A comprehensive search strategy was developed to identify studies reporting implementation-related outcomes of neuromuscular training programs for ACL injury prevention in female athletes. Four major electronic databases were systematically searched from January 1, 2014, to May 31, 2025, to capture contemporary implementation research aligned with modern implementation science frameworks: PubMed (MEDLINE), SPORTDiscus via EBSCOhost, Scopus, and Web of Science Core Collection. Database selection was informed by the multidisciplinary nature of implementation research, which spans sports medicine, implementation science, and behavioural intervention literature (32). PubMed provided comprehensive coverage of biomedical and health sciences literature with robust MeSH vocabulary for precise implementation science concepts. SPORTDiscus offered specialized coverage of sports medicine and exercise science literature, ensuring the capture of sport-specific prevention programs. Scopus and Web of Science provided broad multidisciplinary coverage and citation tracking capabilities, essential for identifying implementation research across diverse academic fields.

Search strategies were developed using a systematic approach combining Medical Subject Headings (MeSH) terms and free-text keywords to maximize sensitivity while maintaining specificity for implementation research (32–34). The search strategy incorporated four main concept groups connected using Boolean operators: (1) ACL injury prevention concepts, (2) neuromuscular training interventions, (3) implementation science terminology, and (4) female athlete populations. Primary search concepts included: ACL injury prevention: [“anterior cruciate ligament”(MeSH) OR “ACL injury” OR “knee injuries/prevention and control”(MeSH) OR “Athletic Injuries/prevention and control”(MeSH)]; Neuromuscular training: [“Exercise Therapy”(MeSH) OR “neuromuscular training” OR “injury prevention program” OR “FIFA 11+” OR “PEP program” OR “KLIP” OR “Sportsmetrics”]; Implementation science: (“implementation” OR “compliance” OR “adherence” OR “adoption” OR “real world” OR “effectiveness” OR “barriers” OR “facilitators” OR “program delivery” OR “sustainability” OR “dissemination”); and Study population: [“Female”(MeSH) OR “female athlete” OR “women” OR “girls” OR “female”]. The Boolean logic combined these concepts using AND operators between concept groups and OR operators within concept groups. Search strategies were adapted for each database platform to optimize retrieval while maintaining consistency across sources (see Supplementary Appendix 1). Database-specific syntax accommodated variations in field codes, proximity operators, and truncation symbols while preserving search logic integrity.

This review was restricted to controlled study designs specifically randomized controlled trials (RCTs), controlled clinical trials, crossover trials, and controlled before-after studies to enable systematic comparison of implementation approaches while maintaining methodological rigor (20, 35, 36). This design restriction was selected to balance the need for comparative evidence (implementation strategy A vs. B, or intervention vs. control) with recognition that implementation research often occurs in pragmatic settings where perfect randomization may not be feasible. Controlled designs provide a level of internal validity necessary for drawing meaningful conclusions about implementation strategy effectiveness while remaining more feasible than strict RCT-only requirements. Real-world effectiveness was captured through implementation outcomes (compliance rates, adherence measures, fidelity assessments, adoption patterns) and injury outcomes measured within these controlled trials conducted in naturalistic settings (schools, community sports clubs, elite training facilities), rather than highly controlled laboratory environments. This approach enabled assessment of program performance under real-world implementation conditions while maintaining comparison group rigor.

Studies were excluded if they focused solely on male athletes or mixed populations without separate female data analysis. Research involving only injured athletes or post-surgical rehabilitation populations was excluded, as were studies not reporting implementation-related outcomes or those using non-controlled study designs such as observational studies, case series, or case reports. While observational studies provide valuable real-world evidence, they were excluded due to the increased risk of selection bias and confounding when comparing implementation approaches without controlled allocation. Conference abstracts, editorials, commentaries, reviews, and studies reporting only biomechanical outcomes without implementation process data were excluded. Laboratory-based studies without real-world implementation context and investigations where implementation data could not be separated from efficacy results were also deemed ineligible.

The study selection process followed a systematic two-stage approach using the EPPI-Reviewer version 6 web platform to ensure transparency and reproducibility (37, 38). Retrieved citations from all database searches were initially imported into Zotero reference management software for comprehensive bibliographic organization and preliminary review. Following initial organization in Zotero, all retrieved records were uploaded to the EPPI-Reviewer version 6 web platform where automated duplicate removal was performed to eliminate redundant citations across databases. Implementation outcomes were operationally defined to guarantee uniform screening choices before title/abstract screening with a structured classification framework. The implementation outcomes were classified into three areas: (1) Quantitative implementation measures, such as compliance rates (percentage of prescribed sessions completed), adherence measures (consistency of program delivery over time), fidelity measures (degree to which programs were delivered as intended), adoption measures (proportion of eligible settings or individuals initiating program use), and reach measures (proportion of target population participating); (2) Implementation process measures, such as descriptions of delivery methods, implementer attributes, training methods, supervision models, and adaptation processes; and (3) Implementation determinants, such as barriers research had to report a minimum of one quantifiable implementation metric in domain 1, or present substantial qualitative data in domains 2 or 3 with enough detail to allow thematic analysis. To calibrate reviewer agreement, a screening decision tree was designed and pilot-tested on 20 randomly selected abstracts before full screening began.

The screening of the titles and abstracts was performed in the EPPI-Reviewer platform (37, 38) by two independent reviewers who evaluated each record in accordance with the established eligibility criteria. Before full screening, each reviewer screened a calibration set of 20 randomly selected abstracts to determine baseline agreement; initial agreement was 85% (17/20 concordant decisions) and discordant cases were discussed to clarify screening criteria interpretation. To achieve complete screening, inter-rater agreement was determined by the Cohen kappa coefficient (20, 39, 40). Title/abstract screening reached κ = 0.82 (95% CI: 0.71–0.93), indicating substantial agreement. Full-text screening had κ = 0.88 (95% CI: 0.76–1.00). During screening, disagreements were resolved by discussion and consensus (n = 12 at title/abstract stage; n = 5 at full-text stage), with third-party adjudication necessary in 2 cases where consensus was not achievable.

A comprehensive data extraction approach was developed specifically to support the narrative synthesis methodology and implementation science theoretical frameworks guiding this review. The data extraction strategy was designed to systematically capture information aligned with both CFIR domains and RE-AIM dimensions, ensuring a comprehensive collection of implementation-related data essential for quality assessment, SWiM-guided analysis, and results synthesis.

The data extraction framework integrated three complementary components. First, structured data extraction captured standardized study characteristics, intervention details, and quantitative implementation outcomes necessary for SWiM reporting guidelines compliance. Second, theory-guided data extraction systematically organized barriers and facilitators according to CFIR’s five domains: intervention characteristics, outer setting, inner setting, individual characteristics, and implementation process. Third, RE-AIM structured extraction captured reach, effectiveness, adoption, implementation, and maintenance outcomes across individual and organizational levels to support comprehensive implementation evaluation.

Studies were eligible if they reported biomechanical measures (e.g., knee abduction moment, ground reaction forces, frontal plane projection angle) alongside implementation metrics. Biomechanical outcomes were treated as secondary indicators of neuromuscular adaptation that may predict injury risk reduction, but were not considered primary real-world effectiveness outcomes. The hierarchy of effectiveness evidence prioritized: (1) exposure-adjusted injury incidence (injuries per 1,000 athlete-exposures or training hours), (2) crude injury incidence, (3) biomechanical surrogate outcomes. This hierarchy reflects the principle that real-world effectiveness is ultimately measured by injury prevention impact in naturalistic settings, while biomechanical adaptations provide supporting mechanistic evidence.

Study identification and design characteristics were extracted including author details, publication year, country, study design, setting type, duration, and follow-up periods. Population characteristics encompassed sample sizes, demographics, sport types, competition levels, and geographic contexts essential for understanding implementation contexts. Intervention details captured program names, components, session characteristics, delivery methods, and implementation strategies, with particular attention to adaptation and modification processes that influence real-world implementation success.

Critical implementation process data included detailed extraction of implementation strategies using established taxonomies, delivery personnel and training approaches, fidelity measurement methods, and adaptation rationale and processes. Implementation outcomes were systematically captured including compliance and adherence rates with clear operational definitions, adoption rates at organizational and individual levels, reach and penetration measures, and sustainability indicators over time. Used together, CFIR and RE-AIM frameworks identify metrics for evaluating implementation success and modifiable factors that explain and enhance implementation outcomes.

Barriers and facilitators were systematically extracted and coded according to CFIR’s five domains to ensure comprehensive theoretical coverage. Intervention characteristics domain captured program complexity, adaptability, trialability, and relative advantage perceptions affecting implementation success. Outer setting factors included policy environment, external pressures, patient needs and resources, and competitive pressures influencing implementation contexts. Inner setting characteristics encompassed organizational culture, leadership engagement, available resources, access to knowledge and information, and readiness for implementation. Individual characteristics of implementers included knowledge and beliefs about the intervention, self-efficacy, individual stage of change, and other personal attributes affecting implementation success. Implementation process factors captured planning approaches, engaging strategies, executing methods, and reflecting and evaluating activities throughout implementation.

Representative quotes and specific examples supporting each CFIR construct were extracted to enable rich thematic analysis and theory development. Quantitative data on barrier and facilitator frequencies were captured when available to support both quantitative and qualitative synthesis approaches. This systematic approach enabled a comprehensive understanding of implementation determinants while maintaining theoretical coherence essential for narrative synthesis using implementation science frameworks.

Implementation outcomes were systematically organized according to RE-AIM dimensions to support structured narrative synthesis. Reach data captured population coverage, representativeness, and participation rates across diverse implementation contexts. Effectiveness outcomes included real-world injury prevention results, biomechanical improvements, and comparative effectiveness data contrasting controlled versus naturalistic implementation settings. Adoption measures encompassed organizational uptake, individual implementer engagement, and setting-level adoption rates across different contexts. Implementation indicators captured program fidelity, adaptation patterns, delivery consistency, and cost considerations affecting sustainable implementation. Maintenance outcomes included sustainability measures, long-term continuation data, and institutionalization indicators essential for understanding implementation durability.

This comprehensive data extraction approach ensured systematic capture of all information necessary for quality assessment using implementation-appropriate criteria, narrative synthesis following SWiM guidelines with transparent reporting of grouping strategies and synthesis methods, and results presentation organized around implementation science frameworks that provide practical guidance for stakeholders seeking to implement ACL injury prevention programs in real-world settings.

Data synthesis followed the Synthesis Without Meta-analysis (SWiM) reporting guidelines, providing a structured methodology for transparent narrative synthesis of heterogeneous implementation evidence. The SWiM framework’s nine reporting items guided synthesis planning, execution, and presentation to ensure reproducible and transparent analysis (22, 23).

Systematic heterogeneity investigation employed multiple analytical approaches. Clinical heterogeneity was assessed through a comparison of population characteristics, intervention components, and outcome measurement approaches. Methodological heterogeneity was evaluated by examining study designs, implementation measurement methods, and follow-up periods. Statistical heterogeneity was assessed where quantitative synthesis was attempted, though meta-analysis was deemed inappropriate due to substantial diversity. Contextual heterogeneity was systematically explored through subgroup analysis by setting type, implementation strategy, population characteristics, and geographic region. Sources of heterogeneity were documented and explored rather than statistically controlled, recognizing heterogeneity as informative for understanding implementation context-dependency rather than methodological limitation.

Laboratory-based implementations demonstrated the highest degree of control and supervision, enabling precise intervention delivery and comprehensive outcome measurement. Ford et al. (47) employed sophisticated technology integration with real-time biofeedback systems, achieving 93.3% completion rates while demonstrating the feasibility of complex intervention delivery in research settings. Shams et al. (53) compared technology-enhanced approaches using mirror feedback versus Mulligan taping techniques, with both methods achieving substantial completion rates exceeding 94%. Hopper et al. utilized professional strength and conditioning supervision to achieve perfect completion rates alongside significant biomechanical improvements in youth netball athletes (49). However, Bulow et al. (45) demonstrated that even perfect implementation fidelity cannot overcome inappropriate intervention selection, achieving 100% completion with zero measurable benefits. These controlled environments established the upper bounds of implementation quality while highlighting the critical importance of intervention appropriateness for target populations.

Educational institution implementations leveraged existing organizational structures to facilitate systematic program delivery across diverse athletic populations. Foss et al. (48) successfully implemented school-based programs across nine institutions using athletic trainer oversight, achieving greater than 95% team compliance and demonstrating significant injury reductions across multiple sports. Sugimoto et al. (54) revealed the implementation bottleneck in educational settings, with coaches achieving only 52.5% compliance despite 87.8% athlete readiness, identifying professional support as a critical implementation success factor. Zebis et al. (57) optimized the educational model through comprehensive teacher training and bi-weekly research team monitoring, achieving 89% session completion alongside measurable neuromuscular adaptations. Yarsiasat et al. (56) demonstrated cultural adaptability within structured educational environments, successfully implementing culturally adapted programs in Thai sports schools with perfect completion rates. Educational institutions emerged as optimal early implementation venues due to their structured environments, dedicated personnel, and established training schedules that facilitate systematic program integration.

Community and club-based studies addressed real-world implementation challenges while examining diverse stakeholder engagement approaches across naturalistic sports environments. McKay et al. (51) conducted the first systematic comparison of implementation strategies, demonstrating that comprehensive support including physiotherapist assistance achieved superior adherence rates compared to workshop-only or educational-only approaches. Ferri-Caruana et al. (46) achieved exceptional compliance through direct research supervision within Spanish club systems, with no participants missing more than one session during the eight-week intervention period. Suits et al. (55) provided revolutionary evidence comparing Knowledge-to-Action framework implementation against passive information delivery, establishing that systematic stakeholder engagement significantly improved implementation rates and demonstrating direct dose–response relationships between implementation frequency and injury prevention effectiveness. These community implementations revealed the complexity of real-world program delivery while identifying stakeholder engagement and professional support as critical success factors for achieving meaningful implementation outcomes in diverse organizational contexts.

Organizational and longitudinal implementation research provided unique insights into program sustainability, institutional adoption, and long-term implementation maintenance across complex multi-level systems. Lindblom et al. (50) conducted the only comprehensive long-term follow-up using RE-AIM framework assessment, revealing high program reach (91–99% familiarity) but concerning fidelity challenges with only 23–26% of users implementing programs without modifications three years post-trial Schoeb et al. (52) demonstrated that self-directed implementation could achieve meaningful injury reductions despite suboptimal adherence rates, with participants averaging 80% of recommended frequency while still experiencing significant reductions in both traumatic and overuse injuries. This study uniquely acknowledged and designed for the “voltage drop” phenomenon inherent in real-world implementation, establishing that perfect adherence may not be necessary for meaningful injury prevention outcomes. These longitudinal perspectives highlighted the tension between implementation fidelity and adaptation flexibility while demonstrating the potential for sustained impact even with imperfect implementation quality.

Implementation strategy diversity across the included studies revealed a spectrum of approaches ranging from minimal educational interventions to comprehensive multi-component support systems with varying degrees of effectiveness. Educational-only strategies consistently demonstrated limited effectiveness, with Sugimoto et al. (54) revealing poor coach compliance and Schoeb et al. (52) showing reduced but still meaningful outcomes through self-directed implementation. Workshop-based approaches with ongoing supervision achieved consistently high implementation quality across diverse settings, while technology-enhanced interventions demonstrated potential for improved precision but raised questions about scalability and resource requirements. The emergence of systematic stakeholder engagement approaches, particularly the Knowledge-to-Action framework demonstrated by (55), represented a methodological advancement that achieved superior implementation outcomes through barrier identification and adaptive program delivery. This implementation strategy evolution from passive information dissemination to active stakeholder engagement reflected the maturation of implementation science approaches in sports injury prevention research.

All twelve randomized controlled trials demonstrated appropriate randomization procedures with adequate sequence generation and allocation concealment (45–57). Baseline group comparability was maintained across studies, with most reporting appropriate demographic and baseline characteristic distributions between intervention and control groups.

Completion rates varied across studies, ranging from 89% [(46) 26/29 participants] to 100% (45, 49, 56) (see Table 3). Several studies experienced attrition challenges in community-based settings, with Ford et al. (47) reporting 6.7% dropout, Shams et al. (53) reporting 5.9% dropout (3/51 participants for attendance/injury reasons), and Ferri-Caruana et al. (46) reporting 10.3% attrition. McKay (51) reported moderate organizational-level attrition in the community soccer club context. Complete outcome data were available for all participants who completed their assigned interventions in controlled laboratory and educational institution studies.

Blinding of outcome assessors was limited across studies due to the behavioural nature of neuromuscular training interventions, representing an inherent challenge in implementation research rather than a methodological flaw. This limitation was noted in the MMAT assessment for all RCTs (Table 1 footnote). Participant adherence to assigned interventions varied substantially. High adherence (≥89%) was achieved in studies with direct professional supervision: Zebis et al. (57) - 89% session completion; Ferri-Caruana et al. (46)- 89.7% completion with 100% attendance (≤1 absence); Ford et al. (47) - 93.3% completion; Shams et al. (53) - 94.1% completion; Bulow et al. (45), Hopper et al. (49), and Yarsiasat et al. (56) - 100% completion. School-based implementations showed variable adherence: Foss et al. (48) achieved >95% team compliance with athletic trainer oversight, while Sugimoto et al. (54) reported coach compliance of 52.5 ± 11.7% and athlete compliance of 87.8 ± 10.6%. Community-based implementations demonstrated strategy-dependent adherence patterns: McKay et al. (51) reported team adherence ranging from 73.5% (control) to 85.6% (comprehensive support); Suits et al. (55) reported significantly higher implementation in KTA intervention versus handout groups (p < 0.001). Self-directed implementation achieved moderate adherence: Schoeb et al. (52) reported 80% of recommended frequency (0.8 ± 0.6/week).

Studies demonstrated variable quality in implementation-specific reporting. Comprehensive barrier and facilitator identification was provided by McKay et al. (51) and Suits et al. (55), while other studies offered limited implementation process details. The descriptive follow-up study (50) provided appropriate sampling strategies and measurements for long-term implementation assessment (RE-AIM framework), though response rates were moderate (57% among trial coaches, 36% among current coaches), potentially introducing nonresponse bias concerns.

Based on MMAT criteria fulfilment, eight studies were classified as High quality (meeting ≥4 of 5 applicable criteria): Bulow et al. (45), Ford et al. (47), Foss et al. (48), Hopper et al. (49), Shams et al. (53), Suits et al. (55), Yarsiasat et al. (56), and Zebis et al. (57). Five studies were classified as Moderate quality (meeting 3 of 5 criteria): Ferri-Caruana et al. (46), Lindblom et al. (50), McKay et al. (51), and Schoeb et al. (52). No studies were classified as Low quality. Domain-level MMAT results are presented in Table 1 to enable transparent quality appraisal.

Implementation compliance rates demonstrated substantial variation across different implementation strategies, ranging from 52.5 to 100% completion rates (Table 4). Educational institution-based implementations with research supervision achieved the highest and most consistent compliance rates, with (49, 56) both achieving 100% completion rates, while (57) reported 89% session completion. Structured educational environments demonstrated high implementation fidelity in these studies. Technology-enhanced laboratory implementations demonstrated consistently high compliance rates, with (47) achieving 93.3% completion despite the complexity of biofeedback systems, and (53) reporting 94.1% completion rates. However, these approaches required intensive resource investment and specialized personnel, potentially limiting their scalability beyond research settings.

The most striking finding emerged from school-based implementations with varying professional support levels. Foss et al. (48) achieved greater than 95% team compliance through athletic trainer oversight, Sugimoto et al. (54) revealed a critical implementation paradox: athletes demonstrated high readiness to comply (87.8% compliance) when programs were delivered, but coaches achieved only 52.5% implementation compliance with minimal ongoing support. This disparity indicates lower coach implementation compared to athlete participation when programs were delivered.

Community-based implementations showed strategy-dependent compliance patterns. McKay et al. (51) demonstrated a clear implementation hierarchy: comprehensive support with physiotherapist assistance achieved 85.6% team adherence, workshop-based education reached 81.3%, while control groups receiving online materials only achieved 73.5% adherence. Suits et al. (55) further validated stakeholder engagement approaches, with Knowledge-to-Action intervention yielding significantly higher implementation rates than educational handout distribution alone (p < 0.001).

Self-directed implementations revealed that motivated populations could achieve reasonable compliance despite minimal supervision. Schoeb et al. (52) reported 80% adherence to recommended frequency among elite youth skiers, while (50) found 74% adoption rates among coaches three years post-trial, though only 23–26% maintained programs without modifications. Research laboratory-controlled implementations achieved high completion rates but demonstrated that perfect implementation cannot overcome inappropriate intervention selection. While Ferri-Caruana et al. (46) achieved 89.7% completion with significant biomechanical improvements, Bulow et al. (45) achieved 100% completion but found no measurable benefits, highlighting the critical importance of intervention appropriateness for target populations.

Analysis of implementation strategies revealed distinct effectiveness hierarchies. Professional supervision combined with educational institution settings consistently produced the highest compliance rates and demonstrated measurable outcomes. Technology-enhanced approaches achieved excellent compliance but faced scalability limitations due to resource requirements (Table 4). Implementation outcomes showed higher compliance rates in studies with ongoing professional oversight (athletic trainers, physiotherapists, qualified coaches) compared to educational-only approaches. This pattern is held across diverse settings, from school districts (48) to sports colleges (57) to research laboratories (49). The relationship between implementation strategy complexity and compliance varied across contexts. Simple self-directed approaches could achieve reasonable implementation rates with highly motivated populations (52), while complex stakeholder engagement processes substantially improved implementation in community settings (55).

Measurement approaches varied considerably across studies, from objective attendance tracking to self-report questionnaires. Studies with direct professional supervision (49, 56, 57) achieved the most reliable compliance assessment, while self-directed implementations relied primarily on participant self-report (50, 52). Implementation outcomes varied by strategy type, population characteristics, and organizational capacity. Educational institutions, elite athlete populations, and settings with existing professional support infrastructure achieved optimal implementation outcomes, while community-based amateur populations required more intensive stakeholder engagement approaches to achieve meaningful implementation levels.

To assess whether methodological quality influenced synthesis conclusions, analyses were repeated restricting inclusion to high-quality studies only [n = 8; MMAT rating ‘High’: Bulow et al. (45), Ford et al. (47), Foss et al. (48), Hopper et al. (49), Shams et al. (53), Suits et al. (55), Yarsiasat et al. (56), and Zebis et al. (57)] compared with moderate-quality studies [n = 5: Ferri-Caruana et al. (46), Lindblom et al. (50), McKay et al. (51), Schoeb et al. (52), and Sugimoto et al. (54)]. Limiting to high-quality studies did not change the direction or consistency of primary findings (Table 9). High-quality studies showed consistently positive implementation compliance outcomes (8/8, 100%) compared to moderate-quality studies (4/5, 80%), with the one mixed finding explained by the coach-athlete compliance difference reported by (54). The implementation strategy hierarchy was maintained at the levels of quality, where professional supervision models had a higher compliance rate (range: 89–100) than educational-only models irrespective of the study quality classification.

The positive directions in injury prevention outcomes were observed in high-quality studies reporting injury outcomes included (48, 55–57) and moderate-quality studies included (52, 54). The finding that implementation ≥2 sessions per week was associated with 85% ACL injury risk reduction was from one high-quality study (55): HR = 0.15, 95% CI = 0.03–0.73, p = 0.019 (Table 10). Likewise, high-quality evidence by Foss et al. (48) (injury rate ratio: 5.34 vs. 8.54/1000 athlete-exposures, p < 0.001) and Zebis et al. (57) (1 vs. 7 lower extremity injuries, p = 0.04) supported the dose–response relationship between implementation quality and effectiveness. The coach implementation pattern, where athlete compliance (87.8%) exceeded coach implementation (52.5%), was observed in one moderate-quality study (54) (Table 9). Similar patterns of higher implementation with professional support were observed across both high-quality and moderate-quality studies, suggesting this pattern was not limited to one quality stratum.

The strength of effect direction categories was also assessed with alternative thresholds. The main analysis grouped positive directions by statistical significance (p < 0.05) or by clinically meaningful differences (at least 10 percent absolute difference in compliance, at least 20 percent decrease in relative risk of injury). Stricter thresholds (p < 0.01 or ≥15% absolute difference) reduced positive classifications to 77% (10/13 studies) but maintained the pattern of no negative effect directions and similar strategy-level patterns (Tables 6, 11). A higher liberal threshold (p < 0.10 or 5% or greater absolute difference) raised positive classifications to 85% (11/13) without changing main conclusions. No negative effect directions were observed under any threshold specification. This pattern was consistent across threshold levels, though interpretation should account for potential publication bias toward positive findings. Sensitivity analyses showed that main synthesis patterns were consistent across study quality strata and effect direction threshold specifications. Key patterns including associations between implementation quality and outcomes, professional support and higher compliance rates, and the implementation frequency-effectiveness relationship observed in one study (55), remained consistent across analytical approaches.

Program complexity represented the most frequently cited barrier across eight studies (45, 47, 48, 51–55), with coaches consistently reporting that 15–30 min session requirements and multi-component learning demands created significant implementation burdens. Conversely, program simplicity and clear progression protocols served as powerful facilitators in eleven studies (46–53, 55–57), with structured approaches like the evidence-based neuromuscular training program achieving 89% compliance through systematic progression and regular monitoring. The complexity-simplicity paradox demonstrates that while comprehensive programs may be more effective, implementation success depends critically on presenting complex interventions through simplified, manageable delivery protocols.

Intervention adaptability emerged as a key implementation determinant, with rigid protocols creating barriers in four studies (45, 51, 53, 54) while flexible delivery options facilitated implementation in nine studies (46–49, 51–53, 55, 56). The self-directed ISPAInt program exemplified successful adaptability by allowing athletes to perform exercises “anytime and anywhere within 20 minutes,” achieving 80% of the recommended frequency despite minimal supervision (52). Sport-specific adaptations proved particularly valuable, with the PEP program’s successful modification for Sepak Takraw demonstrating that cultural and sport-specific customization enhances acceptability without compromising effectiveness (56). This finding underscores that implementation frameworks must balance standardization with contextual adaptation to maximize real-world adoption.

External setting factors revealed substantial organizational influence on implementation success, with supportive policies facilitating implementation in eight studies (47–54, 57) while competitive pressures created barriers in six studies (48, 51, 52, 54, 55, 57). National-level policy support provided crucial implementation infrastructure, as demonstrated by the Swedish Football Association’s formal policies enabling widespread program reach (91–99% familiarity) three years post-trial (50). However, the absence of district-level implementation policies created significant gaps, with 0 % of district associations maintaining formal guidelines despite national support. The tension between performance priorities and injury prevention time allocation consistently emerged as a barrier, particularly in competitive environments where coaches faced pressure to maximize skill development over prevention activities (52, 54).

Inner setting characteristics highlighted the critical importance of organizational resources and leadership engagement for implementation success. Available resources served as facilitators in twelve of thirteen studies (45–53, 55–57), with existing athletic trainer infrastructure proving particularly valuable for school-based implementations. University research partnerships provided essential equipment and expertise, enabling high-fidelity implementation in controlled settings (47, 49). Leadership engagement emerged as highly variable, with strong institutional support facilitating implementation in ten studies (45, 47–52, 54, 55, 57) while absent organizational commitment created barriers in seven studies (45, 50–55). The stark contrast between the >95% team compliance achieved with athletic trainer oversight compared to 52.5% coach compliance with minimal support illustrates the transformative impact of organizational leadership and professional support systems.

Individual characteristics revealed a consistent pattern where knowledge gaps and implementation burdens created coach-level barriers while athlete receptivity remained consistently high. Coach implementation challenges appeared in eight studies (45, 48, 50–55), with knowledge gaps about injury prevention mechanisms and implementation skill deficits limiting program delivery. Paradoxically, athletes demonstrated high compliance (87.8%) when programs were delivered, indicating that implementation failure occurred primarily at the provider rather than recipient level (54). Professional qualifications and comprehensive training served as powerful facilitators, with accredited strength and conditioning coaches achieving 100% completion rates through technical competency and safety prioritization (49). This finding suggests that implementation strategies must focus primarily on enabling and supporting coaches rather than motivating athletes.

Implementation process factors showed associations between ongoing engagement and implementation success. Minimal ongoing support created barriers in seven studies (45, 50–55), while comprehensive training and regular monitoring facilitated implementation in eleven studies (45–51, 53, 55–57). The Knowledge-to-Action framework’s success in achieving significantly higher implementation rates compared to educational handouts exemplifies the value of systematic stakeholder engagement and barrier identification (55). Professional oversight emerged as essential for execution quality, with systematic monitoring enabling successful implementation across twelve studies while its absence led to implementation failures (45–53, 55–57). Patterns across the implementation process domain showed that sustained, systematic support was associated with higher implementation outcomes compared to one-time training or passive information delivery.

Real-world effectiveness outcomes varied by implementation quality and context. Among the three studies reporting ACL-specific data, injury reductions ranged from 65 to 85% when programs were implemented with adequate frequency and supervision (Table 12). Suits et al. (55) found that implementation frequency ≥2 sessions per week was associated with 85% ACL injury risk reduction (HR = 0.15, 95% CI = 0.03–0.73, p = 0.019), providing direct dose–response evidence among included studies for implementation frequency and injury prevention effectiveness in amateur soccer settings. These findings indicate that meaningful injury reductions were achieved when implementation thresholds were met in the contexts studied.

Implementation fidelity emerged as a critical determinant of real-world effectiveness, with substantial variation observed between controlled and naturalistic settings across the included studies. Seven studies (45–47, 49, 53, 56, 57) achieved high implementation fidelity (>90% compliance) through intensive supervision models, while real-world implementations with minimal supervision demonstrated considerably lower adherence rates. Compliance rates varied substantially: Sugimoto et al. (5) reported 52.5% coach compliance with minimal support, while (48) achieved >95% team compliance with athletic trainer oversight (Table 12). Schoeb et al. (52) reported meaningful injury reductions (33.5% traumatic, 30.1% overuse) with 80% of recommended frequency in elite youth skiers, indicating that injury prevention benefits occurred despite suboptimal adherence in this motivated population.

The relationship between implementation strategies and real-world effectiveness revealed significant disparities in outcome achievement across different support models and settings. Educational institution-based implementations showed high effectiveness outcomes when measured, Zebis et al. (57) demonstrated significant reductions in lower extremity injuries (1 versus 7 injuries, p = 0.04) and Yarsiasat et al. (56) showing substantial decreases in knee injury incidence (11.5% versus 34.6%) when programs were delivered with professional supervision and organizational support (Table 12). Community-based implementation outcomes varied by strategy type, McKay et al. (51) demonstrated adherence rates ranging from 73.5 to 85.6% depending on support intensity, while Suits et al. (55) confirmed that active stakeholder engagement significantly improved implementation rates compared to passive information delivery. Self-directed implementations, exemplified by Schoeb et al. (52), achieved moderate effectiveness despite minimal supervision, but this success appeared contingent on elite athlete populations with high intrinsic motivation and organizational support structures.

Perfect implementation fidelity did not ensure intervention effectiveness across all studies. Bulow et al. (45) achieved 100% implementation compliance but demonstrated no measurable benefits in dynamic balance outcomes, while (49) achieved similar fidelity (100% completion) with significant biomechanical improvements (knee abduction reduction p = 0.004, g = 1.43) (Table 12). These contrasting findings indicate that implementation fidelity alone does not determine effectiveness, and intervention appropriateness for the target population appears relevant. Across included studies, effectiveness outcomes varied by implementation quality, intervention characteristics, and population factors.

One study provided long-term sustainability evidence. Lindblom et al. (50) reported that 68–82% of original trial coaches continued program use three years post-trial, though only 23–26% maintained programs without modifications (Table 12). Program use continued in many coaches at three-year follow-up, though modifications to original programs were common. Real-world effectiveness outcomes were observed across multiple settings when adequate implementation support was provided, though long-term fidelity and sustainability data were limited (n = 1 study with ≥3-year follow-up).

Implementation patterns across multiple organizational levels showed differential compliance between implementers and participants in some settings. The findings showed that athletes achieved 87.8% compliance when programs were delivered, compared to only 52.5% coach compliance (54), indicating that implementation failure occurs at the provider rather than recipient level. This coach implementation paradox was further supported by evidence that high levels of coach intent to integrate prevention programs did not translate to effective implementation (16), suggesting that motivation alone is insufficient for successful adoption. The identification of this bottleneck aligns with the updated Consolidated Framework for Implementation Research (24), which emphasizes the critical role of implementation climate and individual characteristics in determining intervention success. These findings challenge traditional educational approaches to implementation and highlight the need for comprehensive coach support systems that address capacity, motivation, and opportunity barriers simultaneously.

Implementation outcomes demonstrated patterns suggesting differential effectiveness across strategy types, with stakeholder-engaged professional support models associated with higher compliance rates than passive educational approaches. Studies employing comprehensive implementation support, including professional supervision (48), knowledge-to-action frameworks (55), and educational institution integration (56, 57), demonstrated superior implementation outcomes compared to educational-only strategies. This hierarchy aligns with implementation science evidence suggesting that active implementation strategies are more effective than passive dissemination approaches (24, 25). The success of professional support models validates the National Athletic Trainers’ Association position statement recommendations for multi-stakeholder involvement in prevention program implementation (59). However, the resource intensity of effective implementation strategies creates tension between implementation effectiveness and scalability, necessitating careful consideration of cost-effectiveness in implementation planning.

The synthesis revealed contextual variation in implementation outcomes, with educational institutions demonstrating higher implementation fidelity compared to community and elite sports settings in the studies reviewed. Studies conducted in school-based settings (48) and sports colleges (57) achieved exceptional compliance rates and injury reduction outcomes, suggesting that structured educational environments provide optimal implementation contexts. This pattern may reflect the inherent characteristics of educational settings, including dedicated time allocation, professional supervision, and organizational commitment to holistic athlete development. Conversely, community-based implementations showed greater variability, with success dependent on comprehensive support provision and stakeholder engagement intensity. Elite sports settings presented unique challenges, with high athlete motivation potentially offset by competing performance priorities and time constraints. These contextual patterns align with implementation science literature emphasizing the importance of inner-setting characteristics in determining intervention success (24, 25, 28, 31).

Emerging evidence from this review suggested that technology-enhanced implementation approaches may improve biomechanical outcomes, though with scalability considerations. Our results demonstrated that real-time biofeedback significantly improved biomechanical outcomes (47), while simple augmentation strategies such as mirror feedback and taping could enhance traditional training approaches (53). These findings align with neurocognitive and ecological motor learning frameworks, which emphasize the importance of feedback modalities in optimizing skill acquisition and retention (61). However, technology-enhanced approaches often require specialized equipment and training, creating potential barriers to widespread implementation. The integration of motor learning principles into implementation strategies represents an evolving area where sports science and implementation science intersect, offering opportunities for innovation in delivery methods while requiring careful consideration of implementation feasibility.

Patterns across included studies suggested relationships between implementation quality and effectiveness outcomes. The ≥2 sessions per week finding represents a minimum dose threshold rather than evidence of a graded dose–response relationship, as few studies tested multiple frequency levels systematically. True dose–response relationships, where effectiveness increases proportionally across multiple dose levels, require investigation in future research (45, 57). High-fidelity implementation resulted in significant neuromuscular adaptations associated with ACL injury risk reduction (57), while perfect implementation of inappropriate interventions yielded no benefits (45). These findings establish implementation fidelity as a critical mediator in the relationship between intervention exposure and effectiveness outcomes. The dose–response evidence provides practitioners with specific targets for implementation success while highlighting the importance of matching intervention appropriateness to target populations. This relationship validates meta-analytic evidence suggesting that compliance represents a key moderator of intervention effectiveness (5, 16).

The review provided strong evidence supporting stakeholder engagement and co-creation approaches in implementation strategy development. Studies employing systematic barrier identification and adaptation strategies (55) achieved superior implementation outcomes compared to standardized approaches. This finding validates the Knowledge-to-Action framework’s emphasis on tailored implementation strategies based on local context and stakeholder needs. The success of stakeholder engagement approaches aligns with implementation science literature demonstrating that participatory implementation strategies improve adoption and sustainability (24, 31, 58). However, the resource intensity of comprehensive stakeholder engagement creates practical challenges for widespread implementation, necessitating the development of scalable engagement approaches that maintain effectiveness while reducing resource requirements.

Effect direction classification thresholds (p < 0.05 for statistical significance; ≥10% absolute difference for compliance/adherence outcomes; ≥20% relative risk reduction for injury outcomes) were defined a priori as pragmatic benchmarks representing clinically meaningful differences in implementation science contexts. The 10% absolute compliance difference threshold was selected based on reviewer judgment that this magnitude of difference is likely to translate into meaningful population-level impact when scaled across programs, though no universal consensus exists for implementation outcome thresholds. The 20% relative injury risk reduction threshold reflects a clinically important effect size in injury prevention research. These thresholds represent pragmatic decisions to enable systematic evidence synthesis, and sensitivity analyses examined whether conclusions changed under alternative specifications (Section 3.7).

Publication bias should be considered when interpreting these findings. Studies demonstrating successful implementation and positive injury prevention outcomes may be more likely to achieve publication than those reporting implementation failures or null results, potentially overestimating the perceived effectiveness of neuromuscular training programs. The absence of negative effect directions across all 13 included studies, while encouraging regarding program safety, could partly reflect selective publication rather than universal effectiveness. Conventional statistical assessments of publication bias (funnel plot asymmetry, Egger regression tests) require pooled effect sizes and could not be applied to this narrative synthesis of heterogeneous implementation evidence.

Selective outcome reporting may also influence findings, as only 38% of studies (5/13) reported adoption and reach outcomes, 46% (6/13) reported injury prevention outcomes, and 23% (3/13) reported ACL-specific injury outcomes. This pattern could indicate selective reporting of positive outcomes rather than comprehensive outcome measurement. Pre-registration of protocols was not uniformly reported across included studies, restricting systematic assessment of reporting completeness. Future implementation research should prioritize prospective registration with full outcome specification to enhance transparency and minimize selective reporting concerns.

The absence of reported negative effect directions should not be interpreted as definitive evidence of safety or absence of harm. Adverse events beyond participant withdrawals were not systematically assessed or reported across studies. No studies explicitly reported systematic adverse event monitoring protocols, limiting confidence in safety conclusions. The finding that no studies reported increased injury risk with neuromuscular training should be stated as “no included studies reported increased injury risk” rather than concluding that programs are universally safe, acknowledging underreporting limitations and the general inadequacy of included studies to detect rare adverse events.