Fatigue-Induced Hip-Abductor Weakness and Changes in Biomechanical Risk Factors for Running-Related Injuries

College-aged males who were active participants in the university's Army Reserve Officers' Training Corps were recruited, as running is consistently used in physical training, and lower extremity overuse injuries are prevalent in this population.^2,8  Before data collection, all participants underwent the informed consent process and signed the consent form approved by the university institutional review board, which also approved the study. All participants completed a brief questionnaire that assessed previous lower extremity injury and a medical history form, both of which were evaluated by an athletic trainer. All participants were considered healthy; those with a physician's orders prohibiting them from being involved in physical activity or who had undergone lower extremity surgery within the previous 6 months were excluded from the study. Only those in the low-risk group according to the American College of Sports Medicine Risk Stratification Categories (https://www.acsm.org/blog-detail/acsm-certified-blog/2019/11/11/acsm-risk-stratification-chart-download) were included in the study. All participants reported right-leg dominance.

A single data-collection session consisted of HABD testing and a biomechanical running analysis before and after a fatiguing run. All participants underwent the following sequence: (1) rested-state HABD testing, (2) rested-state running biomechanics, (3) fatiguing-run protocol, (4) fatigued-state running biomechanics, and (5) fatigued-state HABD testing. The data-collection procedures for HABD testing were identical for the rested and fatigued states. The data-collection procedures for running biomechanics were identical for the rested- and fatigued-state gait analyses, except that participants ran, as opposed to walked, back to the starting position during the latter trials.

All isometric HABD measures were collected by the same nonauthor researcher. Isometric HABD was measured in the right limb with the participant lying on his contralateral side with a pillow between the knees to align the hip in frontal-plane neutral position.²⁷  A Microfet (Hogan Industries, Draper, UT) handheld dynamometer was placed on the lateral femoral epicondyle and secured with a nylon strap anchored to the testing table.²⁷  Each individual was instructed to maximally abduct the hip against the dynamometer for 3 seconds.¹⁹  Five trials were performed: 2 initial practice trials and 3 data-collection trials. The force production during the final 3 trials was averaged for data analysis.²⁸  A 30-second rest was allowed between trials.²⁷ 

Recruits wore their personal running shoes during data collection. Retroreflective markers were attached bilaterally on the thorax, pelvis, and lower extremity in accordance with previously published methods.⁸  A 3-dimensional motion-capture system situated along an 18-m runway and Nexus software (Vicon Motion Systems Ltd, Centennial, CO) were used to capture kinematic data. Kinetic data were recorded using a force platform (Advanced Mechanical Technology Inc, Watertown, MA) embedded flush with the runway. Kinetic data were collected at 960 Hz and time synchronized with kinematic data collected at 240 Hz.

A static calibration trial was conducted before the dynamic trials. Markers on the medial femoral condyles and medial malleoli were removed for the dynamic trials. Running data-collection trials began after the calibration trial, which enabled participants to become familiar with the collection procedures and running velocity. A trial was considered successful if the individual ran at the correct velocity (4.0 m/s ± 10%), as measured by Speedtrap II (Brower Timing Systems, Draper, UT) infrared sensors placed 4 m apart in the middle third of the runway, and landed with the entire foot on the force plate without apparent targeting. Data from 3 successful running trials were collected for the right limb. To assess exertion and fatigue, we obtained ratings of perceived exertion (RPEs) on the Borg scale and ratings of physical fatigue on the fatigue analog scale (FAS) before and after the running trials.

After completing the rested-state running trials, participants began the fatiguing-run protocol, which consisted of a graded exercise test, 5-minute seated rest, and exhaustive run. The protocol was performed on a treadmill (model Quinton Medtrack T65; Cardiac Science Corp, Bothell, WA). Retroreflective markers on the shoes and anterior-superior iliac spines were removed for the treadmill protocol. Metabolic data were collected using a metabolic cart via open-circuit indirect calorimetry (model TrueOne 2400; ParvoMedics, Inc, Sandy, UT). Calibration was conducted before each data-collection session according to the manufacturer's instructions.

The treadmill protocol began with a speed-blinded modified Astrand graded exercise test (GXT) to determine maximum oxygen consumption (o₂max).²⁹  Standardized instructions were given before the GXT, with an emphasis on the importance of maximum effort. The GXT was stopped when the participant reached the point of volitional exhaustion, and o₂max was confirmed based on his meeting 1 of the following criteria: respiratory exchange ratio >1.15, plateau in maximum oxygen output with an increase in work rate, blood lactate >8 mmol, or RPE >17 on termination of the GXT.²⁹  After the GXT, the individual rested while seated for 5 minutes; blood lactate was collected via finger prick during the final minute of the rest period. The RPE and FAS data were collected immediately at the point of volitional exhaustion on the GXT and after the 5-minute rest period.

After the seated rest, the breathing apparatus was refitted, and the participant began the exhaustive run at a speed calculated to elicit 80% of o₂max at a 1% grade. Metabolic data were collected during the first 3 minutes of the run to confirm attainment of 80% ± 5% o₂max. If this metabolic threshold was not met, the speed was increased until it was achieved, and then the breathing apparatus was removed. To gauge the individual's fatigue level, RPE and FAS measurements were collected during the exhaustive run at 5 minutes, the GXT time, twice the GXT time, and the end of the exhaustive run. Participants were encouraged to continue running at the prescribed treadmill speed for twice the length of the GXT or until each reported an RPE >17 or an FAS score >7, whichever came second. Those who met the RPE and FAS criteria but thought they could not maintain the running pace for twice the length of the GXT were allowed to decrease the pace, within a 5–8 mph (8–13 kph) range, as long as the RPE and FAS criteria were maintained until the time criterion was achieved. If the participant needed to discontinue due to volitional exhaustion, the exhaustive run was ended.

Immediately after the fatiguing-run protocol, the retroreflective markers were replaced as needed, a static calibration trail was repeated, and data collection for the fatigued-state biomechanics began, followed by the isometric HABD measures. Collection of the fatigued-state running biomechanical, RPE, FAS, and isometric HABD measures was performed using the same methods as for the rested state, with the addition of running between gait trials. Water was available to each person during all aspects of data collection.

Data from participants who did not meet at least 1 of the criteria for o₂max were excluded from analysis. The kinematic and kinetic data were filtered using a low-pass, fourth-order Butterworth filter with a 10-Hz cutoff frequency. Kinematic variables of interest were peak angles and velocities for hip adduction, hip internal rotation, knee adduction (varus), and knee internal rotation. Kinetic data used for joint moments were filtered before moment calculations. Kinetic variables of interest were peak hip and knee adduction and internal-rotation moments. All moments were expressed as external moments and normalized to body mass. Variables of interest were identified from each of the 3 successful trials, and an ensemble mean was used for analysis. Isometric HABD measures were normalized to body mass (N/kg) and then converted to torque (Nm/kg) by multiplying by the length of the lever arm.^17,28  The lever arm was the distance (meters) from the greater trochanter to the lateral femoral epicondyle, where the dynamometer was placed.²⁸ 

We analyzed the data using SPSS (version 24.0; IBM Corp, Armonk, NY) with the α level set a priori at P < .05. Differences in HABD torque, biomechanical variables, and the RPE and FAS measures between conditions (rested versus fatigued) were compared using paired t tests. Pearson correlation coefficients were calculated to evaluate correlations between (1) rested isometric HABD torque and rested biomechanical variables, (2) fatigued-state isometric HABD torque and fatigued-state biomechanical variables, and (3) percentage change in isometric HABD torque and the biomechanical variables due to fatigue.

A total of 38 males successfully completed the study protocol, and their data were analyzed. Five participants terminated the exhaustive run due to volitional exhaustion, and the exhaustive runs of 33 participants were ended because the termination criteria outlined in the “Methods” were met. Participant demographics and data-collection metrics are reported in Table 1. Rested- and fatigued-state HABD torque and the biomechanical variables of interest are provided in Table 2. The HABD torque decreased from the rested to the fatigued state. Hip-adduction angle, hip internal-rotation velocity, hip internal-rotation moment, knee-adduction (varus) velocity, knee internal-rotation velocity, and knee internal-rotation moment increased from the rested to the fatigued state.

In the rested state, HABD torque was significantly correlated with maximum hip-adduction velocity (r = −0.322, P = .049). In the fatigued state (Figure), HABD torque was correlated with maximum hip-adduction velocity (r = –0.393, P = .015), hip internal-rotation velocity (r = –0.410, P = .01), maximum knee-adduction (r = 0.385, P = .017), and maximum knee-adduction velocity (r = –0.378, P = .019). The percentage change in HABD torque was not significantly correlated (P ≥ .05) with the percentage change in any of the biomechanical variables.

In healthy, physically active males, an exhaustive running bout at 80% of o₂max produced a 15% decrease in HABD torque. In addition, we observed fatigue-induced changes in biomechanical variables that were previously associated with running-related injury,^8,10,11  including increased hip-adduction angle and knee-adduction (varus) and internal-rotation velocities. The fatigued state also elicited significant correlations that were not present during the rested state between HABD torque and the biomechanical variables of interest. Use of the fatigued state to uncover underlying, potentially injurious metrics is endorsed. Compared with control individuals, patients with iliotibial band syndrome displayed increased knee adduction after 30 minutes of running but not at the onset of running.³⁰  Collectively, these findings support the need to evaluate theoretical frameworks for the development of running injuries in a fatigued state.

The influence of fatigue on HABD torque and, subsequently, the relationship between HABD torque and the biomechanical variables previously associated with running-related injury, were our primary interest in the current study. The decrease we noted in HABD torque (15%) was similar to that reported earlier after an exhaustive run (11%).¹⁰  Although HABD torque was correlated with peak hip adduction in participants with patellofemoral pain syndrome,¹⁰  we observed no correlation between these variables in our population of healthy males (rested state: r = –0.263, P = .10; fatigued state: r = –0.189, P = .24). Our findings support the healthy control group findings reported by Dierks et al¹⁰  (before a prolonged run: r = 0.017, P = .472; after a prolonged run: r = 0.051, P = .208). Thus, HABD weakness, which resulted from a fatiguing run in our investigation, was not related to the increased hip-adduction angle. Similarly, Pohl et al³¹  demonstrated no compensatory patterns in walking after a nerve block to simulate HABD weakness. Previous associations between HABD weakness and hip adduction in those with running-related injuries may not have been due to muscle weakness alone; pain may be related to compensatory patterns for increased hip adduction, not strength.^21,22  Muscle weakness and pain from lower extremity overuse injury occur over an extended period, allowing compensatory patterns to develop.^31,32  Decreases in strength from fatigue may not persist long enough for compensatory patterns to develop,³¹  which may explain the absence of a correlation between HABD torque and hip-adduction angle in our research.

The only peak angle significantly correlated with HABD torque was knee adduction in the fatigued state (r = 0.385, P = .017); in contrast, Dierks et al¹⁰  observed no relationship. However, different populations might have contributed to the contradictory findings, as Dierks et al¹⁰  studied a mixed-sex population (5 males, 15 females). Movement patterns associated with running-related injuries have differed between sexes.^33,34  Furthermore, knee adduction was greater in males with patellofemoral pain syndrome than in male control participants and females with patellofemoral pain syndrome.³⁴  Our fatigued-state findings occurred in a healthy population; future investigators should continue to evaluate knee adduction in relation to running-related injury in males.

We assessed HABD torque in relation to not only the magnitude of movement (peak angle) but also the rate of movement (velocity). The relationship between HABD torque and velocities at the hip and knee was examined because prospective findings^8,35  indicated that rate of motion may be a better variable for determining injury risk than joint angles alone. Hip-adduction velocity was significantly correlated with HABD torque in both the rested and fatigued states. In addition, knee adduction and hip internal-rotation velocity were negatively correlated with HABD torque. These results, in relation to evidence supporting the importance of the rate of movement in the development of running-related injuries,^8,35  have applications for clinical injury prevention. In a randomized, controlled trial, Toresdahl et al³⁶  evaluated how a home-exercise strength-training intervention influenced injury rates in first-time marathon participants. The study design allowed the effects of the intervention to be appropriately assessed because the strength-training and non–strength-training groups did not differ in training loads or finishing times.³⁶  Despite 12 weeks of strength training by the experimental group, no difference in injury rate was reported between groups.³⁶  Our findings suggest that the velocity of a task performed for injury-prevention purposes must be considered. Exercises performed in the Toresdahl et al³⁶  intervention were primarily focused on sagittal-plane movements (eg, squats, lunges, and single-legged toe touches) or isometric stability (eg, plank and side plank). Although additional work is needed to identify the role of HABD strength in running-related injury prevention,³⁷  clinicians should take into account both the magnitude and rate of force development when incorporating HABD strength training.

A primary limitation of our research was the use of isometric handheld dynamometry to evaluate HABD torque. During gait, the HABDs are required to act eccentrically to stabilize the pelvis. Thus, isometric measures may lack construct validity.^23,38  Eccentric isokinetic measures may have more ecologic validity; however, we chose isometric testing due to its clinical utility. Significant correlations between HABD torque and joint velocities warrant further investigations using measures of eccentric HABD capabilities. Another measure of muscle performance that we did not explore but which may be related to injury is hip power. Authors³⁹  of a prospective investigation noted that decreased average hip external-rotator power predicted lower extremity injury in females; however, HABD peak torque was not predictive. Including only male participants may have been a further limitation of our study, yet because increasing evidence^33,34  indicates that sex may play a role in pathomechanics, we excluded females. Future researchers should evaluate the effects of fatigue on HABD torque and gait mechanics in females and, ideally, should compare sexes.

In summary, we demonstrated that a fatiguing run resulted in decreased HABD torque and changes in biomechanical variables that were previously associated with running-related injury. The biomechanical variables increased by fatigue were hip adduction, knee-adduction (varus) velocity, and knee internal-rotation velocity. Although HABD torque and peak hip-adduction angle had earlier been associated with running-related injury and were negatively affected by fatigue in our analysis, no relationship between these variables was present. Our investigation was performed in healthy males, but our findings support the growing body of literature indicating that HABD weakness and increased hip-adduction angle during gait may be secondary to injury and not causal. The HABD torque was not correlated with peak hip adduction, yet it was associated with knee adduction in the fatigued state. This result may be unique due to our all-male population. Finally, changes in joint velocities due to fatigue and correlations between HABD torque and hip- and knee-joint velocities highlight the clinical need to consider not only the quantity of HABD strength but also the quality of motion and eccentric control provided by the HABDs.

This research was partially funded by a grant from the Far West Athletic Trainers' Association. We thank Ms Katlin Michaels and Bethany Wisthoff, PhD, ATC, for their assistance with data collection.