An electronic database search of PubMed, SPORTDiscus, CINAHL, and Scopus yielded 1,013 titles and abstracts. After 478 duplicates were removed, the remaining 535 titles and abstracts were screened by two reviewers (JAH & JPW) for initial eligibility (Cohen’s κ = 0.87, 95% CI 0.76–0.98). After which, two reviewers read the full-text of 39 articles to determine eligibility for final inclusion. Of the 39 studies reviewed, 18 were excluded, leaving 21 studies. The reference lists of the remaining studies were inspected for relevant articles, of which two additional articles were found. Thus, a total of 23 studies were included in the final review (Fig. 1).
Of the 23 studies included in this review, eleven were of a retrospective design reporting ACL injury and twelve were cross-sectional studies assessing lower extremity biomechanics. General characteristics of these studies are presented in Tables 2 and 3.
Studies with ACL injury as the outcome
All participants in the retrospective injury risk studies were considered recreationally active. All studies except for one  were predominantly made up of participants under age 30, with two [31, 32] including participants up to age 40. Thus, the ages of the sample populations across studies were rather homogenous. Male participants across studies numbered 979, in contrast to 244 females. The time elapsed between ACL injury occurrence and data collection varied widely between studies, ranging from a few weeks up to five years. Four studies [16, 32, 42] did not report time elapsed between injury occurrence and data collection.
Quality assessment scores for each study are detailed in Table 4. The average Downs & Black Quality Index score in the retrospective injury risk studies was 9.5/14 (range = 8–13). The most commonly neglected items were blinding of the investigator and reporting of statistical power.
Detailed results of each injury risk study are presented in Table 5. Meta-analyses were conducted for the injury risk studies to assess aggregate between-group differences in femoral anteversion, passive internal rotation and external rotation between injured limbs and healthy control limbs. Fewer degrees of passive internal rotation was significantly associated with history of ACL injury based on the compiled data of 10 studies (MD -5.02°; 95% CI [-8.77°—-1.27°]; p = 0.01; n = 10) (Fig. 2). There was no significant effect between passive external rotation and ACL injury when examining the metadata of nine studies (MD -2.62°; 95% CI [-5.66° – 0.41°]; p = 0.09; n = 9) (Fig. 3). A meta-analysis of two studies, both in females, revealed no significant effect between femoral anteversion and ACL injury (MD -0.46°; 95% CI [-2.23° – 1.31°]; p = 0.61; n = 2) (Fig. 4).
Heterogeneity within the retrospective ACL injury studies was highly variable; I2 ranged from 20% for the influence of femoral anteversion on ACL injury to 90% and 92% for the influence of internal and external rotation on ACL injury, respectively.
Cross-sectional studies where lower extremity biomechanics was the outcome
With one exception (not reported) , all participants in the biomechanical studies were deemed physically active, and all were below the age of 30. Unlike the injury studies, which were composed mostly of males, the biomechanical studies included 452 females and 375 males. Of the twelve studies, eleven used a squat or jump landing variation to determine the influence of femoral anteversion and passive hip rotation on biomechanics. Of these eleven, three used a double-leg task [4, 40, 47] and eight opted for a single-leg task [6, 19,20,21, 26, 35, 38, 44]. A single study chose a side-cutting task . Seven of the studies measured valgus motion via 3D biomechanics [9, 19,20,21, 26, 38, 40]; the remaining five [4, 6, 35, 44, 47] measured 2D frontal plane knee motion.
The average Downs & Black score for the cross-sectional biomechanical studies was 10.5/12 (range = 8–12). Though the majority of these studies also omitted mention of statistical power, they generally scored higher on the Quality Index scale than did the retrospective injury risk studies. Because 2D biomechanics represent different movement patterns than 3D mechanics, they are reported separately below.
Biomechanical 2D results
Detailed results of each biomechanical study are presented in Table 6. Of the seven studies assessing 2D biomechanics, three studies [6, 35, 47] examined the association between femoral anteversion and passive internal rotation and 2D medial knee displacement (knee abduction) using single-leg squats [6, 35] or double-leg jump landings  (Figs. 5 and 6). Only one study  detected a small effect (d = 0.3; p = 0.04) during a single-leg squat between groups displaying high and low internal rotation ROM. Three studies [4, 35, 47] examined the association between passive external rotation and 2D medial knee displacement using an overhead squat , a single-leg squat , and a double-leg jump landing  (Fig. 6). All three studies observed greater passive external rotation in participants with medial knee displacement (collectively, participants with greater external rotation ROM displayed 4.77° greater two-dimensional frontal plane knee angle), though Bell (2008)  was the only study in which the external rotation difference was significant (MD 9.40°; CI [2.97 – 15.83]; p = 0.01).
Biomechanical 3D results
Five studies [19, 20, 26, 38, 40] examined the relationship between femoral anteversion and three-dimensional valgus measured as initial contact joint angle , peak joint angle [19, 20, 26, 40], peak joint moment [19, 20, 40] and/or joint excursions  using a single-leg jump landing , single-leg forward landing [19, 20], single-leg squat , or double-leg drop jump . Due to varying functional tasks, a meta-analysis was not possible. However, generally, greater femoral anteversion was associated with greater initial and peak hip flexion (d range = 0.2 – 1.6) , peak knee valgus angle (d range = 0.1 – 1.5) [19, 20, 26, 40], peak hip internal rotation moment (d range = 0.2–0.9) [19, 20, 40], and peak hip internal rotation angle (d range = 0.2 – 0.7) [19, 20, 40]. Four studies [9, 19,20,21] examined the relationship between hip ROM and 3D biomechanics, and generally revealed that greater passive internal rotation was predictive of greater frontal plane hip and knee excursions (d range = 0.7–1.2)  and frontal and transverse plane knee moments (d range = 0.1–1.0) [19, 20].
Evidence for a sex-specific effect
Where available, data were stratified by sex and reported in Table 7. Eight injury outcome studies reported sex-specific data (130 females, 670 males). From these studies, ROM differences between injured and uninjured females could be computed in two studies, and between injured and uninjured males in 6 studies. In females, there were no significant differences in passive internal (MD = 1.60°; p = 0.68) or external (MD = 1.10°; p = 0.33) rotation between ACL-injured and non-injured females [12, 49]. Similarly, in males, there were no significant differences in internal (MD = 3.75°; p = 0.09) or external (MD = 1.99°; p = 0.22) rotation between ACL-injured and non-injured males [3, 11, 16, 32, 42, 49]. Only one of the 8 studies included both males and females , and restricted internal and external rotation was associated with greater injury occurrence only in females.
Sex-specific data were available for four biomechanical studies, including 189 females and 220 males [6, 19, 20, 26, 44]. Only one within study biomechanical sex comparison was possible . During a jump landing, females displaying risky biomechanics (high frontal plane knee projection angle) trended toward greater passive internal rotation ROM than females displaying safer biomechanics, but this was not significant (p = 0.08, d = 0.7). This trend was not observed in the corresponding male cohort (p = 0.95, d = 0.0). Conversely, during a single-leg squat, there was no difference in internal rotation ROM between females displaying greater or fewer frontal plane knee projection angles, while males who had 6.4° less internal rotation ROM displayed more risky frontal plane knee movement (p = 0.04, d = 0.4) .
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.