Association Between Knee- and Hip-Extensor Strength and Running-Related Injury Biomechanics in Collegiate Distance Runners

We used a cross-sectional design and collected data during 2 sessions that were conducted in random order and separated by a 1-week washout period. One session consisted of assessments of isometric strength and running biomechanics, and the other session consisted of an assessment of back-squat strength.

A total of 36 collegiate distance runners (26 men, 10 women) participated in the study (Table 1). Using an a priori power analysis (G*Power 3; Heinrich Heine Universität, Düsseldorf, Germany¹⁹ ), we found that 29 participants would be needed to identify a weak-to-moderate association via stepwise regression (ΔR² = 0.15, ΔP = .05, β = .2, residual variance = 0.50) after accounting for 3 covariates. All participants had a minimum of 6 months of experience in resistance training with the back squat, no injury in the 2 months before the study, and no lower extremity surgery in their lifetime. Injury was defined as missing >1 week of training and not returning to a minimum of 50% of their previous training volume.^20,21  All participants provided written informed consent, and the study was approved by the Institutional Review Board of California State University, Fullerton.

One-repetition maximum testing was completed using the back-squat exercise in a single session according to a previously established protocol.²²  The warm-up protocol involved 10 repetitions with an unloaded barbell and 5 repetitions at 50%, 4 repetitions at 70%, 1 repetition at 90%, and 1 repetition at 100% of the participant's estimated 1RM. The test continued with single repetitions until the participant could not visibly reach 90° of knee flexion, dropped the weight, or voluntarily terminated the test. Participants rested 3 minutes between all sets of repetitions, and the number of attempts was 3 ± 1.9.

Maximal knee- and hip-extensor strength of the dominant limb (ie, preferred limb to kick a ball) was evaluated via MVICs on an isokinetic dynamometer (HUMAC NORM, Stroughton, MA). All participants completed a standardized warm-up protocol consisting of two 5-second submaximal contractions at 50% and 75% of self-perceived maximal effort, followed by 1 practice MVIC at 100% effort. All participants completed 3 MVIC attempts after the warm-up protocol, received oral encouragement and visual feedback on the torque output, and rested 30 seconds between attempts. Knee-extension MVICs were performed with participants seated and the hip (determined as the trunk relative to the thigh) and knee flexed to 110° and 60°, respectively. Hip-extension MVICs were performed with participants positioned prone and the hip and knee flexed to 60° and 90°, respectively.²³  The dynamometer pad was placed 2 fingerwidths from the medial malleolus and femoral condyles for the knee- and hip-joint MVICs, respectively, and the axis of the arm was visually aligned with the joint center.

Running biomechanics were analyzed after the MVICs were completed and approximately 10 minutes of rest. The warm-up consisted of a 10-minute treadmill (Star Trac model 8TR; Core Health & Fitness, LLC, Vancouver, WA) run at the participant's self-selected speed. Participants wore laboratory-standard neutral-cushion running shoes (model Pegasus 32; Nike, Inc, Beaverton, OR) and tight-fitting clothing. Static retroreflective markers were placed on the first thoracic vertebra and lumbosacral joint and bilaterally on the iliac crest, anterior-superior iliac spine, posterior-superior iliac spine, greater trochanter, and acromioclavicular joints. Unilateral markers were placed medially and laterally on the femoral condyles and malleoli, the first and fifth metatarsal heads, and the heel counter of the dominant limb. Dynamic tracking clusters containing 4 retroreflective markers were placed on the thigh, shank, and foot of the dominant limb to track segment motion. The markers on the first thoracic vertebra, lumbosacral joint, and acromioclavicular joints were kept in place to track trunk motion, and markers on the anterior-superior iliac spine and posterior-superior iliac spine were kept in place to track the pelvis.²³  All lower extremity static markers except at the heel counter were removed before the running trials began. All static and dynamic trials were recorded at 240 Hz by using a Qualisys (Gothenburg, Sweden) 9-camera motion-capture system.

Participants performed 5 overground practice running trials on a 20-m runway and through 2 timing gates (model TF100; TrackTronix, Lenexa, KS) to establish a consistent self-selected speed. The participants then performed 5 trials at their self-selected speed (±5%), contacting 1 force plate (Advanced Mechanical Technology, Inc, Watertown, MA) sampling at 2400 Hz. Trials were recorded if the dominant foot contacted the plate under each speed condition. Trials were repeated if the running speed was not within 5% of the previously established speed, the foot missed the force plate, or the participant visually changed his or her gait to contact the force plate.

The 1RM back squat was also normalized to body mass, and the maximal weight lifted was used for analysis. The MVIC torque data were gravity corrected and normalized to body mass, and the peak torque of the 3 trials was used for analysis. Marker position and force data were low-pass filtered at 20 Hz.²⁴  We calculated the hip-joint center as 25% of the distance between greater trochanter markers²⁵  and subsequent joint centers as the center point between the lateral and medial static markers. Joint angles and moments were determined using Visual3D software (C-Motion, Inc, Germantown, MD). Vertical loading rate was computed using a custom LabView program (National Instruments, Austin, TX).

Heel contact and toe-off were identified when the vertical ground reaction force exceeded 20 N and fell below 20 N, respectively, and indicated the length of the stance phase. Joint kinematics were computed as the motion of the distal relative to the proximal segment using the Cardan sequence of flexion-extension, abduction-adduction, and internal-external rotation. Trunk angle was calculated as motion of the trunk segment relative to the global vertical axis. Peak angles and moments (knee flexion, abduction, and internal rotation and hip flexion, adduction, and internal rotation) during the stance phase were extracted for analysis. Vertical loading rate was defined as the peak derivative during the first 13% of the stance phase and normalized to body weight.²⁶  Internal joint moments were determined using inverse dynamics and were normalized to a product of body weight and height. We time normalized all kinematic and kinetic waveforms to 100% of the stance phase and plotted them as ensemble averages for descriptive purposes (Figures 1 and 2). The foot-strike index²⁷  was used to quantify the footfall pattern, which may influence knee and hip biomechanics during running.²⁸  Briefly, the foot-strike index quantifies the footfall pattern via the position of the center of pressure at heel contact expressed as a percentage of the foot-segment length (rearfoot = 0%–33%, midfoot = 33%–67%, forefoot = 67%–100%). Foot-strike index was calculated using a modified foot segment¹⁶  and was a continuous variable.

Descriptive statistics (means and standard deviations) were computed for all outcome variables. Outcomes were assessed for normality using the Shapiro-Wilk test and via inspection of skewness and kurtosis (ratio of standard error to statistic). Outliers were examined using boxplots. We performed separate forward stepwise linear regression models to address the unique association between each strength (predictor) and biomechanical (criterion) outcome after accounting for sex (male = 0; female = 1), running speed (m/s), and foot-strike index (%). Covariates were included because previous research^28,29  indicated that running biomechanics differed between males and females, forefoot and rearfoot runners, and faster and slower speeds. Covariates were entered first (step 1), and the strength variable of interest was entered second. The change in the coefficient of determination (ΔR², ΔP) from step 1 to step 2 was used to represent the magnitude of association (α = .05, β = .2). All analyses were conducted using SPSS (version 26; IBM Corp, Armonk, NY).

All data were normally distributed, and no outliers were identified. Descriptive statistics are presented in Table 1; the sample was predominately male. One participant did not complete the 1RM back-squat assessment (ie, did not return for the second session), so the analyses were limited to 35 participants for this outcome. Similarly, a technical problem arose for 1 participant during the hip-extension MVIC, and the analyses were limited to 35 participants for this outcome. Covariates explained 1.2% to 22.7% of the variance in the running kinematic variables (Supplemental Table 1) and 4.75% to 24.7% of the variance in the running kinetic variables (Supplemental Table 2).

Greater 1RM back-squat strength was associated with a larger peak knee-flexion angle (ΔR² = 0.110, ΔP = .045, total R² = 0.13), smaller peak knee internal-rotation angle (ΔR² = 0.127, ΔP = .03, total R² = 0.36), and smaller peak knee internal-rotation moment (ΔR² = 0.129, ΔP = .03, total R² = 0.38) after accounting for sex, speed, and foot-strike pattern (Table 2). No associations were found between 1RM back-squat strength and hip kinematics or kinetics, trunk-lean angle, or vertical loading rate (all ΔP values > .05).

No associations were present between maximal isometric knee-extension strength and any biomechanical variable (all ΔP values > .05; Table 3). Similarly, no associations were observed between maximal isometric hip-extensor strength and any running biomechanical variable after accounting for sex, speed, and foot-strike index (all ΔP values > .05; Table 4).

The purpose of our study was to determine the associations between knee- and hip-extensor strength (1RM back-squat and isometric knee- and hip-extensor strength) and running biomechanics that have previously been associated with overuse injuries in competitive distance runners. The 1RM back squat was used as a global assessment of lower extremity strength that involves similar muscles and biomechanics to those involved in running.³⁰  We selected knee- and hip-extensor isometric strength tests because they are clinical assessments of strength that are often used to assess distance runners.²³  The primary finding was that greater 1RM back-squat strength was associated with a larger peak knee-flexion angle, smaller peak knee internal-rotation angle, and smaller knee internal-rotation moment after adjusting for sex, running speed, and foot-strike index. However, 1RM back-squat strength was not associated with other kinematic variables or running kinetics. Furthermore, isometric strength was not associated with running kinematics or kinetics. Finally, sex, running speed, and foot-strike index were included as covariates and were associated with a variety of running kinematics and kinetics. As such, future researchers should consider these factors when analyzing relationships between strength and running biomechanics.

We hypothesized that greater 1RM back-squat strength would be related to greater peak knee-extensor moments because greater strength may allow runners to produce a larger torque at the knees during running. Although our findings did not support this hypothesis, we noted a relationship between 1RM strength and greater peak knee-flexion angle during the stance phase. Stronger runners may be able to use greater knee flexion without an increased joint moment, thereby creating a more favorable landing strategy during the early stance phase of running. Sagittal-plane knee-joint stiffness is derived from the change in moment divided by the change in angle and has been prospectively linked to overuse injury in recreational runners.¹²  Therefore, less knee flexion may contribute to additional knee-joint stiffness and may be partially attributable to lower 1RM back-squat strength.

We also found that greater 1RM back-squat strength was associated with smaller peak knee internal-rotation angle and internal-rotation moment. Peak knee internal-rotation angle and internal-rotation moment are associated with overuse knee conditions in runners. For example, female runners with ITBS displayed a larger peak knee internal-rotation angle than did uninjured controls.³¹  The IT band originates on the gluteus maximus and tensor fascia lata muscles and inserts onto the Gerdy tubercle at the tibia. Thus, it can assist with rotary and frontal-plane stabilization of the knee joint,^8,32  and gluteus maximus muscle weakness may contribute to lowest knee-joint stability during running.

Contrary to our hypothesis, we identified no relationship between 1RM back-squat strength and hip kinematics or kinetics during running. The back squat is an exercise used to target the knee extensors rather than the hip extensors.³³  Conversely, the deadlift exercise is hip dominant, and greater sagittal-plane hip-joint moments and power occur during the deadlift than during the back squat.³³  We found a substantial proportion of unexplained variance in knee kinematics despite associations between 1RM back-squat strength and knee-flexion and internal-rotation angles. Distance running is a submaximal activity, and maximal strength may not characterize the task demands. Similarly, movement velocity is slower during a 1RM back squat than during running. Other factors, such as muscular endurance and the rate of force development, may be better indicators of muscle function during running, and future investigators should comprehensively evaluate these characteristics.

We also demonstrated no association between any strength measurement and hip frontal-plane running biomechanics. Iliotibial band syndrome has been linked to altered hip frontal-plane kinematics. Ferber et al³¹  determined that runners with ITBS had larger peak hip-adduction angles than uninjured control participants. Furthermore, the hip-adduction angle has been linked to other knee-joint injuries, such as patellofemoral pain syndrome.³⁴  The squat and the isometric hip-extensor tests used in our study primarily occur in the sagittal plane. Brund et al³⁵  also observed no relationship between hip-abductor eccentric strength and frontal-plane hip or knee kinematics during running. Similarly, the bilateral squat test and non–weight-bearing isometric test likely do not reflect the biomechanical demands of running, which is unilateral and weightbearing.

The discordance in our findings between the 1RM back-squat and isometric strength assessments may be due to contraction type. The 1RM back squat uses a dynamic eccentric muscle action, followed by a concentric muscle action similar to the muscle-contraction demands during running. Conversely, isometric strength may not accurately characterize muscle function during dynamic tasks, such as running. The back squat is a commonly used exercise for strengthening the knee and hip extensors in collegiate distance runners,³⁶  and our results indicated that the 1RM back squat was associated with favorable knee-joint kinematics (ie, greater peak flexion angle and smaller peak internal-rotation angle).

No relationship was evident between strength outcomes and trunk-lean angle during running, which was contrary to earlier findings²³  of a relationship between greater isometric hip-extensor strength and a larger forward trunk lean in recreational runners. The lack of associations in this investigation may have reflected the experience and running speed of the participants. Our competitive runners ran an average of 84.56 km/week at a speed of 4.12 m/s (collapsed across sex). In contrast, the recreational runners studied by Teng and Powers²³  ran an average of 35 to 37 km/wk and were analyzed at a predetermined speed of 3.4 m/s. Peak trunk-lean angle is altered when runners are instructed to run at a speed different from their preferred speed.³⁷  Therefore, the predetermined speed used by Teng and Powers²³  may not have accurately represented the runners' preferred trunk-lean angle. Our hypothesis was based on the work of previous authors³⁸  who identified an association between greater trunk lean and lower patellofemoral contact stress. However, excessive trunk lean may be detrimental to performance and optimal running mechanics, and an optimal trunk lean angle is unknown and should be addressed in future research.

The weak associations between strength and running biomechanics should not detract from other benefits of strength training for collegiate distance runners. Improvements in force-production capabilities of the propulsive muscles (hip and knee extensors and plantar flexors) improve tissue tolerance to mechanical loading¹⁷  and performance in competition.³⁹  Bertelsen et al¹⁸  suggested that injury occurs after the load capacity of a runner is exceeded. Strength training may be an intervention to increase runners' capacity and reduce their future injury risk.¹⁸  Moreover, weight training contributes to large improvements in performance, and the benefits are more profound for well-trained than recreational athletes.³⁹  In a recent meta-analysis, Trowell et al¹¹  indicated that strength training in runners contributed to minimal adaptations in running biomechanics but did improve knee-extensor, knee-flexor, and plantar-flexor strength. In another meta-analysis, Lum and Barbosa³⁹  showed that strength training was associated with faster performances in Olympic time-based sports. Future authors need to examine the associations between other metrics of muscle function (eg, muscular endurance, fatigability, and the rate of force development) and running biomechanics.

Limitations should be considered when interpreting the results of our study. First, the population was predominately male, and running biomechanics may differ by sex.²⁹  However, to account for the unequal sex distribution, sex was a covariate in the regression analyses. Second, our collegiate distance runners were not analyzed at the same point within their competitive seasons. Most runners were analyzed during the offseason and preseason phase for either cross-country or track and field. Thus, the preparation stage and strength-training schedule may have influenced strength capacity. Third, the foot segment we used in our biomechanical analyses resulted in a truncated foot because the distal endpoint was represented by the midpoint of the first and fifth metatarsal markers. Consequently, foot-strike index values can exceed 100%, and the dichotomous classification of runners into footfall pattern categories (ie, rearfoot, midfoot, and forefoot) may not be accurate. Hence, we used the foot-strike index as a continuous variable (%), for which a larger value still represents more of a forefoot than rearfoot position at ground contact. Fourth, we chose the back squat as an assessment of overall lower body strength; it may not reflect performance of all exercises (eg, deadlift) included in a distance-runner's strength-training program. Future researchers should investigate the association between alternative strength-training exercises and distance-running biomechanics. Fifth, the runners were free of injury within the 2 months before testing; however, 4 participants reported a history of stress fracture, and 15 participants described various soft tissue injuries in the year before the study, which may have influenced running biomechanics.¹⁴  Nevertheless, it would be impractical to recruit collegiate distance runners with no history of injury given the high incidence of running-related injuries.

Greater 1RM back-squat strength was weakly associated with a larger peak knee-flexion angle, smaller peak knee internal-rotation angle, and smaller peak knee internal-rotation moment in collegiate distance runners. Conversely, maximal knee- and hip-extensor isometric strength was not associated with any running biomechanical variable. Maximal strength was not strongly related to distance-running biomechanics. However, strength may play a greater role in overall performance and tissue resilience in collegiate distance runners.