Introduction
Improving the quality and equitable delivery of evidence-based interventions, practices, programs, policies, and other health innovations (herein referred to collectively as EBIs) is critical for improving population health and addressing health inequities [Reference Brownson, Kumanyika, Kreuter and Haire-Joshu1]. The field of dissemination and implementation (D&I) science is particularly focused on how EBIs for health are delivered in real-world settings. Dissemination refers to the targeted distribution and communication of information, intervention materials, and evidence to specific audiences, whereas implementation is the adoption and integration of EBIs into routine practice or care [Reference Brownson, Colditz and Proctor2]. Key settings for improving the D&I of EBIs are within communities. Community-level action (i.e., change efforts in community and public health organizations and settings) to improve EBI access and delivery is vital [Reference Pinto, (Ethan), Miles and Ong3], given that opportunities to live a healthy life are largely determined by the communities in which individuals live, work, worship, and play [4]. Community-level action may seek to address barriers to the D&I of EBIs and/or leverage factors that facilitate EBI spread and delivery in community settings (e.g., through coalition building, intervention tailoring, and use of local opinion leaders, among other strategies) [Reference Powell, Waltz and Chinman5,Reference Leeman, Birken, Powell, Rohweder and Shea6].
One opportunity to support community-level action comes from attending to the social processes underlying the D&I of EBIs. Social processes, rooted in the relationships and interactions among a range of actors, play a major role in D&I success. There are many actors involved in community-level efforts, including organizations, decision-makers (e.g., policymakers and community leaders), individuals involved in disseminating and implementing EBIs (e.g., practitioners, educators, clinicians, and implementation support professionals), and EBI recipients (e.g., students, patients, and families) [Reference Luke, Morshed, McKay, Combs, Brownson, Colditz and Proctor7,Reference Burke, Lich, Neal, Meissner, Yonas and Mabry8]. In the context of D&I efforts, actors’ relationships within networks facilitate the exchange of information and resources, influence, advice sharing, mentoring, capacity building, and collaboration, among other social processes [Reference Pinto, (Ethan), Miles and Ong3,Reference Luke, Morshed, McKay, Combs, Brownson, Colditz and Proctor7].
To study the network structure of D&I actors’ relationships within communities and how these relationships influence D&I behaviors and social processes, we can utilize social network analysis (SNA) [Reference Light and Moody9,Reference Valente10]. SNA focuses “nodes,” which can be individuals or organizations, and the relationships between them (referred to as “ties”). Both nodes and ties can be defined by characteristics. For example, if nodes are individual people, they can be characterized by sociodemographic information. Ties can be characterized by the strength or directionality of the relationship (e.g., how often two individuals are in contact with each other). SNA can inform D&I efforts by helping to identify and characterize efficient pathways by which EBIs can be spread and scaled, explore social influences on decision-making processes, predict and explain tie formation patterns, examine connections between network characteristics and attitudes held by network members, and more [Reference Luke, Morshed, McKay, Combs, Brownson, Colditz and Proctor7,Reference Glegg, Jenkins and Kothari11]. For example, Palinkas and colleagues utilized SNA to examine the structure of a network of social service organizations in relation to decision-making around adoption (or rejection) of a novel EBI. Their mixed-methods evaluation offered (a) a holistic view of the patterning of connections within the network and the impact of that patterning, and (b) an understanding of what flowed across network connections [Reference Palinkas, Holloway, Rice, Fuentes, Wu and Chamberlain12]. Unlike conventional analytic approaches, which treat individuals as independent actors, SNA offered the opportunity to leverage the interconnected nature of network members in the analysis.
SNA provides a foundational understanding of networks’ structure, activities, and performance; this knowledge can be used to consider how to strengthen the effectiveness of networks for D&I efforts [Reference Luke, Morshed, McKay, Combs, Brownson, Colditz and Proctor7,Reference Light and Moody9,Reference Luke, Powell and Paniagua-Avila13,Reference Luke and Harris14]. Valente’s 2012 publication in Science on “network interventions” highlighted how the structure and characteristics of social networks can be leveraged to initiate, change, or maintain network behavior to positively impact health. Valente described four categories of interventions: identifying individuals to act as change agents (e.g., opinion leaders, champions, and individuals who bridge different groups); delivering an intervention to segments or subgroups in a network based on relationship structure (e.g., clusters or cliques) or attributes (e.g., sociodemographic characteristics); inducing or prompting change using network structure (e.g., word-of-mouth tactics for disseminating information, snowballing); and altering or changing the network structure itself by adding or removing people, groups, or connections [Reference Valente10]. Such network change interventions allow for purposeful adaptations that improve network function and outcomes. Despite the potential of these interventions, much of the work in the field of public health has focused on network interventions as applied to health behavior change interventions [Reference Hunter, de la and Murray15–Reference Gesell, Barkin and Valente17], rather than the using network interventions within D&I research.
In D&I research, network interventions offer a way to affect systems change and improve the utilization and dissemination of EBIs. Indeed, network change interventions are similar in concept to several established implementation strategies (methods and techniques to promote the implementation and sustainability of EBIs). For example, the Expert Recommendations for Implementing Change includes a category of “developing stakeholder interrelationships,” which includes specific implementation strategies like building a coalition, creating a learning collaborative, developing new clinical teams, identifying champions and early adopters, and promoting network weaving [Reference Powell, Waltz and Chinman5]. Similarly, in the taxonomy of D&I strategies offered by Leeman and colleagues [Reference Leeman, Birken, Powell, Rohweder and Shea6], the fundamental emphasis on the delivery system, support system, and synthesis/translation system inherently supports incorporating SNA perspectives. Further, a 2023 scoping review explicitly sought to identify network change strategies that could serve as implementation strategies and offers additional insight. Based on a review of 53 studies, Bunger and colleagues proposed a multilevel typology of network interventions that includes (a) actors (e.g., by changing an actor’s prominence in the network or motivation to connect with others), (b) relationship creation, strengthening, or dissolution (e.g., through incentives or training), and (c) context, defined as changing the environment, creating groups, and changing network composition or membership [Reference Bunger, Yousefi-Nooraie, Warren, Cao, Dadgostar and Bustos18].
SNA is not new to D&I research [Reference Provan, Veazie, Staten and Teufel-Shone19,Reference Valente, Chou and Pentz20], and networks were foundational to the Diffusion of Innovations theory [Reference Rogers21] upon which much of D&I is grounded. However, much of the literature connecting SNA and D&I focuses on the impact of network connections on organizational functioning (e.g., trust) rather than the spread or integration of EBIs [Reference Glegg, Jenkins and Kothari11]. Additionally, few extant D&I network studies focus on community-level action to help improve the D&I of EBIs [Reference Glegg, Jenkins and Kothari11]. To advance this work, it is critical to incorporate a solid theoretical foundation to strengthen how the field understands and uses implementation strategies that target network change processes [Reference Bunger, Yousefi-Nooraie, Warren, Cao, Dadgostar and Bustos18]. There are many D&I theories, models, and frameworks (TMFs), but it is unclear whether these TMFs provide sufficient specification regarding how to identify and leverage networks in D&I strategies aiming to improve the reach, utilization, spread, and sustainability of EBIs.
This is a lost opportunity, particularly given the importance of networks and relationships in D&I science and practice. Therefore, the objective of this research was to (1) examine the explicit presentation of social network concepts in D&I TMFs (i.e., concepts that address the structure of relationships and how these relationships influence behavior and social processes) that include community-level action and (2) identify opportunities to advance the study, specification, and use of networks in D&I efforts. This work is an important step toward moving beyond reporting on networks as part of context to strategically intervening on networks.
Methods
We conducted a narrative review of network concepts in D&I TMFs that target community-level change. Compared to other review methodologies, a narrative review aligns with our study objective as an approach to explore, identify, and descriptively summarize a broad range of concepts at the intersection of networks, communities, and D&I TMFs.
Selection of TMFs and TMF publications
The current work leverages an extant review published in 2021 by Pinto et al. of 74 community-level TMFs that describe aspects of community engagement in D&I research and practice – including communication, partnership exchange, community capacity building, leadership, and collaboration [Reference Pinto, (Ethan), Miles and Ong3]. Given (a) our similar focus on community-level change in D&I efforts and (b) the inherent role of social networks in community engagement processes (e.g., forging new collaborations), we built on the Pinto et al. review by bringing an explicit attention to networks in D&I TMFs. We used the collaborative, web-based Covidence platform [22] to manage the database of TMF publications cited in the Pinto et al. review, including grouping multiple publications for a single TMF in the same record to streamline data extraction. One TMF in the Pinto et al. review was listed in duplicate and merged for the current study. We contacted authors of original articles when necessary (i.e., to request full-text articles that we could not access via institutional subscriptions). From the existing Pinto et al. sample, we selected 66 TMFs for the current review, each with at least one corresponding publication, that met the following inclusion criteria (Supplementary File 1): relevant to community-level dissemination and/or implementation efforts (n = 5 excluded); available as full-text articles (n = 1 excluded); and available in English (n = 1 excluded). Potential exclusions from the Pinto et al. list were discussed as a team.
Data extraction of network concepts
We iteratively developed our data extraction template. We first conducted a preliminary review of nine commonly used TMFs in D&I science [Reference Rogers21,Reference Aarons, Hurlburt and Horwitz23–Reference Michie, van Stralen and West31] and noted network constructs and relevant information. The network information gathered in this preliminary review drove our data extraction template, which we pilot-tested and iteratively refined with six TMFs [Reference Aarons, Hurlburt and Horwitz23,Reference Moullin, Dickson, Stadnick, Rabin and Aarons24,Reference Kilbourne, Neumann, Pincus, Bauer and Stall32–Reference Farkas, Jette, Tennstedt, Haley and Quinn36]. Data extraction focused on descriptions of the TMF (i.e., the article narrative explaining the features and constructs of the TMF, in addition to the TMF figure or visual if applicable). Information in the included articles that described formative research, case studies, examples, and applications of the TMF were excluded from the data extraction process.
Using Covidence, two coders independently completed data extraction, with a third coder conducting consensus and the team reviewing any outstanding disagreements. For each included TMF, we extracted the TMF title and the year the TMF was first published. In collecting network information, our first step was to determine whether the TMF explicitly addressed networks – that is, the TMF elaborated on network characteristics, network structure, and/or network activities beyond simply acknowledging that relationships or partnerships are important for D&I efforts. For example, we determined that the Conceptual Model for the Diffusion of Innovations in Service Organizations is a “network-explicit TMF” because it includes several constructs describing relationship characteristics and their network structure (e.g., homophily, decentralization) [Reference Greenhalgh, Robert, Macfarlane, Bate and Kyriakidou26]. Conversely, while the Dynamic Sustainability Framework emphasizes the need for stakeholder partnerships to maintain the use of EBIs [Reference Chambers, Glasgow and Stange27], this TMF does not offer further information about network concepts and therefore we did not determine it to be network-explicit. Other examples of relational or social concepts that could be inferred from TMFs – but would not qualify as network-explicit without further network detail – include considerations of communication, social influence, sharing values and goals, social support, and implementation facilitation.
Table 1 defines key network concepts applicable to our TMF data extraction. For each network-explicit TMF, we extracted the following information:
Network concepts extracted from network-explicit TMFs
D&I = dissemination and implementation; EBI = evidence-based intervention; TMFs = theories, models, and frameworks.
-
Types of network actors and organizations (e.g., recipients, implementers);
-
Whether networks occur within and/or across organization(s) involved with D&I efforts;
-
What is built or exchanged across the network (e.g., information, innovation) and/or the outcomes of network activity (e.g., trust, social capital);
-
How relationships are characterized (e.g., based on tie strength or weakness);
-
How network structure is considered (e.g., size, centrality);
-
Guided by Valente’s taxonomy of network change interventions [Reference Valente10], ways in which the TMF proposes ways to change the structure or characteristics of networks.
A copy of our Covidence data extraction template is available in Supplementary File 2.
Data analysis
Among the network-explicit TMFs, we synthesized network concepts using descriptive statistics and narrative summaries.
Results
Of the 66 TMFs in our review sample, about one-third (n = 24; 36%) included explicit attention to social networks [Reference Rogers21,Reference Aarons, Hurlburt and Horwitz23,Reference Moullin, Dickson, Stadnick, Rabin and Aarons24,Reference Greenhalgh, Robert, Macfarlane, Bate and Kyriakidou26,Reference Damschroder, Aron, Keith, Kirsh, Alexander and Lowery29,Reference Kilbourne, Neumann, Pincus, Bauer and Stall32,Reference Yuan, Nembhard, Stern, Brush, Krumholz and Bradley35–Reference Hanson, Schoenwald and Saunders60] (Supplementary File 3). These network-explicit TMFs were initially published between 1969 and 2016. Below we summarize our findings according to network concepts outlined in Table 1. We describe example TMFs to demonstrate the range of network considerations.
Nodes
As shown in Table 2, the 24 network-explicit TMFs included a diverse range of actors. Commonly included actors were implementers (n = 22), adopters or decision-makers (n = 22), innovation developers (n = 18), implementation support professionals (n = 14), and innovation recipients (n = 12). Fewer frameworks discussed the network roles of policymakers (n = 11) and funders (n = 5). At the organizational level, most TMFs described implementing or disseminating agencies (n = 20), and more specifically, community-based, or faith-based organizations (n = 16) involved in D&I efforts.
Types of actors in network-explicit TMFs (n = 24)
TMFs = theories, models, and frameworks.
Note: Types of actors are listed in descending order based on frequency. Multiple selections were allowed.
Ties within and across organizations
One TMF (4%) addressed intra-organizational networks only, three TMFs (13%) addressed inter-organizational networks only, and 20 TMFs (82%) addressed both (Table 3). Some TMFs such as the Consolidated Framework for Implementation Research (CFIR) [Reference Damschroder, Aron, Keith, Kirsh, Alexander and Lowery29,Reference Damschroder, Reardon, Widerquist and Lowery39] and Exploration, Preparation, Implementation, and Sustainment (EPIS) framework [Reference Aarons, Hurlburt and Horwitz23,Reference Moullin, Dickson, Stadnick, Rabin and Aarons24] describe intra- and inter-organizational network structure according to the inner and outer context for implementation. Notably, the Conceptual Model for the Diffusion of Innovations in Service Organizations emphasizes the importance of networks at multiple levels and phases of implementation, including internal and external networks that support knowledge sharing as a “system antecedent for innovation,” networks of the adopter within the “user system” where implementation occurs, and inter-organizational networks in the outer context that influence norms and the broader implementation environment [Reference Greenhalgh, Robert, Macfarlane, Bate and Kyriakidou26].
Characteristics of network-explicit TMFs (n = 24)
TMFs = theories, models, and frameworks.
Note: TMFs are listed in ascending order by year first published. Further details about each framework are available in Supplementary File 3.
Tie type and purpose
Network function varied across TMFs, but commonly included the flow of resources, knowledge, trust, norms, goals, and values to influence the dissemination and/or implementation of innovations. For example, the Stages of Research Utilization Model [Reference Davis, Peterson, Helfrich and Cunningham-Sabo40] emphasizes the exchange of information and knowledge about the program innovation between the “resource system” consisting of innovation developers and the “user system” consisting of implementing organizations. Additionally, several frameworks – including the AIMS (Approach/Engagement, Implementation, Monitoring, Sustainability) model [Reference Smith, Wise, Rosen, Rosen, Childs and McManus57], the CFIR [Reference Damschroder, Aron, Keith, Kirsh, Alexander and Lowery29,Reference Damschroder, Reardon, Widerquist and Lowery39], and the Framework of Dissemination in Health Services Intervention Research [Reference Mendel, Meredith, Schoenbaum, Sherbourne and Wells53] – discussed how networks build collective efficacy for D&I efforts by empowering network members and fostering a sense of community.
Tie characteristics
Approximately half of the network-explicit TMFs (n = 13; 54%) described factors that characterize ties or relationships (i.e., qualifying information beyond the existence of ties) (Table 3). For example, Diffusion of Innovations [Reference Rogers21] and the AIMS model [Reference Smith, Wise, Rosen, Rosen, Childs and McManus57] make the distinction between strong and weak ties. Strong ties made of trusting relationships are important for providing social support, whereas weak ties may be better suited for spreading novel information [Reference Light and Moody9]. Several TMFs described tie strength based on the level of trust between network members. For example, the Framework for Knowledge Translation posits that knowledge transfer between the research and user groups will be more successful if their relationships have a higher level of trust, rapport, and prior history of working together [Reference Jacobson, Butterill and Goering52]. Many TMFs [Reference Rogers21,Reference Greenhalgh, Robert, Macfarlane, Bate and Kyriakidou26,Reference Damschroder, Aron, Keith, Kirsh, Alexander and Lowery29,Reference Damschroder, Reardon, Widerquist and Lowery39,Reference Dearing, Maibach and Buller41,Reference Dearing, Rogers and Meyer42,Reference Hanson, Saunders, Ralston, Moreland, Peer and Fitzgerald50,Reference Hanson, Schoenwald and Saunders60] also use the concept of homophily (or conversely, heterophily) in characterizing ties based on the similarity (or difference) of connected actors according to a given characteristic [Reference McPherson, Smith-Lovin and Cook61]. In these TMFs, homophilous ties based on concordant occupation, professional role, and/or cultural background may facilitate the adoption of innovations.
Network structural characteristics
Most network-explicit TMFs (n = 17; 71%) characterized element(s) of network structure, such as size, centrality, density, and bridging (Table 3). Related to the role of networks in dissemination strategies, the Blueprint for Dissemination emphasizes (a) the importance of network size in generating a threshold of participating organizations to maximize information exchange, and (b) a “nodal structure” consisting of a centralized organizing body at the national level paired with local organizations to drive progress in dissemination [Reference Yuan, Nembhard, Stern, Brush, Krumholz and Bradley35]. The Conceptual Model for the Diffusion of Innovations in Service Organizations discusses network structure at length, including the distinction between decentralized, horizontal networks that are helpful for informal diffusion versus centralized, vertical networks that are often needed for active dissemination [Reference Greenhalgh, Robert, Macfarlane, Bate and Kyriakidou26]. Related to network density, the Convergent Diffusion and Social Marketing Approach to Dissemination describes the difference between cohesive, interconnected “ecologic communities” made up of local “cooperating and competing organizations” versus the less integrated network structure of “societal sectors” (e.g., government agencies, non-profit organizations) [Reference Dearing, Maibach and Buller41,Reference Dearing, Rogers and Meyer42]. Several TMFs emphasize the specific role of actors as “bridges” across groups or “boundary spanners,” including those that connect the inner setting where implementation occurs with external organizations in the outer setting [Reference Aarons, Hurlburt and Horwitz23,Reference Moullin, Dickson, Stadnick, Rabin and Aarons24,Reference Greenhalgh, Robert, Macfarlane, Bate and Kyriakidou26,Reference Glisson and Schoenwald49,Reference Robinson, Elliott and Driedger56]. For example, a key network feature of the Linking Systems framework is the presence of a “linking agent” who brokers knowledge and connects the gap between innovation developers and users [Reference Robinson, Elliott and Driedger56].
Network change interventions
From most TMFs, we could infer at least one approach to changing the structure or characteristics of networks involved in D&I efforts (n = 21; 88%). Following Valente’s categorization of network change interventions [Reference Valente10], the most common network change approach was identifying individuals – for example, opinion leaders, innovators, early adopters, program or organizational champions, change agents, coaches, bridges, and knowledge brokers – to lead or support D&I efforts (n = 20) [Reference Rogers21,Reference Aarons, Hurlburt and Horwitz23,Reference Moullin, Dickson, Stadnick, Rabin and Aarons24,Reference Greenhalgh, Robert, Macfarlane, Bate and Kyriakidou26,Reference Damschroder, Aron, Keith, Kirsh, Alexander and Lowery29,Reference Kilbourne, Neumann, Pincus, Bauer and Stall32,Reference Yuan, Nembhard, Stern, Brush, Krumholz and Bradley35,Reference Farkas, Jette, Tennstedt, Haley and Quinn36,Reference Bowen and Zwi38–Reference Dearing, Rogers and Meyer42,Reference Elliott, O’Loughlin and Robinson45,Reference Farkas and Anthony47–Reference Smith, Wise, Rosen, Rosen, Childs and McManus57,Reference Hanson, Schoenwald and Saunders60]. Several TMFs also described network change approaches related to segmentation (n = 10) [Reference Rogers21,Reference Aarons, Hurlburt and Horwitz23,Reference Moullin, Dickson, Stadnick, Rabin and Aarons24,Reference Greenhalgh, Robert, Macfarlane, Bate and Kyriakidou26,Reference Damschroder, Aron, Keith, Kirsh, Alexander and Lowery29,Reference Kilbourne, Neumann, Pincus, Bauer and Stall32,Reference Yuan, Nembhard, Stern, Brush, Krumholz and Bradley35,Reference Damschroder, Reardon, Widerquist and Lowery39,Reference Dearing, Maibach and Buller41,Reference Dearing, Rogers and Meyer42,Reference Havelock51,Reference Pettigrew, Ferlie and McKee54,Reference Robinson, Elliott and Driedger56], induction (n = 8) [Reference Rogers21,Reference Aarons, Hurlburt and Horwitz23,Reference Moullin, Dickson, Stadnick, Rabin and Aarons24,Reference Greenhalgh, Robert, Macfarlane, Bate and Kyriakidou26,Reference Damschroder, Aron, Keith, Kirsh, Alexander and Lowery29,Reference Yuan, Nembhard, Stern, Brush, Krumholz and Bradley35,Reference Damschroder, Reardon, Widerquist and Lowery39,Reference Dearing, Maibach and Buller41,Reference Dearing, Rogers and Meyer42,Reference Hanson, Saunders, Ralston, Moreland, Peer and Fitzgerald50,Reference Pettigrew, Ferlie and McKee54,Reference Hanson, Schoenwald and Saunders60], and alteration (n = 11) [Reference Aarons, Hurlburt and Horwitz23,Reference Moullin, Dickson, Stadnick, Rabin and Aarons24,Reference Yuan, Nembhard, Stern, Brush, Krumholz and Bradley35,Reference Anderson, Cosby, Swan, Moore and Broekhoven37,Reference Elliott, O’Loughlin and Robinson45,Reference Fixsen, Naoom and Blase48,Reference Hanson, Saunders, Ralston, Moreland, Peer and Fitzgerald50–Reference Jacobson, Butterill and Goering52,Reference Pettigrew, Ferlie and McKee54–Reference Smith, Wise, Rosen, Rosen, Childs and McManus57,Reference Hanson, Schoenwald and Saunders60]. Segmentation, for example, is described in the Replicating Effective Programs framework with the intentional tailoring of intervention approaches and dissemination efforts for different recipient populations and implementing organizations [Reference Kilbourne, Neumann, Pincus, Bauer and Stall32]. Considerations of induction are often related to the deliberate spread of innovations or information throughout a network. This approach is exemplified in the Blueprint for Dissemination, which offers guidance for sharing best practices among peer organizations in a learning network and “generating a threshold of participating organizations that maximizes network” [Reference Yuan, Nembhard, Stern, Brush, Krumholz and Bradley35]. Network alteration approaches included the creation of inter-organizational relationships, for example, through building coalitions [Reference Yuan, Nembhard, Stern, Brush, Krumholz and Bradley35,Reference Elliott, O’Loughlin and Robinson45,Reference Riley, Stachenko and Wilson55] or “Community Change Teams” as described in the Community-based Learning Collaborative Model [Reference Hanson, Saunders, Ralston, Moreland, Peer and Fitzgerald50,Reference Hanson, Schoenwald and Saunders60]. Changes to network structure were also discussed in the context of knowledge translation or brokerage by establishing new ties to exchange information across groups or communities [Reference Jacobson, Butterill and Goering52,Reference Robinson, Elliott and Driedger56].
Discussion
This review examined the presence and presentation of core social network concepts in community-level D&I TMFs to identify opportunities to study and leverage social networks in D&I science. Among 66 relevant TMFs, we found that about one-third (n = 24; 36%) explicitly attended to social networks, and among those that did, there was an emphasis on network change involving individual actors. Our examination of network-explicit TMFs highlights opportunities to think more broadly about network change interventions and outcomes. Readers interested in applying network concepts in D&I research and practice may use the results of our review to help select appropriate TMFs to guide network-related inquiry in several areas or entry points. This may include considerations of existing organizational structures or groups (e.g., coalitions, community advisory boards, learning collaboratives, government), network features or characteristics (e.g., power and hierarchy, common language and values, organizational context and culture, relationships within and across sectors), and outcomes of network activity on practice and policy (e.g., knowledge translation, capacity building, fostering trust, provision of technical assistance).
There is interest in D&I science to incorporate perspectives of whole implementation systems and affect systems change among organizations and communities involved in disseminating and implementing EBIs [Reference Luke, Morshed, McKay, Combs, Brownson, Colditz and Proctor7,Reference Gesell, Barkin and Valente17,Reference Valente, Chou and Pentz20,Reference Valente, Palinkas, Czaja, Chu and Brown63]. Our findings highlight several D&I TMFs focused on community-level change that give explicit attention to networks, which can support opportunities for intervention. For example, the systems-oriented nature of the EPIS framework – which focuses on a range of actors (including policymakers, funders, advocacy groups, and payors) and inter-organizational networks (including contractual links or academic–community partnerships) – offers a strong foundation for guiding network change in support of pre-implementation, implementation, and sustainment phases [Reference Aarons, Hurlburt and Horwitz23,Reference Moullin, Dickson, Stadnick, Rabin and Aarons24]. As a result, empirical projects grounded in EPIS have examined and intervened on social networks to support the adoption and high-quality implementation of EBIs [Reference Palinkas, Holloway, Rice, Fuentes, Wu and Chamberlain12]. Building on EPIS, the Community-based Learning Collaborative Model [Reference Hanson, Saunders, Ralston, Moreland, Peer and Fitzgerald50,Reference Hanson, Schoenwald and Saunders60] provides additional guidance on specific network characteristics and change processes throughout various stages of implementation. D&I researchers aiming to understand implementation determinants from a network and systems perspective may consider the Updated CFIR published in 2022, with attention to constructs delineating actors’ roles, network connections between the inner and outer setting, and “teaming” processes to collaboratively implement an innovation [Reference Damschroder, Reardon, Widerquist and Lowery39]. The results of this work do not suggest that a broad range of TMFs are inappropriate for use in D&I science. Instead, a large number may benefit from expansion to incorporate a network perspective. Of course, as noted by Nilsen, combining theoretical concepts requires careful consideration of the underlying philosophies and worldviews represented to ensure coherence in the integrated framework [Reference Nilsen, Nilsen and Birken64].
As noted earlier, a range of network change interventions may be important potential implementation strategies, such as building coalitions [Reference Powell, Waltz and Chinman5,Reference Valente10]. However, the observed emphasis on network change strategies focused on individual behavior change points to the need to think more broadly about network-based D&I strategies. The focus on individual-level strategies is unsurprising given the well-documented roles of opinion leaders [Reference Flodgren, Parmelli and Doumit65] and champions [Reference Miech, Rattray, Flanagan, Damschroder, Schmid and Damush66], among other change agents, in the D&I literature. This finding also aligns with that of a 2023 review of network change strategies for implementation, which also observed a high prevalence of individual-level network targets [Reference Bunger, Yousefi-Nooraie, Warren, Cao, Dadgostar and Bustos18]. That work highlighted the potential for change in the network context and the network environment to reshape systems supporting community-level efforts to implement or disseminate EBIs [Reference Bunger, Yousefi-Nooraie, Warren, Cao, Dadgostar and Bustos18]. This broadened focus beyond individual action may also help advance equity by shifting expectations of who is involved and/or at the center of D&I networks. Leveraging existing networks of partners, such as community coalitions [Reference Provan, Veazie, Staten and Teufel-Shone19,Reference Valente, Chou and Pentz20] and professional learning communities [Reference Bunger, Hanson, Doogan, Powell, Cao and Dunn67], may also provide an important foundation to promote equity. By considering implementation systems more holistically, D&I scientists can take advantage of the potential for D&I efforts to not only support the spread of a given innovation through a system but also reshape that system to be more equity-promoting [Reference Ramanadhan, Aleman and Bradley68].
We found that most network-explicit TMFs included a dual focus on ties within and across organizations. This emphasizes the theoretical importance of understanding and considering intra- and inter-organizational relationships for D&I research and how such diversity of tie structure may influence implementation outcomes. For example, with a focus on building social capital for D&I efforts, the Implementation Capital Framework published in 2019 [Reference Neal and Neal69] highlights the impact of high bonding social capital (i.e., strong within-group ties and clustering) on sharing norms related to an EBI’s acceptability and appropriateness. This could be highly relevant for intra-organizational connections, as an example. At the same time, the framework emphasizes how high levels of bridging social capital (i.e., strong between-group ties) can afford actors access to diverse information and resources and impact an EBI’s feasibility and fidelity.
Another important insight from this review is the need for future research to explore the dynamic (and malleable) nature of social networks. Thus, after network change interventions are deployed, changes in social networks could serve as an important implementation outcome. To further this line of inquiry, clear specification of the purpose and expected impact of network-focused implementation strategies is needed. Measurement will also be strengthened by pairing network theory with existing SNA tools that describe, visualize, and model network changes over time [Reference Luke, Morshed, McKay, Combs, Brownson, Colditz and Proctor7]. There are increasing opportunities for researchers to harness SNA tools through the growth of open science/data and open-source initiatives and platforms [70,71]. The push for publicly available research data means that a growing numbers of social network datasets and analysis programming code can serve as roadmaps for researchers to conduct SNA. Likewise, growing knowledge in open-source programming languages, such as R and Python, opens to door for social network experts to collaborate on packages specifically designed to aid in SNA [Reference Csárdi, Nepusz and Müller72]. This extends to open-source tools like Gephi Lite which allow for a user-friendly web-based interface to upload and interact with their social network data [Reference Lite73]. These technological advances drive the feasibility for researchers to incorporate and evaluate network-focused implementation strategies.
We also acknowledge that the quality, relevance, and impact of SNA efforts in D&I science are likely to be much higher if a diverse range of actors are involved throughout [Reference Luke, Morshed, McKay, Combs, Brownson, Colditz and Proctor7,Reference Ramanadhan, Aleman and Bradley68,Reference Carothers, Allen and Walsh-Bailey74]. Further details about participatory SNA work in D&I are available elsewhere [Reference Luke, Morshed, McKay, Combs, Brownson, Colditz and Proctor7]; however, connected to participatory work is understanding social networks in the context of advancing health equity. There are a wide range of opportunities, from increasing inclusion of diverse expertise through strategic partnership composition [Reference Ramanadhan, Salhi and Achille75] to seeking transformational change to reduce inequities for the focal condition as well as others [Reference Ramanadhan, Aleman and Bradley68]. Future research is needed to explore these opportunities further.
In sum, our review offers several important research opportunities for SNA in D&I science: improving examination of networks as determinants; further developing strategies for network change and network-focused implementation strategies; advancing use of network measures to understand the impact of such network change strategies; and broadening our understanding of how networks can help advance health equity. Advancing D&I science in these areas may benefit from participatory approaches that engage diverse community partners representing the real-world settings and contexts in which the D&I of EBIs occur [Reference Ramanadhan, Aleman and Bradley68]. Partnerships with community-based organizations, public health departments, and local policymakers, for example, can inform (a) the section and potential tailoring of appropriate TMF(s) to guide network-related inquiry and practice, and (b) the selection, tailoring, and testing of network-focused implementation strategies on equity-focused outcomes related to dissemination, implementation, and community health.
This narrative review has important limitations. First, our findings are based on community-level TMFs included in the Pinto et al. sample of TMFs [Reference Pinto, (Ethan), Miles and Ong3]. While this sampling frame is appropriate given our focus on community-level change for disseminating and/or implementing EBIs, other relevant TMFs may have been excluded from the Pinto et al. review, and by extension, our review. Another source of potential TMFs for interested readers is the “Dissemination & Implementation Models in Health” interactive webtool (which includes a community-level filter) [76]. This repository builds on the Tabak et al. 2012 review of D&I TMFs [Reference Tabak, Khoong, Chambers and Brownson77], which was a key data source for the Pinto et al. article. Second, our review included TMFs originally published over the past several decades, and given our sampling approach, we may have excluded more recently developed TMFs. However, older TMFs represent the important history and multidisciplinary foundations of D&I science. Many older TMFs remain highly salient by guiding current empirical D&I work, the development of new TMFs, and ongoing TMF refinement. Third, our review included articles that presented TMFs and not the range of applications that may include additional detail regarding networks’ roles in the dissemination or implementation of specific EBIs. However, the study’s goal was to examine the guidance offered generally through TMF descriptions (which other researchers can draw upon); thus, the pool of included data sources seemed appropriate. Fourth, as with any narrative review, the data reflect what team members were and were not able to parse from the TMFs as originally presented. We minimized the risk of omission or miscoding by utilizing dual independent coding, having a third consensus reviewer, and team discussions.
At the same time, the strengths of this review offer important benefits. We provide a novel network lens to a large set of community-level D&I TMFs. Our documentation of network characteristics in these TMFs may help D&I researchers identify entry points and opportunities for integrating network perspectives in their work. We also identify key areas for future SNA and D&I research to help purposively guide the field moving forward.
Conclusions
Our findings highlight several community-level D&I TMFs that emphasize networks. Among frameworks with an explicit network focus, the variability of network actors and activities suggests an opportunity to advance the understanding of diverse social drivers of D&I processes. Further research is needed to examine networks (at multiple levels) as determinants, network change interventions as implementation strategies, and/or network change as an implementation outcome in community-level D&I efforts.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/cts.2024.614.
Acknowledgments
We thank Madison Rivard for her help with data extraction and project management.
Author contributions
ARK and SR conceptualized the review. All authors contributed to article screening and data extraction. ARK, JLC, NRS, and SR synthesized review findings. All authors helped interpret findings. ARK and SR wrote the first draft of the manuscript. All authors (ARK, JLC, NRS, RRJ, MC, WS, and SR) provided substantive feedback and read and approved the final manuscript. ARK takes responsibility for the manuscript as a whole.
Funding statement
This work was supported by the National Institutes of Health (NRS, grant number K99CA277135; grant numbers 5P50CA244433, P50CA244431, 5T32CA057711) and the Centers for Disease Control and Prevention (grant number U48DP006395).
Competing interests
The authors declare that they have no conflicts of interests.
Open access
