Discussion

This study is the first meta-analysis to systematically synthesize video-based evidence on anterior cruciate ligament (ACL) injury situations across sports. We included 39 studies comprising 1551 video-documented ACL injuries, with football, basketball, and COF (American Football, Australian Rules football, rugby) contributing the majority of cases, alongside netball, handball, and judo. By pooling and comparing data across these diverse contexts, the present work provides the most comprehensive quantitative overview to date of video-analyzed movement patterns and game situations associated with ACL injuries. Understanding these patterns is of critical importance, as they reveal the situational contexts in which ACL loading and subsequent rupture most often occur, thereby offering essential guidance for sport-specific surveillance and the design of targeted prevention strategies [8, 60].

Contact Mechanisms

In our video-based meta-analysis of predominantly team sports, noncontact mechanisms accounted for 44% of ACL injuries, indirect contact for 36%, and direct contact for 20%. Moderator analyses indicated that the distribution of injury mechanisms varied by sport. Across team sports, direct‑contact injuries were most frequent in COF (32.1% [95% CI 26.4–37.9], n = 5), intermediate in football (22.5% [15.6–30.3], n = 11), and uncommon in basketball (6.1% [1.7–12.8], n = 8). By contrast, indirect‑contact injuries comprised nearly half of basketball ACL injuries (48.9% [36.7–61.2], n = 7) but a smaller fraction in football (35.0% [29.6–40.7], n = 10) and COF (26.8% [15.4–40.0], n = 5); noncontact proportions did not differ significantly between sports. Smaller samples in other sports showed specific extremes (e.g., judo skewed toward direct contact, with 64.7%, whereas netball presented low direct contact involvement). These sport-specific patterns align with the most recent systematic review, which reported that basketball is dominated by indirect contact mechanisms (~ 60%), football is primarily noncontact (~ 47%), and American Football/rugby have larger contact contributions [13]. However, this review synthesized video data together with athlete/medical staff reports and did not meta-analyze proportions, which may explain the differences in magnitude [13].

Movement Patterns

In our team-sports-dominated sample of studies, noncontact ACL injuries most frequently occurred during cutting (53.8%), followed by pressing/tackling (50.2%), decelerating (38.9%), and landing (30.1%). Indirect contact mechanisms were most frequently associated with being tackled (56.1%), followed by situations with ball possession (44.8%) and landing (30.0%). For direct contact injuries, pooled action-specific estimates were available for being tackled (23.9%) and pressing/tackling (24.2%). Moderator analyses indicated sport-specific differences for tackling-related patterns. Pressing/tackling was more frequent in soccer than in COF when pooling all injuries and when restricting analyses to noncontact cases. For indirect contact injuries, the proportion of cases classified as being tackled differed between sports (higher in COF than in soccer), although this finding should be interpreted cautiously owing to the limited number of eligible studies in some sports. No significant moderator effects were observed for direct contact injury patterns, but action-specific data were sparse. Notably, tackling-related situations showed sport-dependent patterns. In soccer, pressing/tackling accounted for a substantial proportion of injuries, including in noncontact cases, which may reflect high-demand defensive actions (e.g., pressing or attempted tackles) that do not necessarily involve knee-directed contact. In contrast, indirect contact injuries in COF were frequently classified as being tackled, suggesting that perturbations and body–contact dynamics during tackles may be a more dominant feature in these codes. Given the small number of eligible studies in some sports and the conceptual overlap between “pressing” and “tackling,” these findings should be interpreted with caution and motivate more granular, consensus-driven reporting (e.g., separating pressing from tackling and specifying whether contact occurred, where, and when, relative to the injury).

Overall, these findings confirm that cutting, pressing/tackling, deceleration, and landing are the key high-risk movement contexts when no contact with opponents is involved, whereas being tackled and pressing/tackling are the most common contact-related scenarios. Our findings corroborate those of Sunderg et al. [13], who, based on both video- and non-video sources, also identified cutting and landing as key mechanisms underlying noncontact ACL injuries in team sports. However, by incorporating more recent, larger-sample studies [27, 38], we identified pressing/tackling and being tackled as frequently occurring movement patterns, particularly in football and COF, which might have previously not been recognized.

Movement Velocity

Across the pooled team-sport sample, ACL injuries occurred most frequently at high horizontal velocity (53.8%), with additional contributions from low horizontal velocity (34.8%) and no horizontal motion (6.2%). The majority of injuries were sustained with no vertical component (63.5%), while fewer occurred at low (19.0%) or high (11.5%) vertical velocity. These distributions suggest that both rapid horizontal movements and vertical loading scenarios, particularly landings, are important contexts for ACL injury. Previous work [13] has emphasized the relevance of high-horizontal speed play in several sports; our pooled estimates extend this by demonstrating that noncontact injuries can also arise in stationary or low-horizontal-velocity situations, potentially due to high vertical ground reaction force impulses. However, the assumption that noncontact, low-horizontal-velocity ACL injuries were due to high vertical velocity remains speculative at the moment, because the included studies rarely reported the results in a way that would allow this assumption to be directly checked. Future studies should report results in a way (ideally on a per-injury basis) that allows for a joint interpretation of movement speed and contact mechanism, as well as situational patterns to clarify under which precise conditions contact, indirect, and noncontact ACL injuries occur.

Game Situations

When considering the broader game context, 67.5% of ACL injuries occurred during ball possession and 55.4% during offensive play. Noncontact injuries showed a predominance in offensive phases (58.5%), whereas indirect contact injuries had a comparatively larger share in defensive phases (42.7%). Moderator analyses of injury timing indicated that time-of-match distributions differed by sport: Injuries clustered in the first quarter in football (37.5%; n = 2) and netball (37.8%; n = 1) were closer to a uniform distribution in contact‑oriented football (COF; 25.9%, 95% CI 15.4–41.8%; n = 3), and were less frequent early in basketball (16.0%; n = 2). Because a “first quarter” represents roughly 25% of total match time, the football and netball figures suggest over‑representation of early phase injuries, whereas basketball shows under‑representation. These differences may reflect sport-specific opening-phase demands, warm-up strategies, or distinct match structures (e.g., substitution patterns, stoppages, and set-play density in basketball and COF).

Practical Implications

Injury prevention strategies in sports should be mechanism‑specific, with movement scenarios and game situations refining where the primary leverage lies (rules/policy, coaching and technique, physical preparation, or equipment) [8]. In the sections that follow, we aim to derive the potentially most effective prevention strategies for each sport from our findings. We derived prevention strategies for football, basketball, and COF, as the findings for these sports were likely most robust, given that they are based on the largest number of included studies in our analysis.

For football, noncontact mechanisms were the predominant driver of ACL injuries, with the highest noncontact frequencies observed during cutting (53.8%), pressing/tackling actions (50.2%), decelerating (38.9%), and landing (30.1%). Accordingly, preventive work in football should incorporate scenario-based training that refines and enhances cutting [61], braking [62], and single- and double-leg landing techniques [63, 64]. Given the energy absorption requirements of the identified high-risk movement patterns, these drills should be coupled with the development of sufficient eccentric strength capacities in the muscle groups involved in these tasks [65]. Because injuries frequently occurred during ball possession (67.5%), and noncontact cases were more common during offensive play (58.5%), these drills should routinely include the ball and neurocognitive decision-making (i.e., executive function) demands to promote automatic control under realistic game conditions [66].

Indirect‑contact injuries were also common in football (35.0%), most often while being tackled (56.1%), with additional contributions from landing (single‑ and double‑leg landing combined 30.0%) and single‑leg landing (25.7%). To address these situations, we recommend integrating perturbation elements into football-specific cutting and landing tasks (e.g., controlled pushes or band pulls while changing direction or receiving the ball), with explicit emphasis on trunk control, single-leg stability, and rapid recovery of alignment when destabilized. Embedding these perturbations within ball‑in‑possession drills (also to emphasize an external focus of attention) reflects the contexts in which indirect‑contact injuries frequently arise. In an ACL injury context, perturbation training has been mainly applied during RTS settings [67]. However, for preventive approaches, perturbations may need to be adjusted in intensity to induce the desired prevention effects.

Although direct‑contact injuries represent a smaller share in football (22.5%), they warrant targeted countermeasures. From a policy and coaching perspective, this includes stricter sanctioning and consistent officiating for knee-directed challenges (e.g., late or low tackles into a planted limb or knee-to-knee collisions), alongside technical coaching on safer entry and exit from duels (reducing planted-leg exposure and preparing to yield or redirect contact). Finally, moderator analyses also indicated an early match clustering in football, with 37.5% of injuries occurring in the first 25% of the game. This counterintuitive pattern (unlikely to be fatigue-driven) suggests re-examining pre-kick-off and on-field activation, e.g., keeping the warm-up closer to kick-off and adding brief, game-speed cut/deceleration/landing sequences. Future work should confirm this with time-stamped exposure data in relation to descriptions of warm-up strategies.

In basketball, direct‑contact mechanisms were uncommon (6.1%), whereas indirect‑contact injuries were the largest share across sports (48.9%). High horizontal speed was relatively infrequent (19.6%), and landing featured prominently among movement patterns (32.5%), underscoring the potential impact of more vertical‑impulse related contexts. Because indirect‑contact events predominate, incorporating perturbation elements into jump‑landing and cutting drills (e.g., controlled shoulder bumps, elastic‑band pulls, aerial/body contact at or just after landing), with explicit training of trunk control, single‑leg stability, and rapid recovery of alignment when destabilized, could be a valuable strategy for ACL injury prevention in basketball. Coupling these drills with ball-in-hand scenarios to mirror the contexts in which indirect perturbations more generally arise may further increase their effectiveness.

As in other sports, the prevention of noncontact injuries should emphasize jump- and land-specific technique (single and double leg), coached at game speed and under visual-attention demands, with eccentric strength work to support controlled decelerations on take-off and landing. Given the overall prominence of ball possession and offensive phases in the pooled data, we recommend integrating ball handling and passing tasks to foster an external focus of attention while maintaining safe knee alignment under realistic decision‑making. Given the low frequency of direct‑contact ACL injuries in basketball, rule‑based levers are likely limited. No early match clustering was evident in basketball (16.0% in the first quarter), providing no immediate signal to modify pre-tip-off warm-up timing.

In COF, direct‑contact mechanisms accounted for the highest share among the team sports analyzed (32.1%). Accordingly, prevention might prioritize tackle-related exposures: refine tackle technique (minimize knee-directed impact to a planted limb), strengthen officiating and sanctioning for knee contact, and consider equipment strategies (e.g., task-adaptive knee bracing) where feasible. Precedents in these codes (e.g., recent tackle‑law initiatives in rugby to reduce concussion risk [68]) illustrate how targeted rule adjustments might shift behavior; however, changes must balance player safety with the fabric of the game, and their effects should be monitored for actual prevention effects and unintended consequences [69]. Indirect and noncontact injuries also contributed meaningfully to ACL injuries in COF. Consequently, the movement- and context-focused measures outlined for football and basketball (cutting/braking and landing technique at high speed, coupled with adequate eccentric strength and decision-making demands) are directly applicable here and should be incorporated alongside contact-focused measures.

Evidence for netball derives from only two studies, so estimates carry substantial uncertainty. Within this limited sample, landing (single and double leg) accounted for 81.2% of ACL injuries, and noncontact deceleration was comparatively frequent (58.0%), indicating that prevention should prioritize jump and landing technique (single and double leg) and high‑quality deceleration into landings at match speed, supported by adequate eccentric braking capacity. In one study that reported horizontal velocity, injuries were often preceded by high horizontal velocity (81.3%), suggesting that drills rehearsing rapid approach-deceleration sequences may be beneficial [54]. Given that indirect‑contact events can still arise in congested spaces, it is reasonable also to include perturbations during landing and ball‑handling tasks to train rapid recovery of alignment. Finally, moderator analyses indicated early match clustering (37.8% in the first quarter). While the causal drivers are unclear, aligning warm-up/field activation with brief, game-speed landing/deceleration exposures closer to the start may be prudent.

Limitations

Some methodological aspects warrant consideration. The included studies showed significant variability in the analyzed sports and methodologies. This finding highlights the need for standardized definitions and reporting methods, which would benefit future research. We believe that the 18 items of the recently developed QA-SIVAS scale [11] could be a guideline for these methodological improvements.

The quality assessment of the included studies in this review, using the QA-SIVAS scale, revealed a broad spectrum of methodological rigor, with scores ranging from 33 to 83%. Two studies were classified as high quality (81–100%), while the majority fell within the good (71–80%) and moderate (60–70%) categories. However, ten studies were deemed to have low quality (< 60%). Certain aspects of the studies were consistently well reported, including the objectives, methodological descriptions, main results, injury context, and discussions within the current evidence base. However, there was substantial room for improvement in areas such as representative sample recruitment, reporting information about the sample, video source/quality, medical report information, and descriptions of the raters’ backgrounds. These findings underscore the importance of applying quality criteria more stringently in future research to ensure the reliability and validity of video analysis studies in sports injury research. This includes the necessity for consistent reporting standards of injury mechanisms (i.e., contact, indirect contact, noncontact injuries) and the use of standardized definitions and terms for movement situations and patterns, at least for specific sports. A further limitation concerns the granularity with which defensive actions were described in the primary studies. For meta-analytic pooling across sports, we used harmonized categories (e.g., “pressing/tackling” and “being tackled”); however, “tackling” can encompass mechanistically distinct scenarios that likely differ between codes (e.g., noncontact pressing or attempted tackles in soccer versus body-contact tackles in COF, and injuries sustained by the tackler, such as when the lower limb becomes pinned, versus injuries to the ball carrier). Although we revisited the original papers to determine whether a more fine-grained classification (e.g., pressing versus tackling and contact versus noncontact tackling) was feasible, most studies did not provide injury-level descriptors with sufficient detail (player role, tackle type, and whether/where/when contact occurred relative to the injury). Consequently, some misclassification across categories cannot be excluded, and pooled estimates for these defensive actions should be interpreted with caution. Future video-analysis studies would benefit from consensus-driven reporting that explicitly separates pressing from tackling and documents, on a per-injury basis, the presence, location, and timing of contact and whether the injured athlete was the tackler or the player being tackled, to improve comparability and synthesis across sports. More broadly, this illustrates that comparability across sports will depend on more explicit operational definitions and reporting of whether contact occurred, where it occurred, and when it occurred relative to the injury event, alongside clearer separation of defensive actions. To improve clarity, future studies should, e.g., differentiate contact on the basis of (1) force magnitude and direction, distinguishing minor interactions from those significantly altering postural stability or knee-joint loading; (2) timing, as contact occurring several strides before the injury may differ biomechanically from contact immediately preceding or during the injurious movement; and (3) biomechanical consequences, such as contact-induced balance shifts or increased valgus loading. Standardizing these classifications through predefined criteria or biomechanical thresholds will enhance comparability across studies and sports, ultimately refining injury prevention strategies. Owing to the different classifications and descriptions of movements and situations, meta-analytical approaches and comparative studies are challenging to implement. Initial approaches to sport-specific categorization of injury-inciting events have recently been developed for football and netball [70, 71].

Another limitation of this study was that our moderator analysis for noncontact ACL injuries in different sport-specific situations was hindered by the small number of eligible studies, which prevented us from drawing definitive conclusions. Hopefully, this limitation can be mitigated by future high-quality studies that differentially assess the situational circumstances leading to ACL injuries in sports other than football.

The overall lower number of injury cases in female athletes and studies only reporting on one sex is a limitation that needs to be addressed in the current sex data gap in sports medicine research [72]. As injury rates in team sports [73] and certain risk factors [74] differ between sexes, we call for the collection of data for both female and male athletes in the future to provide sex-specific injury risk reduction programs [75]. Furthermore, video analysis studies were almost entirely conducted on samples of highly trained athletes (Table 1). Consequently, the findings of our analyses may not necessarily be applicable to recreational athletes.