Trunk and Lower Extremity Biomechanics in Female Athletes With and Without a Concussion History

This study is part of an ongoing clinical research program in conjunction with athletics to biomechanically screen incoming student-athletes before they began their collegiate athletic careers. All participants provided written informed consent, and the study protocol was approved by the Institutional Review Board of the University of Georgia (PROJECT00000876).

Given that some of our data (concussion history, demographics) came from the Concussion Assessment, Research and Education Consortium of the National Collegiate Athletic Association and US Department of Defense, athletes had to consent both to the clinical research project study and the Concussion Assessment, Research and Education Consortium to be included. Athletes were included if they were varsity level and medically cleared for competition. We did not exclude athletes based on injury history or orthopaedic surgery history, as is common in stricter biomechanics research.²⁰  The only exclusion criterion was not consenting to participate in the research. Two participants reported a concussion history but no date (month or year) and were removed from all analyses. The final sample (n = 64) included 26 athletes with and 38 athletes without a concussion history.

Concussion history was self-reported and collected using the Michigan TBI History form.²⁶  Athletes self-reported concussion history as diagnosed or undiagnosed. Three participants only reported the year of their most recent concussion. The month of June was used for analyses of these participants. Injury and orthopaedic surgery histories were also self-reported by athletes. Injury was defined as a traumatic injury to a ligament, such as a tear, meniscus damage, or fracture to the lower extremity. Minor injuries such as sprains, in which athletes immediately returned to activity, and upper extremity injuries were not considered in the analysis. Surgery was defined as any kind of surgery to the lower extremity. Dominant limb was defined as the preferred limb for kicking a soccer ball for distance.

A 30-cm-high box was placed a distance half the athletes’ height away from 2 force plates (Bertec). Athletes were instructed to stand on the box and get into an athletic position (ie, knees and hips slightly flexed) when the researcher (E.S. or J.O.) said, “Get set.” A green light was illuminated randomly by a member of the research team (E.S. or J.O.) within 5 seconds of athletes assuming the “get set” position.²⁷  The illumination of the light was the cue for the athletes to jump forward, not up, toward the 2 force plates. Athletes were instructed to react as quickly as possible to the light. For the jump-landing task, they were required to land with both feet simultaneously and fully on the 2 force plates (1 foot per plate) and, immediately on landing, to jump straight up as high as possible. For the cutting task, athletes were required to land on 1 limb, immediately and explosively cut to the opposite direction at a 45° angle (eg, left limb landing, right-cut direction), and run through a set of cones. Both tasks were considered unsuccessful and were repeated if participants jumped up for height from the box rather than straight forward, they stepped rather than jumped off the box, or their feet did not land fully on the force plate(s). All athletes were given at least 1 practice trial for each task and could practice until they were comfortable with the task. Three successful jump landings and cuts off both limbs (6 total cuts, 3 in each direction) were collected.

Retroreflective markers were placed bilaterally on the acromioclavicular joint, iliac crest, greater trochanter, anterior-superior iliac spine, medial and lateral femoral epicondyles, medial and lateral malleoli, calcaneus, fifth metatarsal, and second metatarsal. A marker was also placed on the sternal notch. A cluster of noncollinear markers was placed on the posterior-superior iliac spine and sacral body and placed bilaterally on the thigh, shank, and foot. Static calibration markers were removed before jumping and cutting. Marker position data were sampled at 240 Hz with an 8-camera (Qualisys MIQUIS; Qualisys Systems) motion capture system. Force plate data were sampled at 1200 Hz.

Raw marker position data and force plate data were exported to Visual 3D (C-Motion Inc) software for analysis. All data were processed with a fourth-order, low-pass Butterworth filter at 10 Hz. The anterior- and posterior-superior iliac spines defined the pelvis, hip joint centers were estimated using the method of Bell et al, knee-joint centers were estimated using the midpoint between the medial and lateral femoral epicondyles, and ankle-joint centers were estimated using the midpoint between the medial and lateral malleoli.²⁸  Euler-Cardan angles (Y-X-Z rotation sequence) were used to calculate the hip, knee, and ankle angles. Hip motions were defined as the thigh relative to the pelvis, knee motions were defined as the shank relative to the thigh, and ankle motions were defined as the foot relative to the shank. Trunk motion was defined relative to the global axis system (Y-X-Z rotation). Rotation about the y-, x-, and z-axes was flexion-extension, abduction-adduction (trunk lateral bending), and internal-external rotation, respectively.

For the jump landing, trunk lateral bending was calculated as peak bending toward either limb. In other words, the largest displacement toward (away from straight up-down [0°]) either the dominant or nondominant limb was used in the analysis. During cutting, the largest value toward the cutting limb (ie, planted limb) for peak trunk lateral-bending angle was used in the analysis.

Joint moments were calculated with standard inverse dynamics and resolved in the proximal segment coordinate system. Vertical loading rate was calculated as the first derivative of the vertical ground reaction force slope. Joint moments were normalized to the product of body weight and height (BW × HT). Vertical ground reaction force and loading rate were normalized to body weight (×BW).

Peak kinetics (vertical ground reaction force, vertical loading rate, external knee-abduction moment, and external knee-flexion moment) and kinematics (ankle-dorsiflexion angle, knee-flexion angle, knee-abduction angle, hip-flexion angle, trunk-flexion angle, and trunk lateral-bending angle) were obtained during the eccentric portion of the task (ie, initial ground contact to peak knee-flexion angle). Initial ground contact was defined as the point at which vertical ground reaction force exceeded 10 N. Kinetic and kinematic variables were averaged across all 3 trials for each task. Limbs were analyzed separately rather than averaged together, as previous researchers have shown asymmetries in movement after concussion and different results for each limb compared with controls.^18,19 

Demographic information such as age, height, mass, and body mass index was compared using a series of independent t tests. The proportions of lower extremity injuries or surgeries (yes or no), injured or surgical limb (none, both, dominant, or nondominant), limb dominance (left or right), and sport (soccer, gymnastics, or volleyball) were compared between groups (concussion history or no concussion history) using the Fisher exact test (Table 1).

Separate 2 (group) × 2 (limb) between-within analyses of covariance (ANCOVAs) with Bonferroni adjustments were used to compare jump-landing and cutting biomechanics. The between factors were athletes with and without a concussion history. The within factors were the dominant and nondominant limb. The covariates were mean centered time since the most recent concussion and the limb that had a history of MSKI (coded as none, both, dominant, or nondominant).¹²  Mean centering was calculated by subtracting each athlete’s time since their most recent concussion from the group mean time since their most recent concussion, then assigning the control group a value of 0.¹²  Trunk flexion and lateral bending during jump landing were compared using a 1-way ANCOVA as both limbs contacted the ground simultaneously. Partial η² (η_p²) effect sizes were interpreted as small (≤0.05), medium (0.06–0.13), and large (≥0.14) and reported with 90% CIs. We chose a 90% CI because η_p²  values cannot be <0 because F tests are 1-sided; therefore, using a 95% CI may include zero when the P value is significant, causing confusion.^29–31  Covariate adjusted means with 95% CIs and mean differences with 95% CIs are reported. An α level of .05 was established a priori. We used IBM SPSS Statistics for Windows (version 28.0; IBM Corp) for analyses.

We performed a sensitivity analysis to confirm that the 3 athletes who reported only the year of their concussion were not influencing the results. We removed the 3 athletes, recalculated the mean centered time since their most recent concussion, and rechecked the ANCOVA results. Our sensitivity analysis results were the same as the full analysis results.

None of the demographic data were different between groups (Table 1). Athletes with a concussion history reported the following injuries and surgeries: 1 lateral meniscus tear, 1 Achilles tendon tear, 1 knee arthroscopic surgery, and 3 anterior cruciate ligament tears. Athletes without a concussion history reported the following injuries and surgeries: 1 medial collateral ligament tear, 1 shank fracture, 2 ankle fractures, 6 anterior cruciate ligament tears, 1 meniscus tear, 1 femur fracture, and 2 unspecified ankle surgeries. A Fisher exact test revealed no differences in the proportion of injury or surgery type between groups (P = .67). Raw (non–covariate-adjusted), time-normalized waveforms are shown in Figures 1 and 2 for all biomechanics outcomes.

We observed a group-by-limb interaction for peak knee-abduction angle (F_1,60 = 4.6, P = .04, η_p²  = 0.071 [90% CI = 0.002, 0.185]). The peak knee-abduction angle in the nondominant limb was larger in athletes with than those without a concussion history (mean difference = 4.35°; 95% CI = 0.63°, 8.08°; P = .02, η_p²  = 0.083 [90% CI = 0.006, 0.201]). Athletes with a concussion history displayed larger peak knee-abduction angles in their nondominant than dominant limb (mean difference = 4.30°; 95% CI = 1.08°, 7.51°; P = .01, η_p² = 0.107 [90% CI = 0.014, 0.229]). No other interactions were found (P range = .09–.89; Table 2; Supplemental Table 1, available online at https://dx.doi.org/10.4085/1062-6050-0259.23.S1).

No group main effects were observed for any jump-landing outcomes (P range = .14–.77; Table 2; Supplemental Table 1).

We observed a limb main effect for peak vertical ground reaction force (F_1,60 = 4.6, P = .041, η_p² = 0.068 [90% CI = 0.002, 0.185]), peak knee-abduction angle (F_1,60 = 4.2, P = .046, η_p² = 0.065 [90% CI = 0.001, 0.177]), and peak external knee-flexion moment (F_1,60 = 7.3, P = .009, η_p² = 0.108 [90% CI = 0.015, 0.231]). However, post hoc analysis showed no difference between limbs for peak vertical ground reaction force (mean difference = 0.09 ×BW; 95% CI = −0.01, 0.19 ×BW; P = .08, η_p² = 0.052 [90% CI = 0.000, 0.159]). The peak knee-abduction angle (mean difference = 1.75°; 95% CI = 0.16°, 3.34°; P = .03, η_p² = 0.075 [90% CI = 0.003, 0.190]) and peak external knee-flexion moment (mean difference = 0.02 BW × HT; 95% CI = 0.01, 0.02 BW × HT; P < .001, η_p² = 0.382 [90% CI = 0.215, 0.497]) were greater in the nondominant than the dominant limb. We observed no other limb main effects (P range = .05–.78; Table 2; Supplemental Table 1).

No group-by-limb interactions were found (P range = .07–.88; Table 3; Supplemental Table 2).

We observed a group main effect for peak trunk lateral-bending angle (F_1,60 = 8.7, P = .005, η_p² = 0.126 [90% CI = 0.023, 0.252]). Athletes with a concussion history displayed a smaller peak trunk lateral-bending angle toward the planted limb compared with athletes without a concussion history (mean difference = 4.75°; 95% CI = 1.52°, 7.97°; P = .005, η_p² = 0.126 [90% CI = 0.023, 0.252]). No other group main effects were found (P range = .13–.80; Table 3; Supplemental Table 2).

A limb main effect was found for peak external knee-abduction moment (F_1,60 = 15.3, P < .001, η_p² = 0.206 [90% CI = 0.067, 0.332]) and peak trunk lateral-bending angle (F_1,60 = 7.0, P = .01, η_p² = 0.104 [90% CI = 0.014, 0.226]). Peak trunk lateral bending toward the planted, nondominant limb was greater during cuts with the nondominant than dominant limb (mean difference = 2.24°; 95% CI = 0.97°, 3.532°; P < .001; η_p² = 0.171 [90% CI = 0.047, 0.300]). Peak external knee-abduction moment was smaller during cuts with the nondominant than dominant limb (mean difference = 0.006°; 95% CI = 0.004°, 0.008°; P < .001; η_p² = 0.315 [90% CI = 0.151, 0.434]). We observed no other limb main effects (P range = .10–.88; Table 3; Supplemental Table 2).

Athletes with a concussion history displayed larger nondominant-limb knee-abduction angles during jump landing and less trunk lateral bending during cutting tasks compared with athletes without a concussion history. Athletes with a concussion history also displayed larger knee-abduction angles during a jump landing with their nondominant than dominant limb. These results were not observed among athletes without a concussion history.

The larger knee-abduction angle during jump landing for athletes with a concussion history may be due to deficits in neuromuscular control. Our results agree with those reported in previous literature, and larger knee-abduction angles are predictive of anterior cruciate ligament tears.^10,21  The causes of larger knee-abduction angle can vary. Researchers have shown that females have larger knee-abduction angles than males during landing, which may be a potential reason for the increased risk of MSKI postconcussion.^6,32  Athletes may have different jump-training–specific backgrounds, which have been shown to decrease the knee-abduction angle during landing.³³  In addition, a large predictor of knee-abduction angle is hip-abductor muscle strength and activation.³⁴  Neural drive is a large component of muscle strength, and individuals postconcussion have shown lower levels of electromyographic activity compared with controls during hand movements.³⁵  Individuals have displayed altered cortical measurements of excitability and inhibition for up to 30 years postconcussion.^36,37  Cortical excitability measures also play a key role in muscular strength.^38,39  These factors may have influenced knee-abduction angles in our athletes with a concussion history. Future research on the lower extremity neural control of movement postconcussion, which we did not collect, is warranted.

Although the peak knee-abduction angle throughout stance phase (initial ground contact to toe-off) during a jump-landing task is a risk factor for future anterior cruciate ligament tear, some researchers have reported other findings.¹⁰  For instance, the knee-abduction angle at initial ground contact is a risk factor for anterior cruciate ligament tears.⁴⁰  Furthermore, peak knee-abduction moment is also a common risk factor for anterior cruciate ligament tears and other MSKIs but was not different in our study.^10,40,41  In other words, although we found a difference in the knee-abduction angle between individuals with and those without a concussion history, it is not clear if this difference truly means the risk for MSKI is increased. Longitudinal research is needed to explore this connection and should include both peak and initial ground-contact outcomes, as time since concussion has been shown to be related to other landing biomechanics, such as knee-flexion angle.¹⁴ 

The reduced trunk lateral bending for athletes with a concussion history was unexpected yet in agreement with findings reported in previous literature.¹²  Greater trunk lateral bending toward the planted limb (results from our study) and less trunk flexion are considered dangerous landing positions and risk factors for lower extremity MSKI.^12,15  Lynall et al reported greater trunk flexion among individuals with than those without a concussion history.¹²  One possible explanation is that athletes with a concussion history performed the task more quickly (ie, better performance), as researchers have shown better performance metrics are related to more hazardous biomechanics.⁴²  In addition, Blackburn and Padua focused on group differences between a natural and instructed trunk-flexion position with safer lower extremity biomechanics, but a curvilinear relationship may exist between general trunk control and landing biomechanics that has not been explored.⁴³  Furthermore, the existence of a previous injury or risk factor (in this case, concussion) may moderate the relationship between what is considered safe or less hazardous. We see this in ankle injuries, for which univariate analyses have indicated ankle-torque asymmetry is not predictive of ankle injuries in a military population; however, the combination of body mass (a known MSKI risk factor) and ankle-torque asymmetry in a single model is predictive of MSKI.^44,45  In other words, when several risk factors are explored together, injury prediction improves. Future research should be done to investigate postconcussion trunk neuromuscular control between male and female athletes. Female athletes tend to have greater lateral trunk displacement during various jumping tasks compared with male athletes, as reported in a systematic review of athletes with unknown concussion histories.⁴⁶  However, we observed that athletes with a concussion history had less high-risk trunk biomechanics. Trunk biomechanics may not be implicated in future MSKI postconcussion for female athletes; however, longitudinal research is needed for confirmation.

Limb asymmetry is common among healthy individuals.⁴⁷  Injuries often lead to more extreme biomechanical asymmetries, and resolving these asymmetries is a common component of rehabilitation programs.⁴⁸  Our overall sample, both athletes with and those without a concussion history, may benefit from programs specifically addressing these asymmetries, especially as asymmetries have been related to an increased risk for lower extremity MSKI.¹⁷  However, athletes with a concussion history exhibited larger peak knee-abduction angles during a jump landing in their nondominant than dominant limb, whereas athletes without a concussion history showed no difference. This indicates greater asymmetry in knee-abduction angle for athletes with a concussion history. Lynall et al reported differences between athletes with and those without a concussion history for specific limbs.¹⁸ 

In our study, values that were different had medium to large effect sizes. Implementing dynamic movement–focused training programs may be useful for certain individuals during their return-to-play protocols. Jump-training programs have been shown to reduce MSKI in general and reduce fear of movement (kinesiophobia) after anterior cruciate ligament reconstruction.^49,50  Jump training could be important to implement postconcussion because kinesiophobia is associated with an increased risk for MSKI postconcussion.⁵¹  However, no longitudinal evidence supports the widespread adoption of landing-specific training to reduce the risk of lower extremity MSKI postconcussion specifically. Later stages of the concussion return-to-play protocol include dynamic movements and other noncontact sport-specific drills.^52,53  However, the duration and specific exercises in this stage for a given athlete are highly individualized by the medical practitioner treating the athlete. Future research may be done to explore if more time spent in a given stage or step of the return-to-play protocol or specific training protocols reduces MSKI risk postconcussion as jump training reduces MSKI in other populations.⁴⁹  Preliminary evidence (Cohen d = 0.480) showed that athletes with no MSKI postconcussion spent a longer time in their return-to-play protocols compared with athletes who went on to experience MSKI postconcussion.⁵⁴ 

Limitations for this study include self-reported concussion and injury history. Although we used self-report measures similar to those used in a previous study, this does not remove the risk of athletes misreporting their histories.²⁶  In addition, our sample reported a long, heterogeneous time postconcussion, which limits our ability to discuss short- versus long-term landing biomechanics. However, we covaried for time since concussion to help alleviate some of these concerns. Future researchers should focus on these outcomes longitudinally for a comprehensive understanding of movement changes postconcussion. Furthermore, previous injuries are known to influence dynamic movements and neuromuscular control.⁴⁸  Although we covaried for previously injured limb(s), this is still a limitation. In addition, our sample did not include male athletes, so we cannot state that female athletes postconcussion demonstrate different landing mechanics than male athletes that may predispose them to a higher risk of MSKI.⁶  At the same time, the inclusion of an all-female sample is a strength, as female athletes are highly underrepresented in concussion literature and have a higher risk for lower extremity MSKI.^6,55  Although our sample comprised only female athletes, we did not collect menstrual cycle information. Different menstrual cycle phases alter landing biomechanics and increase the risk for MSKI regardless of concussion history.^56,57 

Athletes with a concussion history displayed larger knee-abduction angles in the nondominant limb during jump landing and less trunk lateral bending during jump-landing and cutting tasks than athletes without a concussion history. This finding may be due to impairments in neuromuscular control postconcussion, but future research is needed to better understand potential deficits. The reason for less trunk lateral bending among athletes with a concussion history is not entirely clear. A smaller trunk lateral-bending angle toward the planted limb is typically considered a safer landing pattern. Perhaps the existence of a previous risk factor (eg, concussion history) changes what is considered a safe movement pattern. Future longitudinal evidence is needed to explore this concept and the link between landing biomechanics and MSKI postconcussion.

This project was supported, in part, by the Grand Alliance Concussion Assessment, Research and Education Consortium, funded in part by the National Collegiate Athletic Association and the Department of Defense. The US Army Medical Research Acquisition Activity, 820 Chandler Street, Fort Detrick, MD 21702-5014 is the awarding and administering acquisition office. This work was supported by the Office of the Assistant Secretary of Defense for Health Affairs through the Psychological Health and Traumatic Brain Injury Program under Award NO W81XWH-14-2-0151. Opinions, interpretations, conclusions, and recommendations are those of the author and are not necessarily endorsed by the Department of Defense (DHP funds).

Supplemental Table 1. Jump-Landing Model Characteristics.

Supplemental Table 2. Cutting Model Characteristics.

Found at DOI: https://dx.doi.org/10.4085/1062-6050-0259.23.S1