was informed by Rogers' Diffusion of Innovations [6] and various organizational theories and theories from social science [7] and generated inductively by working with clinical staff to help them understand the practical nature of getting evidence into practice. The PARIHS framework was initially published in 1998 [5] and updated based on a conceptual analysis in 2002 [8] and further primary research [9]. A further refinement was undertaken in 2015 [10], resulting in the integrated or i-PARIHS. Articles using the revised version are not included in the citation analysis reported here. The PARIHS framework has been described as a determinant framework in that it specifies determinants that act as barriers and enablers influencing implementation outcomes [2]. Skolarus et al. [1] identified Kitson et al. [5] as one of the two primary originating sources of influence in their citation analysis of dissemination and implementation frameworks.
Despite the growing number of citations of theoretical frameworks in scientific articles, the detail of how frameworks are used remains largely unknown. Systematic reviews of the application of two other commonly used frameworks [1], the Knowledge to Action framework [11] and the Consolidated Framework for Implementation Research [12], both reported that use of these frameworks, beyond simply citation, was uncommon. While PARIHS has been widely cited, it has also been scrutinized; in 2010, Helfrich et al. published a qualitative critical synthesis of studies that had used the PAR-IHS framework [13], finding six core concept articles and 18 empirical articles. One of the reported findings was that PARIHS was generally used as an organizing framework for analysis. At the time, no studies used PARIHS prospectively to design implementation strategies [13]. A systematic review applying citation analysis to map the use of PARIHS (similar to those undertaken for the Knowledge to Action framework (KTA) [11] and the Consolidated Framework for Implementation Research (CFIR) [12]) has not yet been performed.
Systematic reviews can contribute to the development of existing theoretical frameworks by critically reviewing what authors state as their weaknesses and strengths; they can also direct future and current users of frameworks to examples of using the frameworks in different ways. To contribute to this development from the perspective of the PARIHS framework, we undertook a citation analysis of the published peer-reviewed literature that focused on the reported use of PARIHS (and its main elements), in what contexts the framework has been applied, and what scholars who have used the PARIHS framework (and its main elements) report as its strengths, limitations, and validity.

Methods
The method used for this study is citation analysis, i.e., the examination of the frequency and patterns of citations in scientific articles, in this case articles citing the core PARIHS framework publications. A team of researchers with engagement in the development and/or use of the PARIHS framework was constituted. Initially, the group decided on the core publications for the citation analysis. Four articles were selected as they represented the key stages of the framework's development, namely the original paper that described PARIHS, plus • The findings underline that descriptions of the use of the framework generally were not that transparent and often partial.
• Findings also point at difficulties in using the framework, such as lack of guidance on key steps to overcome barriers and support implementation • Identifies the need of common guidelines on how theories, models, and frameworks should be reported in research articles.

Citation search
Citation searches were performed by an information specialist (KG) to retrieve published articles citing any of the four core articles. The searches were performed in two citation databases: Web of Science and Scopus. The first searches were performed between 31 March 2016 and 1 April 2016. Later, 6 September 2019, additional searches were performed in respective databases. These searches were limited to citations that were published 1 April 2016-31 August 2019 to update the result from the first searches. All citations that were published September 1998 (i.e., when Kitson et al 1998 was published)-31 August 2019 (i.e., prior to the search date) in respective databases were collected in EndNote Library. Endnote was used for checking duplicates and retrieving full texts. To manage the scope of the citation analysis, we opted to only include articles in English published in peer-reviewed scientific journals. The searches in Web of Science were, because of the subscription, limited to Web of Science Core Collection without Book Citation Index.

Data extraction
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram [14] for the data extraction is provided in Fig. 1. Initially, an assessment to identify the articles that used the PARIHS framework in any other way than merely referencing one or more of the core articles was performed (Additional file 1). For this initial assessment, all articles were read in full. After identifying articles where the PARIHS framework was used, data extraction was undertaken using a tailor-made data capture form (Additional file 1).
The data capture form was developed and piloted in iterative cycles by the research team. Apart from capturing information about where (country/countries and setting/s) and with whom (professional groups and roles) PARIHS had been applied, the form included questions on whether PARIHS was used in one or more of the following ways: 1) In planning and delivering an intervention, 2) In data analysis, 3) In the evaluation of study findings, and/or 4) In any other way.
Each of these questions was followed by an openended item for extracting information on how this was reported [15]. To enhance reliability and data richness, each reviewer copy-pasted sections of the article corresponding to the open-ended reply into the data extraction form when appropriate and indicated page, column, and row. Two additional items captured whether the PARIHS framework had been tested or validated, as well as any reported strengths and weaknesses of the framework. Thus, we report on what the authors of the included articles claim to have done, rather than a judgment as to how and to what extent they actually used the PARIHS framework.
For data extraction and validation, the research team was divided into four pairs, ensuring that each article was assessed separately by at least two research team members. The pairs received batches of 20 articles at a time. Variations in the assessments were discussed until consensus was reached within the pair(s). Further, queries detected within the pairs were raised and discussed with the whole research team, until consensus was achieved. Regular whole-team online meetings were held to consolidate findings between every new batch of articles and throughout the development and analysis process. In total, the group had > 20 online meetings and four face-to-face meetings from the initial establishment of the group in January 2015.

Data analysis
Categorical data were analyzed using descriptive statistics, whereas the open-ended items were analyzed qualitatively [16], including the collated extractions of data to illustrate each of the four types of use (i.e., how the PARIHS framework was depicted in terms of (1) planning and delivering an intervention, (2) analysis, (3) evaluation of study findings, and/or (4) in any other way).
Applying a content analysis approach [17], members of the research team worked separately with the texts extracted from the reviewed articles. The extracts for each open-ended item were read and reread, to get a sense of the whole. Next, variations were identified and formed as categories. Findings for each question were summarized in short textual descriptions, which were shared with the whole team. In a face-to-face meeting, the data relating to each question were critically discussed and comparisons were made between the findings for each question, to identify overlaps and relationships about how PARIHS has been used.

Results
After duplicate control, 1613 references remained. These were sorted by language and type of publication. In this phase, 131 references categorized as books, book chapters, conference proceedings, and publications written in non-English language were excluded. Also, three of the four core articles (i.e., the three citing Kitson et al. [5] which was the starting point for development of the PARIHS framework and therefore did not appear in the citation search) were excluded from the database [8,9,18], as were four articles expanding and refining PAR-IHS [19][20][21][22]. Accordingly, 1475 articles remained, and after the assessment excluding those merely citing PAR-IHS, a further 1108 articles were excluded, leaving 367 articles that cited one or more of the core articles, and made explicit use of the PARIHS framework (see Fig. 1 and Table 1).
In the following sections, references have been added to the categorical items in the data extraction while we have opted only to provide examples of references to the findings from the qualitative exploration of how the PARIHS framework was operationalized in detail.

Timing of different types of articles
The types of articles published using the PARIHS framework changed over time, with an increase in the number of empirical studies from 2004 onwards, as illustrated in Fig. 2. As the search for articles for this review only included the first eight months of 2019, the graph is limited to full years (i.e., 1998 through 2018). Figure 3 depicts how PARIHS was used by type of article. Although authors frequently claimed that PARIHS was used in one or more ways, details as to how the framework was used were often lacking.

The application of PARIHS to plan and deliver an intervention
In total, 117 (32%) articles claimed to use the PARIHS framework to plan and deliver an intervention [23-46, 58-82, 84-143, 145, 146, 339, 340, 386-388]. Predominantly, these were empirical studies (n = 91) [58-82, 84-143, 145, 146, 386-388] but also two opinion/theoretical articles [339,340] and 24 protocols . Of the 117 articles, about half stated that the framework was used for theoretically informing, framing, or guiding an intervention (e.g., [82,103,105,134,386]). However, in these studies, PARIHS was referred to only in a general sense, in that the core elements of the framework were said to have informed the planning of the study. There was a lack of detail provided about what elements of the framework were used and how they were operationalized to plan and deliver an intervention. In the other half of the 117 articles, it was described more specifically that one or more elements of the framework had been used. Most commonly the facilitation element (e.g., [58,80,92,98,110]) was referred to as guiding an implementation strategy. The articles that provided explicit descriptions of interventions using facilitation employed strategies such as education, reminders, audit-andfeedback, action learning, and evidence-based quality improvement, and roles including internal and external facilitators and improvement teams to enable the uptake of evidence (e.g., [23,79,125,142]). Some articles drew on the PARIHS framework more specifically, to understand the role of organizational context in implementation (e.g., [34,63,145,340]).
We found two main approaches to how the PARIHS framework was used in the evaluation of study findings. First, PARIHS was used to organize the discussion of the findings (e.g., [73,87,109,166,214]), where the framework and/or its elements were used to provide a structure for reporting or generally discussing the findings, or both, for example in stating that the key elements of PARIHS were reflected in the study findings. Second, the framework was used to consider the implications of the study's findings (e.g., [81,84,170,207,361]), where the framework or its elements (varying between one (e.g., [75,195,211]), two (e.g., [71,86,105]), and all the three main elements (e.g., [80,261,269])) enabled authors to elaborate on findings, or reflect on the implications of their study to evaluate the PARIHS framework itself. Specifically, we found some empirical articles reported evaluating the PARIHS element "context" by means of context tools (e.g., [155]). In addition, an evaluation of the study findings using the framework was identified in 18 opinion/theoretical articles [339,[341][342][343][344][345][346][347][348][349][350][359][360][361][362][363][364][365] and 18 empirical review studies [13, 313, 314, 316-320, 323-331, 389]. Among the opinion/ theoretical articles, there were papers evaluating other theoretical constructions in relation to the PARIHS framework (e.g., [364]).
The articles varied in detail, depth, and quality in terms of descriptions of how they went about testing the validity of the PARIHS framework. Approaches ranged from general observations of whether the research teams/users found the elements and sub-elements easy to use (e.g., [62,188,203]), to studies that used elements of context described in the PARIHS framework to validate new context measures across settings and groups (e.g., [150,155,207]). As one example, the Alberta Context Tool started from the PARIHS conceptualization of context to include dimensions of culture, leadership, and evaluation.
Regarding the strength and limitations of the PARIHS framework, about one third of the included articles reported on its strengths and about 10% commented on perceived limitations. The identified strengths included: ▪ Holistic implementation framework (e.g., [141,164,209,258]) that is perceived as intuitive and accessible.
▪ Both practical and theoretical and therefore feasible to use by both clinicians and researchers; also seen as intuitive to use and accessible (e.g., [117,209,255]).
▪ Can be used as a tool: diagnostic/process/evaluative tool; predictive/explanatory tool or as a way to explain the interplay of factors (e.g., [93,205,255,285,379]).
▪ Can accommodate a range of other theoretical perspectives (including approaches such as social network theory, participatory action research, coaching, change management and other knowledge translation frameworks) (e.g., [93,105,245,246]).
▪ Can be used successfully in a range of different contexts (low-and middle-income countries) [150] and services and for various groups of patients (disability, aged care) (e.g., [80,113,248,312]).
Limitations of the PARIHS framework included: ▪ Poor operationalization of key terms leading to difficulties in understanding and an overlap of elements and sub-elements (e.g., [165,285,376]).
▪ Lack of practical guidance on steps to operationalize the framework (e.g., [209,254]) with a subsequent lack of tools.
▪ Lack of information on the individual and their characteristics (e.g., [209,361]) and their lack of understanding of evidence (e.g., [204,390]).
▪ Too structured and does not acknowledge the multidimensionality and uncertainty of implementation (e.g., [143,214]).
▪ Lack of acknowledgement of wider contextual issues such as the impact of professional, socio-political, and policy issues on implementation (e.g., [115,143,285,354]).
▪ Not providing support in how to overcome barriers to successful implementation (e.g., [88]).

Discussion
In a recent survey among implementation scientists, the PARIHS framework was found to be one of the sixth most commonly used theoretical frameworks [4]. Yet, in our review, about 23% (n = 367) of the identified 1614 articles citing any of the four selected core PARIHS articles used the framework in any substantial way. Similarly, a review of the CFIR found that 26/429 (6%) of articles citing the framework were judged to use the framework in a meaningful way (i.e., used the CFIR to guide data collection, measurement, coding, analysis, and/or reporting) [12]. A citation analysis of the KTA framework found that about 14% (146/1057) of screened abstracts described using the KTA to varying degrees, although only 10 articles were judged to have applied the framework in a fully integrated way to inform the design, delivery, and evaluation of implementation activities [11].
PARIHS has been used in a diverse range of settings but, similarly to other commonly used implementation frameworks, most often superficially or partially. The whole framework has seldom been used holistically to guide all aspects of implementation studies. Implementation science scholars have repeatedly argued that the underuse, superficial use, and misuse of implementation frameworks might reduce the potential scientific advancements in the field, but also the capacity for changing healthcare practice and outcomes [4]. The rationale for not using the whole PARIHS framework could be many, including the justified reason of only being interested in a particular element. As such, partial use cannot always be considered as inappropriate. Simultaneously, many researchers entering the field might be overwhelmed with the many frameworks available and the lack of guidance about how to select and operationalize them and using their elements [2,4,391]. The current citation analysis can thus help remedy a gap in the literature by revealing how the PARIHS framework has been used to date, in full or partially, and thus provides input to users of its potential use.
The use of theoretical frameworks in implementation science serves the purpose of guiding researchers' and practitioners' implementation plans and informing their approaches to implementation and evaluation. This includes decisions about what data to gather to describe and explain implementation, their hypotheses about action steps needed, how to account for the critical role of context, and providing a foundation for analysis and discussion [7]. The advancement of theoretically informed implementation science will, however, depend on much improved descriptions as to why and how a certain framework was used, and an enhanced and betterinformed critical reflection of the functionality of that framework. This review shows that the PARIHS framework has rarely been used as a whole; rather, certain elements tend to be applied, often retrospectively as indicated in Fig. 3 underlining the use of PARIHS in the evaluation of study findings, which resonates with the findings of reviews about the use of the KTA [2] and CFIR [11] frameworks. This could be as a result of a lack of theoretical coherence of some frameworks making them difficult to apply holistically, and/or a function of a general challenge that researchers face in operationalizing theory. However, this could also be a result of publishing constraints. While the PARIHS framework may have guided implementation or been implicitly used in the study design, it was rarely the focus of the publications. Further, the aims and scopes of scientific health care journals have historically prioritized clinical outcomes over implementation outcomes where one could expect a more detailed description of the use of theoretical frameworks. This may have resulted in authors not fully reporting their use of, e.g., the PARIHS framework.
The number of empirical studies using the PARIHS framework has steadily increased over the review period. There is also evidence to show that more research teams have contributed to critiquing the framework in terms of reporting on its strengths and limitations and its validity. The pattern of investigation is moving from studies on context, to more systematic explorations of facilitation, thus contributing to a more detailed understanding of the elements and sub-elements of the framework. The lack of focus on "evidence" identified in this review highlights the need for researchers and clinicians to focus on the multi-dimensionality of what is being implemented. Common patterns emerging in this review support the changes made to the most recent refinement of the PARIHS framework [359].
Consistent with other reviews of the use of theoretical frameworks in implementation science, we found that PARIHS was often not used as intended. Further, it was not always clear why the particular framework was chosen. Frequently, authors merely cite a framework without providing any further information about how the framework was used. The lack of clear guidance on how to operationalize frameworks might be one of the underlying reasons for this. Lastly, to enable a critical review of frameworks and further build collective understanding of implementation, we urge authors to be more explicit about how theory informs studies. Development and adoption of reporting guidelines on how framework(s) are used in implementation studies might assist in sharpening the link between the used framework(s) and the individual study, but could potentially also enhance the opportunities for advancing the scientific understanding of implementation.

Limitations
To increase study reliability during the review process, more than one person identified, assessed, and interpreted the data. We had regular meetings to discuss potential difficulties in assessing included articles, and subsequently, all decisions were resolved by consensus to enhance rigor. We used a rigorous search strategy, which was undertaken by an information specialist. The standardization of our processes across the team was also enhanced by the creation of an online data extraction form via Google. However, as the form was not linked to other software (e.g., Endnote), this added timeconsuming processes.
As we did not include articles that were not written in English, we may have limited the insights about the application of the PARIHS framework, particularly with relevance to different country contexts. Additionally, we did not search the grey literature for practical reasons concerning the size of the literature, which may also have provided some additional insights not reflected in this publication. We also limited our search to two databases, which may mean we missed some relevant articles. However, we are confident that we found the majority of relevant published evidence to address the review questions because of a rigorous approach to retrieval. Thus, we think the findings of our citation analysis on the use of PARIHS are generalizable for studies in English published in peer-reviewed journals.

Conclusions
The importance of theoretically underpinned implementation science has been consistently highlighted. Theory is important for maximizing the chances of study transferability, providing an explanation of implementation processes, developing and tailoring implementation interventions, evaluating implementation, and explaining outcomes. This review of the use of the PARIHS framework, one of the most cited implementation frameworks, shows that its actual use and application has been frequently partial and generally not well described. Our ability to advance the science of implementation and ultimately affect outcomes will, in part, be dependent on better use of theory. Therefore, it is incumbent on theory developers to generate accessible and applicable theories, frameworks, and models, and for theory users to operationalize these in a considered and transparent way. We propose that the development and adoption of reporting guidelines on how framework(s) are used in implementation studies might enhance the maturity of implementation science.
Additional file 1: Form for initial assessment and form for data extraction