Five years of the Implementation Research Logic Model: a citation analysis and systematic scoping review of its use in implementation research and practice Authors: James L Merle, Catlin A Golden, Rebecca Simmons, Elizabeth A Sloss, Olutobi A Sanuade, Teresa Winning Verry, Anthony J Nixon, Mary M McFarland, Rebecca Lengnick-Hall, Rosemary Meza, Jennie L Hill, Paul A Estabrooks, Miriam R Rafferty, Dennis Li, Justin D Smith Published: 2026 ABSTRACT AbstractBackgroundThe Implementation Research Logic Model (IRLM) was developed to enhance the rigor, transparency, and reproducibility of implementation research by integrating key elements of implementation research and practice—contextual factors/determinants, implementation strategies, mechanisms, and implementation and health outcomes—into a unified framework. Since its publication in 2020, the IRLM has been widely applied to plan, evaluate, and communicate implementation research. This paper synthesizes diverse IRLM use cases to advance its application and accelerate discovery in implementation science.MethodsWe conducted a citation analysis and systematic scoping review to assess the uptake, application, and adaptation of the IRLM. A forward citation search identified 331 citing articles, and a scoping review identified 124 additional studies using other logic models. Articles were coded for IRLM use, phase of application, construct linking, data sources, community engagement, and framework integration. Descriptive statistics and thematic/pattern analysis were used to summarize findings.ResultsOf the 331 citing articles, 118 included a completed IRLM. Most were used during the planning phase (68%), with fewer applied to executing (6%), reporting (22%), or synthesis (4%). The IRLM was cited across diverse disciplines and countries, but few studies used it across multiple phases or reported its use in reporting or synthesis. Only 14% demonstrated comprehensive causal logic linking across core elements. In contrast, other logic models often lacked integration of implementation-specific constructs. Community and partner engagement was reported in 61% of studies.ConclusionsThe IRLM can be a valuable tool for the field of implementation science but remains underutilized across the phases of implementation research with inconsistent applications of the recommendations from the 2020 article. Enhancing guidance on construct linking, strategy mechanisms, and partner engagement may improve its utility and impact in advancing equitable and rigorous implementation efforts.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13012-026-01514-2. This is the first comprehensive literature review of how the Implementation Research Logic Model (IRLM) has been used in research and practice. This is the first comprehensive literature review of how the Implementation Research Logic Model (IRLM) has been used in research and practice. We show that the IRLM has been widely adopted across the field but is mostly applied during planning, with limited use for later phases (i.e., executing, reporting, and synthesizing). We show that the IRLM has been widely adopted across the field but is mostly applied during planning, with limited use for later phases (i.e., executing, reporting, and synthesizing). Our review highlights important gaps that point to areas for refinement and increased guidance for IRLM users, such as causal linking approaches, inclusion of mechanisms, and involvement of community and implementation partners. Our review highlights important gaps that point to areas for refinement and increased guidance for IRLM users, such as causal linking approaches, inclusion of mechanisms, and involvement of community and implementation partners. By comparing the IRLM with other logic models, we show how it uniquely integrates key implementation science concepts using a generalized theory. By comparing the IRLM with other logic models, we show how it uniquely integrates key implementation science concepts using a generalized theory. Introduction Background and rationale Implementation science aims to improve the uptake, integration, and sustainability of evidence-based practices. Central to this goal is the need for tools that enable researchers and practitioners to systematically specify, track, and evaluate key elements of implementation research—chiefly among them are contextual determinants, implementation strategies, strategy mechanisms of action, and implementation outcomes [1–11]. As the field has matured, scholars have increasingly called for structured methods to enhance the transparency, rigor, and reproducibility of implementation research, enabling insight into how strategies work, the contextual conditions influencing them, and creating generalizable knowledge [12–14]. The Implementation Research Logic Model (IRLM) was developed to address these challenges by offering a logic model specific to implementation research that integrates the aforementioned key elements [1]. Designed with flexibility in mind, the IRLM provides a generalized theory of implementation research and a semi-structured method to guide the planning, execution, reporting, and synthesis of implementation studies [1]. The IRLM is also useful for implementation practice in the absence of formal research or evaluation [15]. Since its publication in 2020, the IRLM article has been accessed over 153,000 times, ranks in the top 5% of all research outputs ever scored by Altmetric [16], and has been required in several funding opportunity announcements by federal funding agencies [17]. Furthermore, the relative citation ratio, which represents a citation-based measurement of research influence [18], is 28, which is in the 99.7th percentile compared to all other NIH funded papers [19]. It has also been disseminated through educational and training programs, workshops, and institutes, while being cited across diverse health and human service domains [20–23]. Despite its growing use, there has been no comprehensive synthesis of the application or potential improvement of the IRLM to increase its scientific and practical utility for both research and practice. Revisions and updates to implementation research methods, models and frameworks are common as the field advances and new insights emerge about the complex processes underlying implementation. For example, the Consolidated Framework for Implementation Research (CFIR), Reach, Effectiveness, Adoption, Implementation, Maintenance (RE-AIM) framework, and integrated-Promoting Action on Research Implementation in Health Services (i-PARIHS) framework underwent formal reviews that resulted in updated versions published more than a decade after their initial introductions [2, 11, 24]. The Exploration, Preparation, Implementation, Sustainment (EPIS) framework published an updated version after 7 years (2012–2019) [25], and the 2015 update of the Expert Recommendations for Implementing Change (ERIC) taxonomy replaced an earlier, less expansive taxonomy [5]. Although assessing a widely used implementation model/framework is labor-intensive, it serves important purposes (a): raising awareness and facilitating uptake (b), guarding against misuse and informing its target guidance, and (c) pooling learnings to build capacity for more rigorous, impactful use. Several important aspects of the IRLM require examination to understand how current use can inform ongoing improvements to the method and its application. First, understanding how the IRLM is being cited and conceptualized can illuminate its role in shaping implementation research and practice and inform areas for refinement [26, 27]. Although citation counts indicate growing visibility, a rigorous citation analysis is needed to understand whether references to the IRLM reflect superficial mention, methodological guidance, or substantive use. Second, the IRLM has utility across the phases of implementation research and practice, yet when it is applied are unclear—whether in planning, executing, reporting, or synthesizing efforts. Third, the types of data sources used to populate IRLMs may vary widely across studies and have not been systematically categorized or described, making it difficult to assess the rigor and consistency of their application. Fourth, consistent with the generalized theory of implementation research, the degree to which IRLM users specify the “links” between the core elements in a process are referred to as Context, Strategies, Mechanisms, Outcomes (CoSMOs) linking [28]. CoSMOs is a defining feature of the IRLM that we posit as being critical to maximizing its utility and producing the desired rigor, reproducibility, and transparency of implementation research and practice. Last, although the IRLM was designed to support participatory and constituent-driven approaches [1], the extent of community/partner involvement is unknown. This gap is salient as the field moves toward greater emphasis on equity, co-creation, and accountability to community and implementation partners [14, 29, 30]. Although frameworks exist to evaluate stakeholder engagement in implementation research [31, 32], no review has assessed the level of engagement specifically at it pertains to the IRLM. There are currently no standards for how to develop IRLMs in collaboration with community partners in ways that are locally relevant, include generalizable hypotheses, and are actionable in practice. Study aims The aim of this citation analysis and systematic scoping review was to capture: (a) how the IRLM has been applied in the literature, (b) during which phases of implementation, (c) the templates/versions used most often, (d) the thoroughness of IRLM completion by users, and (e) user-reported limitations and suggested improvements to enhance usability. We also sought to identify the sources of data used to populate or inform the IRLM (e.g., literature review, theory, interviews), and to assess the extent of community/partner engagement, using a structured community engagement spectrum [32]. Method Citation analysis As described in the study protocol [33, 34], a citation analysis was conducted to map the landscape and assess the scientific spread and scholarly impact of the IRLM. To identify citing articles, a direct, forward citation search within Google Scholar was conducted [35]. Coders (CG, RS, EAS, OS, TWV, AJR, JLM) used content-based citation analysis, which includes both semantic and syntactic analysis of citing articles, such as where in the manuscript the citation appears (syntactic) and the context or purpose of the citation (semantic) [36]. For example, we extracted which section of the manuscript the IRLM citation and/or figure appeared (i.e., introduction, methods, results, discussion, supplemental file) as well as the reason for the citation (e.g., to discuss logic models generally, to support use of their own IRLM). The citation analysis was conducted on studies published between January 1, 2018 (following the first national presentation of the IRLM) and December 31, 2024. Scoping review We conducted our systematic scoping review with guidance from the JBI Manual for Evidence Synthesis [37]. Protocol registration at http://osf.io/2v6u5/(33). We chose the scoping review methodology for its structured process to identify and map the evidence on logic modeling in implementation research or practice, utilizing the framework as outlined by Arksey and O’Malley [38, 39]. We adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for scoping review reporting guidelines (PRISMA-ScR) [40–42]. Searches were conducted in Medline (Ovid), Embase (Elsevier), CINAHL Complete (Ebscohost), Cochrane Library (Wiley), APA PsycINFO (Ebscohost), ProQuest Dissertations & Theses Global, Scopus (Elsevier) and Web of Science Core Collection (Clarivate) on 09-19-24 with a date limit from 2018. See Additional File 1 for the search strategy and Additional Files 2 and 3 for PRISMA-S and PRISMA-ScR checklists. We excluded conference proceedings in our search due to our reliance on IRLM images for complete evaluation [20–23]. To select studies, we included records that (a) were conducted during or after the year 2018, which was when the IRLM was first presented at a national conference [43] and (b) were related to logic modeling and implementation research or practice. Non-English studies were excluded at full-text review due to funding constraints for translation services. Records that used the IRLM identified in the scoping review were removed due to duplication with the citation analysis. See Table 1 for inclusion/exclusion criteria. Trained readers screened titles and abstracts, then reviewed the full text using Covidence review software [44]. Two readers from a pool of seven independently reviewed each record, and discrepancies were reconciled by the senior author (JDS). Assessing the quality of the IRLM itself rather than the overall study in which it was conducted was a formal aim of the review. See Fig. 1 for the PRISMA routes. Table 1Study inclusion and Exclusion criteriaCitation AnalysisScoping ReviewInclusionExclusionInclusionExclusionCited 2020 IRLM PaperNot in EnglishConducted after the year 2018Conference proceedingsRelated to logic modeling and implementation research or practiceUsed the IRLMPublished in EnglishDid not include a logic model Study inclusion and Exclusion criteria PRISMA diagram. Note: *articles that cited/used the IRLM and were identified in the scoping review were removed due to duplication with the citation analysis. Source: page MJ, et al. BMJ 2021;372:n71. doi: 10.1136/bmj.N71. This work is licensed under CC BY 4.0. To view a copy of this license, visit https://creativecommons.org/licenses/by/4.0/ Review team characteristics and training The review team consisted of doctoral- and/or postdoctoral-level scientists trained in implementation science and two graduate students. Senior-level researchers, including the IRLM co-developers (JDS, DHL, MRR) and two others (PAE, JLH) oversaw early-career implementation science faculty, postdoctoral researchers, and graduate students (JLM, EAS, OAS, RLH, RM, CAG, RGS, TWV, AJN). Standing meetings were used to clarify concepts and to discuss questions about the coding protocol, codebook refinement and prevent coder drift. Data extraction and coding Study-level data from the corpus of records identified by the citation analysis and the scoping review were extracted using Covidence [44] (e.g., author, title, journal, year published, study setting). If an IRLM figure was included, the following categories were extracted: (a) the IRLM template/version used; (b) the study design phase(s) in which the IRLM was completed; (c) the comprehensiveness of data elements included in the IRLM, as indicated by including information and performing comprehensive (all elements) or partial (only some elements) CoSMOs linking (e.g., linking context-strategy, strategy-mechanism, and mechanism-outcome relationships using text color or superscripts); (d) whether the valence or direction of the contextual factors or determinants is indicated; (e) which frameworks or taxonomies were used; (f) the degree to which the study team engaged a community or implementation partner; and (g) any shortcomings and/or recommendations for improving the IRLM. For category (f) we used an adapted set of criteria from the Clinical and Translational Science Awards Consortium’s Principles of Community Engagement to code each article’s engagement level specifically with the IRLM on a scale from 0 (no engagement) to 5 (shared leadership), with higher scores indicating greater community engagement [32] (Table 2). If no IRLM figure was included, studies were coded for (a) citation purpose and location of citation in the article and (b) any noted benefits, deficiencies, or recommendations for improving the IRLM. For records identified via scoping review, we extracted which logic model was used (e.g., program evaluation logic model), the stated purpose for using the logic model, and unique features that were different from the IRLM. Table 2Partner engagement categoriesCategoryDefinition0: No EngagementAuthors or researchers did not engage community partners or use any of their data in the development or use of the IRLM.1: OutreachOne-way communication to provide information or collect input; entities co-exist.2: ConsultResearchers seek feedback or input from community partner but retain full decision-making control3: InvolveCommunication flows both ways; entities cooperate with each other; Stakeholders are included throughout key stages to ensure their concerns are considered.4: CollaborateShared planning and decision-making; researchers and partners work together as co-creators of the IRLM; bi-directional communication channels are established; partnership with community evident on each aspect of the IRLM from development to solution.5: Shared LeadershipStrong Bidirectional Relationship; Final decision making is at the community level; Community or implementation partners lead or co-lead the effort; entities have formed strong partnership structures. Partner engagement categories Our full codebook (Additional File 4) includes expanded definitions and examples as well as decision rules for all codes and extracted data. Each study was double coded with a consensus round conducted by the senior research team (JLM & JDS). Extraction training involved synchronous training sessions followed by assigned practice articles. Additional details as well as the final codebook with operational definitions is included in our protocol [33]. Analysis For the citation analysis, yearly citation trends were analyzed, and journals were used to conduct a field analysis (i.e., what disciplines are citing the IRLM, which country the journal is associated with) using SCImago (https://www.scimagojr.com/). Both broad subject areas and more granular subject categories were collected as well as the Country associated with the journal. For the scoping review, the extracted quantitative data were analyzed through descriptive statistics. Key insights were summarized, and general thematic/pattern analysis was used to capture strengths and limitations of the IRLM and purpose for citing the IRLM. Results The PRISMA diagram (Fig. 1) displays the cascade of articles from the citation analysis and scoping review methods. Citation analysis A total of 331 records were included in the citation analysis. Most records were journal articles (n=275), followed by theses/dissertations (n=38), book chapters (n=8), pre-prints (n=7), internal reports (n=1), websites (n=1) and a letter to the editor (n=1). Full references of included and excluded records are included as Additional Files 5 and 6. Since publication in 2020, citations of the IRLM paper have grown each year in a nearly linear fashion (Fig. 2.). Annual citations of the IRLM article Results from the field analysis using SCImago revealed that, among journal articles, there were 151 unique journal titles represented. The most cited journal was Implementation Science Communications in the subject area of Medicine. The top five journals, broad subject areas and specific discipline subject categories are displayed in Table 3. The full list of journals, subject areas, subject categories, and countries are included as Additional Files 7 and 8. Table 3Field analysis: top 5 journals, SCImago subject areas and categoriesJournalCount (n) Implementation Science Communications28 BMJ Open17 Implementation Science12 Journal of Acquired Immune Deficiencies Syndrome12 Frontiers in Public Health9SCImago Subject AreaCount (n) Medicine103 Social Sciences25 Health Professions22 Nursing16 Psychology13SCImago Subject CategoryCount (n) Public health, Environmental and Occupational Health23 Medicine (miscellaneous)17 Health Policy15 Education12 Physical Therapy; Sports Therapy and Rehabilitation10Note: SCImago Subject areas represent broad umbrella fields, while subject categories represent specific disciplines within subject areas Field analysis: top 5 journals, SCImago subject areas and categories Note: SCImago Subject areas represent broad umbrella fields, while subject categories represent specific disciplines within subject areas Articles that included an IRLM Of the 331 included citations, 118 included a completed IRLM figure either in the main text or in supplemental material. Study setting. Most IRLMs were used in the context of non-primary health care (e.g., hospitals, n=68, 58%) followed by primary care settings (n=17, 14%), community settings (n=15, 13%), multi-sector (n=7, 6%), and educational contexts (n=4, 3%). There were also seven studies in which context was not specified. The continent/region and country where each study took place are displayed in Table 4. Most studies took place in a single country (n=79, 67%), with n=31, 26% across multiple countries. Table 4Continent and country where included studies took placeContinentn (%) North America57 (48%) Oceania17 (14%) Europe10 (9%) Asia9 (8%) South America5 (4%)Countryn (%)United States59 (50%)Australia15 (13%)India4 (3%)United Kingdom4 (3%)Nigeria3 (3%)Peru3 (3%)Canada2 (2%)Brazil2 (2%)China2 (2%)New Zealand2 (2%)Ethiopia2 (2%)Germany2 (2%)Kazakhstan1 (1%)Sweden1 (1%)Norway1 (1%)Malawi1 (1%)Singapore1 (1%)Malaysia1 (1%)Liberia1 (1%)Switzerland1 (1%)Ireland1 (1%)Note: N’s do not add up to 118 given that studies often took place across multiple countries or entire regions (e.g., across multiple low and middle-income countries; across several countries in Europe) Continent and country where included studies took place Note: N’s do not add up to 118 given that studies often took place across multiple countries or entire regions (e.g., across multiple low and middle-income countries; across several countries in Europe) Template used. Most IRLMs used one of two core templates from the original IRLM article: the standard template or version that includes the clinical/preventive/health promotion intervention box was most common (n=50, 42%), followed by the standard version (Fig. 1 in Smith et al. 2020) without the intervention included (n=39, 33%), and 25% (n=29) customized the template. No studies used other published IRLM variants, such as multi-context, strategy-comparison, or optimization/adaptive-trial versions. Use across phases of implementation. Most IRLMs were used during a single phase (n=90, 76%), with most users indicating that they used it only for planning the implementation study (n=61, 68%), followed by reporting (n=20, 22%), executing (n=5, 6%), and synthesizing (n=4, 4%). There were 28 (24%) studies that indicated that they used the IRLM across multiple phases: most common was planning and executing (n=8, 29%), followed by planning and reporting (n=7, 25%), reporting and synthesizing (n=3, 11%), and use across all four phases (n=3, 11%). The change in use over time is displayed in Fig. 3. Use of IRLM by phase over time. note: IRLM could be used across one or more phases COSMoS linking. Most studies did not complete CoSMOs linking in their IRLM (n=78, 66%), meaning that there were no explicit links between contextual factors/determinants, strategies, mechanisms, or outcomes. Of the 40 studies that included any linking, n = 23 (19%) included partial linking, and n = 17 (14%) completed comprehensive linking (i.e., links across all sections). Most partial linking occurred between mechanisms–outcomes (n=9, 39%) followed by contextual factors/determinants–strategies (n=7, 30%). Across the 40 IRLMs that included any linking, superscripts were the most common method (n=16, 33%) followed by color coding (n=15, 31%), numbering/lettering (that was not superscripted) (n=7, 15%), font or text effects (n=4, 8%), arrows/lines (n=3, 6%), text or box/sub-box alignment (n=2, 4%), and subscripts (n=1, 2%). There were nine studies that incorporated more than one of the above linking methods. IRLM data sources and community and implementation partner engagement. Data sources for populating the IRLM were varied. Most studies indicated that they used a single source for data collection across IRLM elements (n=55, 47%), followed by those who used multiple methods (n=54, 46%), and n = 9 (8%) did not specify their data source. The most common methods used were interviews with implementation and community partners (n=55), followed by surveys of various partners, including researchers (n=39), systematic, scoping, or literature reviews (n=27), focus groups (n=22), observational assessments (n=20), workshops or work groups (n=14), and advisory panels or teams of varying compositions of members (n=10). Forty-six (39%) articles did not include or describe engagement with the community in the creation of use of their IRLM, followed by Outreach (n=28, 24%), Consult (n=22, 19%), Collaborate (n=11, 9%), Involve (n=9, 8%), and Shared Leadership (n=2, 2%). Frameworks. Most studies cited CFIR as their determinant framework (n=87, 74%); EPIS (n=4, 3%), i-PARIHS [24] (n=3, 3%), and the Theoretical Domains Framework (TDF) [45] (n=2, 2%) were used more than once. The Capability, Opportunity, Motivation, Behavior (COM-B) model [46], the Health Equity Implementation Framework (HEIF) [47], and the Tailored Implementation for Chronic Diseases (TICD) Framework [48] were all used once (<1%). Nineteen articles (16%) did not indicate a specific determinant framework. Half of the studies (n=59) reporting implementation strategies did not cite or indicate a strategies taxonomy. Among those who did, the ERIC taxonomy was used most (n=55, 47%). The TDF was used in four studies (3%), with the Effective Practice and Organisation of Care (EPOCH) [49] and the School Implementation Strategies, Translating ERIC Resources (SISTER) [50] frameworks each cited once (<1%). When reporting implementation outcomes, most studies cited the IOF (n=47, 40%), followed by RE-AIM (n=40, 34%). Notably, 11 studies (9%) cited both the IOF and RE-AIM. EPIS and CFIR outcomes addendum [51] were each cited once (<1%), and 42 studies (36%) did not cite a framework. Most studies included mechanisms (n=94, 80%), although 24 (20%) studies did not. Notably, of the 94 that included mechanisms, n = 77 (82%) did not specify a framework or a source for their mechanisms. Seven studies (7%) cited the COM-B/TDF to identify hypothesized mechanisms and one cited EPIS (1%). Other stated sources of mechanisms included researcher theories/hypotheses (n=5, 5%), empirical evidence (n=2, 2%), and literature review (n=1, 1%). Evidence-based practice. There was a wide variety of evidence-based clinical/preventive/health promotion interventions being implemented across studies. These included non-cancer screening or diagnostic practices, including clinical practice guidelines, rehabilitation/physical therapy, HIV-related practices (HIV screening, pre-exposure prophylaxis implementation, retention in care), maternal/child health, chronic disease treatment and prevention (hypertension, chronic kidney disease, cardiovascular disease, diabetes mellitus, mental/behavioral health programs, health IT/informatics/EHR-based interventions, cancer screening/oncology, shared decision-making/patient education, digital health, and healthy weight programs. Articles that cited the 2020 article but did not include an IRLM There were 213 articles that cited but did not include an IRLM figure. Of these, 170 were journal articles, 8 book chapters, 30 theses/dissertations, 3 preprints, 1 website, and 1 internal report. Among articles following the standard structured research reporting format (i.e., introduction, methods, results, discussion), IRLM citations most often appeared in the discussion section (n=53), followed by the methods (n=52) and background/introduction (n=50), with fewer citations in the results (n=6). Articles could cite the IRLM in more than one section; the extent of such overlap was not quantified. The indicated purpose for the citation varied: many cited the IRLM as part of a general description or discussion of logic models and how they can be useful tools for planning or designing studies. Others cited the IRLM as a useful tool to, for example, (a) map determinants to identify implementation strategies, (b) provide theoretical grounding and developing theories of change, and (c) evaluate causal pathways. Several studies noted plans to use the IRLM in future work as part of their implementation protocol. Noted strengths, limitations and recommendations for improving the IRLM Reported strengths and limitations are included as Additional File 9. The IRLM was often recognized as a useful tool for providing a structured approach for aligning key elements of implementation, including context/determinants, strategies, mechanisms of action, and outcomes [52, 53]. It was described as a valuable reference point during evaluation for assessing and refining implementation efforts [54]. The IRLM’s standardized structure for documenting implementation components was frequently cited as enhancing generalizability, rigor, reproducibility, and supporting cross-study comparisons [55, 56]. Several articles also highlighted its role in guiding the integration of implementation science into intervention planning and evaluation [57]. Additionally, the IRLM was described as promoting conceptual clarity through visualization, helping to map complex processes and identify hypothesized causal pathways and mechanisms of action [58, 59]. Articles emphasized its utility in supporting the development, selection, and tailoring of implementation strategies, particularly when used alongside determinant frameworks to improve alignment between strategies and identified barriers [60]. The IRLM was also recognized for fostering meaningful partner engagement through visual tools that promote shared understanding and communication across stakeholder groups [61]. Although the IRLM is a valuable tool for structuring implementation efforts, several recommendations were identified to enhance its practical application. First, the theoretical complexity and scientific language of the IRLM may reduce its accessibility for practitioners and community-based implementers, particularly for those working in contexts that require multi-level, multi-pronged approaches [62–64]. Additional guidance may be needed to help users with less experience in implementation science understand the overarching logic that connects various implementation frameworks [65]. To address this, authors recommended the IRLM and its supporting materials include simplified language, applied examples, and implementation-specific tools to improve usability across diverse settings [66]. It was also recommended that implementers avoid introducing the IRLM to clinical or community partners prematurely and that the model should include the contextual information (i.e., determinants and potential strategies) prior to engaging partners [67]. This approach may improve partner comprehension and facilitate more meaningful participation, including the prioritization of strategies that align with key barriers. Although not always explicitly stated, there was an implied need for the IRLM to include temporal guidance on the sequencing or timing of implementation activities and IRLM components [68]. This addition could enhance and support more realistic planning and monitoring and increase the utility of the IRLM. Moreover, one article cautioned that, although useful for simplifying complex change processes, using logic models risks oversimplifying complex phenomena [69]. Lastly, although the IRLM was reported as useful for organizing concepts, it does not inherently provide a clear methodology for specifying the mechanisms linking determinants to outcomes or the mechanisms of action through which strategies operate [70]. Scoping review results for articles not citing or using the IRLM In total, 124 articles were identified from a pool of 4900 through the systematic scoping review procedures that neither cited nor used the IRLM (see Additional Files 10 and 11 for included and excluded references). Among these, most articles (n=77, 62%) included logic models that followed the ‘typical’ domain structure of a pipeline logic model [71] (i.e., inputs→activities→outputs→outcomes). The most commonly cited logic model frameworks were the Kellogg process evaluation logic model (n=25, 20%) [72], the intervention or implementation mapping logic model (n=14, 11%) [73, 74], the CDC Program Evaluation Framework (n=13, 11%) [75], Moore’s Process Evaluation Logic Model (n=9, 7%) [76], and McLaughlin’s logic model (n=7, 6%) [77]. Most articles in this sample created a logic model primarily for structuring an implementation or process evaluation (n=55, 44%). Approximately one-third were used to develop their intervention and plan implementation (n=45, 36%). Others used logic models to establish or display their theory of change (n=10, 8%), and there were 14 (11%) that did not specify their purpose for creating a logic model. Similarities and differences to IRLM. Implementation related elements varied across non-IRLM logic models. Frequently, authors reported implementation strategies mixed with intervention components in the Activities domain [78, 79]. Many others included contextual factors (e.g., barriers, existing resources) related to implementation in the Inputs domain [80]. Implementation outcomes were inconsistently reported but appeared in both the Outputs and Outcomes domains [81]. The most reported implementation outcomes were adoption (n=16, 13%), followed by reach (n=15,12%), fidelity (n=12, 10%), acceptability, appropriateness, or feasibility (n=6, 5%), and sustainability/sustainment (n= 4, 3%). Logic models included domains that were unique from the IRLM including: Performance Objectives, or measurable accomplishments (outputs) or changes that demonstrate progress toward achieving specific outputs or outcomes [82]; and Goals of the program and Assumptions [83]. Also, unique to the Intervention/Implementation Mapping approach was to include two logic models: one used to illustrate the problem and another to illustrate change [84]. Regarding similarities to the IRLM, n=18 (15%) included partial linking of elements; however, no studies had comprehensive linking across all elements in the logic model. Methods for partial linking primarily included arrows/lines (n=11, 61%) [79, 83], displaying their logic model in a table with separate rows aligning the elements (n=5, 28%) [85], and color-coding (n=2, 11%) [86]. Discussion This citation analysis and scoping review provide the first comprehensive synthesis of how the Implementation Research Logic Model (IRLM) has been applied in the published literature. Our findings indicate the IRLM’s influence and utility in aligning key elements of implementation efforts and supporting research/evaluation, and to a lesser extent implementation practice, while also highlighting important gaps and opportunities for refinement. As of December 31, 2024, the IRLM has been cited in over 330 publications. Since the publication of the seminal IRLM article on September 25, 2020, the number of citations has increased each year in a linear fashion. Among these, 118 articles included a populated IRLM figure, while 213 cited the IRLM article but did include an IRLM figure. The IRLM has been adopted in diverse fields (e.g., medicine, social sciences, psychology, toxicology and pharmaceutics, computer science, business, biochemistry, genetics and molecular biology, neuroscience, and engineering), and applied in 21 different countries, demonstrating the broad reach typical of a multidisciplinary field. Although primarily used for study planning, the IRLM has demonstrated utility for other purposes, such as being used as a roadmap for executing the study [87, 88], reporting study findings [15], and synthesizing findings across studies [15, 55, 89, 90]. Use of the IRLM across these different phases was relatively consistent over time. From 2021 to 2024, most studies (~70%) used the IRLM during the planning phase. Notably, more studies used the IRLM across multiple phases in 2024 than previous years. Explicit documentation of using the IRLM for execution or synthesis purposes is relatively rare in the current literature. Researchers may be applying it in these phases but often do not clearly describe this in publications. As a result, despite the IRLM’s value as a semi-structured reporting tool for implementation research, published examples of such applications remain limited. Use of the IRLM longitudinally throughout the study period and across related projects is also scarce [91]. Therefore, we encourage users to revisit, update, and use their IRLM to guide implementation efforts as they occur. With the IRLM in use for only five years, many projects may yet to have published on later-phase applications. The limited use in later phases may stem from insufficient time to reach implementation and sustainment phases in the identified studies or from limited awareness or guidance on applying the IRLM for execution, reporting, and synthesis purposes. We anticipate that use of the IRLM at multiple phases will increase over time, particularly for reporting and synthesizing findings across multiple studies. Based on these results, the IRLM appears to be fulfilling its intended role as a planning tool, offering a semi-structured approach to organizing implementation research elements. Moreover, future work should interrogate the feasibility of multi-phase use. Despite the IRLM’s emphasis on linking constructs across the Context, Strategies, Mechanisms, and Outcomes (CoSMOs) domains, most studies did not include comprehensive linking, and mechanisms were often underdeveloped or unsupported by theoretical frameworks. This suggests a need for clearer instructions or exemplars to support rigorous application. Although the 2020 IRLM article identified linking as a guiding principle, it provided limited instruction on this process, and comprehensive linked IRLMs did not appear until 2021. Alternatively, IRLM users may be hesitant to link mechanisms given uncertainty about the causal processes through which strategies achieve their effect. However, we encourage users to put forth hypothesized mechanistic inferences in their IRLMs that can be tested and adjusted iteratively. Although 80% of studies identified mechanisms in their IRLMs, sparse use of theoretical frameworks to select or justify them suggests a need for clearer guidance. A growing body of conceptual and empirical research into mechanisms of change related to dissemination and implementation outcomes [92–96] indicates an opportunity to provide explicit guidance on strategy-driven mechanisms of outcome change. Omitting or underspecifying mechanisms in an IRLM may signal that there is a fundamental gap in the field, including a priori identification of hypothesized relationships that can be tested to support theory testing and generalizable knowledge [28]. Moreover, there appears to be a lack of clarity around why a strategy was selected, what it is expected to accomplish, how it is expected to work, and how it may function when used in combination with other strategies—all critical questions for facilitating science and practice. Our findings point to a need for practical guidance on what mechanisms are, their theoretical basis, how they can be operationalized in an IRLM, their value for causal reasoning and strategy selection, how to identify them with community partners, and how to link them to other IRLM elements Engagement with community and implementation partners to develop an IRLM was inconsistently reported, with 39% reporting no engagement. Understanding implementation context requires engaging those who make decisions, deliver services, and benefit from those services [97]. In translational science, community engagement has been shown to include, and move beyond, understanding context to create actionable models of co-design and shared leadership [98]. For example, King and colleagues’ [99] scoping review demonstrates that engagement ranges from simple consultation to shared leadership, with deeper engagement linked to implementation and system-level research. Although some projects reported a high degree of community and implementation partner involvement [58, 60, 64, 100], practical guidance is needed for meaningful partners engagement in IRLM development and use. This can include co-creating models with partners to identify context, strategies, and mechanisms; using varied engagement processes to ensure decision-maker, doer, and beneficiary perspectives are heard and consolidated; and embedding equity and sustainability principles, such as those from Community-Partnered Participatory Research, to ensure actionable and sustainable pathways. Relatedly, several included studies drew on realist-informed methods [101, 102], which share IRLM’s focus on articulating how interventions work, for whom, and under what conditions [103]. Realist approaches use Context-Mechanism-Outcome configurations to develop and refine program theory, while IRLMs may provide a complementary structure for organizing these relationships into an actionable implementation model. However, explicit standards for developing IRLMs remain limited, particularly when co-developed with community partners. More explicit reporting of this process could also improve comparison across studies and support refinement of IRLM guidance over time. IRLMs also align with broader guidance for complex intervention research, including the updated MRC framework, which emphasizes programme theory, context, mechanisms, implementation, and adaptation across settings [104]. Positioning IRLMs within this wider framework may strengthen their utility for planning, implementing, and evaluating complex interventions. Comparison of IRLM with other logic models Our scoping review of studies that used alternative logic models for implementation research revealed that many lacked the conceptual clarity and integration of the central elements of implementation research (determinants, strategies, mechanisms, outcomes) that comprise the IRLM. Common limitations of alternative logic models included conflation of intervention(s) and implementation strategies, absence of specific linking of the elements in the logic model beyond the basic pipeline or other structure of the logic model, and inconsistent inclusion of implementation outcomes. Perhaps unsurprisingly, one of the impetuses for the development of the IRLM from the perspective of lead author, JDS, was his team’s experience trying to use traditional program evaluation logic models for implementation research—they were simply of limited utility without adaptation. The results of this review suggest a continued need to demonstrate the utility of the IRLM compared to alternative logic models. Limitations and future directions This citation analysis and systematic scoping review has several limitations. We restricted our analysis to English-language publications, which may have excluded relevant studies. We also did not assess the overall methodological quality of included studies, which is consistent with systematic review methodology, but we did assess the degree to which those that included an IRLM figure followed the recommendations of the 2020 IRLM article. Moreover, our search strategy in eight databases focused on articles indexed with both implementation and logic model in their title and/or abstract and subject terms, thus missing potential studies that applied an implementation-related logic model not indexed by the database. Finally, some studies may have used the IRLM without explicitly reporting or citing it, leading to potential underestimation of its use. To better understand the extent of IRLM use that is not evident in the published literature (e.g., use in a grant application), we are conducting a user survey and interviews [34]. Additional future research includes an exploration of the longitudinal use of the IRLM within and across projects with a focus on methods for updating the IRLM across phases and its use as a blueprint or protocol for implementation practice. Findings from this study will be used to expand and enhance resources that support the development and use of the IRLM, particularly guidance for the CoSMOs linking process, mechanisms section, and how to engage community partners [91]. Our citation analysis identified more than 100 completed IRLMs. Developing a curated repository of these examples spanning diverse contexts, disease areas, and types of implementation efforts (e.g., trials, pilot studies, quality improvement initiatives) could support the synthesis of generalizable knowledge on implementation processes. Our findings concerning the completeness of IRLMs lends support for creating a checklist and reporting guidelines for the IRLM when included in publications and proposals [34]. Conclusion The IRLM can help fill key conceptual and methodological gaps in implementation research by improving integration of key elements and enhancing rigor, transparency, and reproducibility. It has increasingly been adopted across diverse fields, settings, and countries, with applications including semi-structured planning, use as a blueprint for study execution, strengthening theorization, guiding the reporting of findings, evaluating the logic of implementation research and practice, and evidence synthesis. Indeed, the IRLM is intended to guide implementation research projects from conceptualization through ongoing execution and reporting. Its structured approach assists users in systematically and iteratively addressing the complexities inherent in implementation science. Addressing these gaps will be critical for enabling more consistent application of the IRLM across contexts, improving comparability across studies, and enhancing its value for both research and practice. We hope the findings of this citation analysis and scoping review contribute to the evolution of the IRLM such that it contributes to rigorous, transparent, and equitable implementation research. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary material 1 Acknowledgements We thank Tallie Casucci, Associate Librarian, J. Willard Marriott Library for her peer review of the search strategy. Abbreviations Implementation Research Logic Model Consolidated Framework for Implementation Research integrated-Promoting Action on Research Implementation in Health Services Reach, Effectiveness, Adoption, Implementation, Maintenance Exploration, Preparation, Implementation, Sustainment Context, Strategies, Mechanisms, Outcomes Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for scoping review reporting guidelines Implementation Outcomes Framework Capability, Opportunity, Motivation, Behavior Theoretical Domains Framework Electronic Healthcare Record Health Equity Implementation Framework Tailored Implementation for Chronic Diseases Effective Practice and Organisation of Care School Implementation Strategies, Translating ERIC Resources Author contributions JDS conceived of the study and acquired funding. JLM, OAS, PE, JH, DL, EAS, RS, RL-H, RM, MRR, and CAG contributed to the development of the methodology, including the study procedures and codebook development. MMM, JLM, and JDS developed the search strategy, and MMM conducted the search. JLM, EAS, OAS, RS, TVW, AJN, and CAG performed the data extraction and coding. JLM developed the data extraction templates, conducted data curation and drafted the final initial of the paper. All authors provided review and editing of the manuscript and approved the final manuscript. Funding This study was funded by a grant from the Ben B. and Iris M. Margolis Foundation to JDS. Additional support was provided by the University of Utah Translational Research: Implementation, Analysis, and Design (TRIAD) and Systematic Review Core (SR Core), with funding in part from the National Center for Advancing Translational Sciences of the National Institutes of Health under Award Number UM1TR004409. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Data availability Study materials and data are available by request to the corresponding author. Declarations Ethics approval and consent to participate Ethical approval for the scoping review and citation analysis is not applicable as only data from published literature will be used and no original data will be collected. Consent for publication Not applicable. Competing interests J.D. Smith is an Associate Editor of Implementation Science. J.D. Smith, Dennis Li, and Miriam Rafferty co-developed the IRLM. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary material 1 Data Availability Statement Study materials and data are available by request to the corresponding author.