Neuromechanics of Middle-Distance Running Fatigue: A Key Role of the Plantarflexors? : Medicine & Science in Sports & Exercise

Journal Logo


Neuromechanics of Middle-Distance Running Fatigue: A Key Role of the Plantarflexors?


Author Information
Medicine & Science in Sports & Exercise 53(10):p 2119-2130, October 2021. | DOI: 10.1249/MSS.0000000000002695


Identifying the neuromechanical changes during high-intensity running to fatigue may highlight the biomechanical limitations to performance and indicate mitigation/training strategies.


This study aimed to investigate the changes in lower limb kinematics, kinetics, and muscle activation during a high-intensity run to fatigue (HIRF).


Eighteen male and female competitive middle-distance runners performed a HIRF on an instrumented treadmill at a constant but unsustainable middle-distance speed (~3 min) based on a preceding maximum oxygen uptake (V˙O2max) test. Three-dimensional kinematics and kinetics were collected and compared between the start, 33%, 67%, and the end of the HIRF. In addition, the activation of eight lower limb muscles of each leg was measured with surface EMG (sEMG).


Time to exhaustion was 181 ± 42 s. By the end of the HIRF (i.e., vs the start), ground contact time increased (+4.0%), whereas flight time (−3.2%), peak vertical ground reaction force (−6.1%), and vertical impulse (−4.1%) decreased (all P < 0.05), and joint angles at initial contact became more (dorsi)flexed (ankle, +1.9°; knee, +2.1°; hip, +3.6°; all P < 0.05). During stance, by the end of the HIRF: peak ankle plantarflexion moment decreased by 0.4 N·m·kg−1 (−9.0%), whereas peak knee extension moment increased by 0.24 N·m·kg−1 (+10.3%); similarly, positive ankle plantarflexion work decreased by 0.19 J·kg−1 (−13.9%), whereas positive knee extension work increased by 0.09 J·kg−1 (+33.3%; both P < 0.05) with no change in positive hip extension work. Hip extensor surface EMG amplitude increased during the late swing phase (+20.9–37.3%; P < 0.05).


Running at a constant middle-distance pace led primarily to the fatigue of the plantarflexors with a compensatory increase in positive work done at the knee. Improving the fatigue resistance of the plantarflexors might be beneficial for middle-distance running performance.

Middle-distance running, exemplified by races over 800 and 1500 m, can be classified as high/severe-intensity exercise, and world class male runners achieve average speeds of >7 m·s−1. High-intensity running involves significant anaerobic energy metabolism (1) leading to substantial metabolic disturbance (2) and fatigue, defined as a loss of force or power production from the active musculature (3), that impairs performance and ultimately leads to task failure. However, the neuromechanical changes during a high-intensity run to fatigue (HIRF) remain largely unknown, and thus the specifics of how fatigue manifests as a neuromechanical limitation has yet to be clearly elucidated. Understanding the neuromechanical basis of fatigue may inform methods (e.g., training and technique) for mitigating those neuromechanical effects and potentially enhance performance.

Previous research has almost exclusively investigated exhaustive running to fatigue at slower speeds/intensities (e.g., durations from 10 min [4] up to ultramarathons [5]) and how this affects kinematics and neuromuscular activation. However, fatigue effects may vary between slow and fast running because of their differing neuromechanics; for example, the contribution of the ankle, knee, and hip joint muscles to the total work done during stance and swing is known to change with running speeds ranging from 3.5 to 8.95 m·s−1 (6). In addition, neuromuscular activation assessed with surface EMG (sEMG) during cycling to exhaustion increased at high/severe-intensity (i.e., above critical power), but first increased and then decreased at intensities below critical power, indicating intensity-dependent activation responses (7). These differences suggest that neuromechanical changes with fatigue during high-intensity running likely differ from lower-intensity running, although the fatigue effects of high-intensity running (i.e., > V˙O2max) remain largely unknown.

Moreover, existing research of neuromechanical changes with running fatigue has also typically used self-paced time trials, where running speed is inherently variable (e.g., [8–10]). Given the known effect of running speed on kinematics (11), ground reaction force (GRF), and joint moments (6), this approach has made it difficult to differentiate the effects of fatigue from those of speed and, thus, confounded our understanding of the isolated influence of fatigue on running neuromechanics. Nonetheless, a few studies using exhaustive constant pace runs at 80%–95% V˙O2max have reported increasing ground contact times (GCT) and decreasing flight times throughout the run (4,12), which indicates changes in sagittal plane kinematics. Furthermore, increased GCT when fatigued has been associated with more knee flexion during initial contact (IC) and at midstance (13–16) as well as lower peak vertical GRF (4,17).

Overall, to date there have been very few assessments of how joint moments and precipitant muscle activations, which are ultimately responsible for the changes in running kinematics, alter with fatigue at any intensity. A small preliminary study (n = 6) of a 600-m running time trial used inverse dynamics and revealed decreased stance phase ankle joint plantarflexion moment that appeared to explain lower peak vertical GRF (18). A larger study of constant-paced running to exhaustion at a lower speed of 10-km race pace found a similar decline in stance phase positive work by the ankle joint muscles, which was compensated for by an increase in positive work from the proximal knee and hip joint muscles (19). Moreover, during unfatigued running, the plantarflexors appear to operate at a higher proportion of their maximum muscle forces and perform a larger proportion of positive work during stance than other muscle groups (20,21), and this disproportionately large contribution (~60%) may make the plantarflexors particularly susceptible to fatigue. The fatigue of individual muscles may also be inferred from increases in neuromuscular activation, assessed with EMG, as activation is known to increase during fatiguing submaximal contractions (22). Conversely, running to exhaustion at the velocity of V˙O2max exhibited decreased EMG of the plantarflexors, gastrocnemius medialis, lateralis, and soleus during late swing and stance, suggesting decreased muscle activation with fatigue (23). Further mechanistic examination of the kinetic and muscle activation changes that underpin the observed kinematic effects of fatigue is clearly required to help identify the specific muscle groups that are most affected by a HIRF.

Thus, the aim of this study was to investigate the changes in sagittal plane running kinematics and kinetics, particularly underlying joint moments and joint work, as well as muscle activation strategies during an HIRF. Specific hypotheses were that with fatigue (i.e., at the end vs the start of the run), there would be (i) an increase in GCT and a decrease in flight time; (ii) an increase in knee flexion angle during stance; and particularly (iii) decreases in plantarflexion moment, and positive and negative work, during stance with a concomitant decrease in vertical GRF.



Eleven males (age, 21.9 ± 3.6 yr; body mass, 67.5 ± 3.9 kg; height, 1.82 ± 0.05 m) and 7 females (age, 25.4 ± 4.5 yr; body mass, 54.0 ± 4.1 kg; height: 1.65 ± 0.02 m) took part in this research study. The participants were national standard runners reflected by International Association of Athletics Federations points of 898 ± 98 (males) and 952 ± 99 (females) based on season’s best 800–1500 m times, equivalent to 1500 m times of 3:56 ± 0:08 min:s (males) and 4:33 ± 0:13 min:s (females). All runners were injury free, in regular training, and completed a health questionnaire before providing informed consent and participating in this study, which was approved by the Loughborough University Ethics Approvals (Human Participants) Sub-Committee.

Experimental design

Participants visited the laboratory for two sessions, which were 1–2 wk apart: first for preliminary assessments and familiarization and second for a main measurement session. Participants were asked to avoid severe-intensity exercise (i.e., running above critical speed) the day before both testing sessions. All running was performed on an instrumented treadmill (3DI, Treadmetrix, Park City, UT) at a 1% gradient to simulate the higher energetic cost of outdoor running (24). The preliminary session consisted of a submaximal step test and a V˙O2max test. In between the two running tests, there was a 20-min rest break, during which 45 anthropometric measurements comprising lengths, widths, and perimeters of body segments were taken from which subject-specific inertia parameters were calculated (25) for use in the inverse dynamics analysis. During the submaximal step test and V˙O2max test, recordings of respiratory gases and blood lactate were made to determine oxygen cost, V˙O2max, and the running velocity at lactate turn point (LTP). Finally, after the running tests, for familiarization, participants performed maximum voluntary isometric contractions (MVIC) of four different muscle groups (plantarflexors, knee flexors, knee extensors, and hip extensors) with one leg.

The main measurement session involved skin preparation and placement of sEMG sensors on eight muscles of each leg (see below), before MVIC of the four muscle groups were performed by each leg unilaterally. sEMG was recorded during the MVIC to facilitate the normalization of the sEMG amplitude recorded during the subsequent HIRF. After a 20-min break, all participants performed a standardized warm-up of 10 min running at 60% of the speed of the HIRF, a 15-s practice run at the speed of the HIRF, and a 5-min stretching. The prescribed running speed of the HIRF was determined as the final speed achieved during the V˙O2max test (less 0.5 km·h−1 if participants completed <30 s at the final speed) as an individualized measure of running performance. Before the HIRF, the treadmill was set at the prescribed running speed for the duration of the run, and the runner was given a countdown to mounting the treadmill and starting the clock. During the HIRF, verbal encouragement was given, and total distance covered by other runners as well as real-time feedback of their own distance covered, calculated every 5 s based on their running speed on the treadmill, was shown for motivation. When the runner could no longer maintain the prescribed running speed, they dismounted the treadmill. GRF, sEMG, and three-dimensional whole-body kinematics were captured synchronously throughout the run with Vicon Nexus and analyzed over 10 strides of running at four time points: at the start (i.e., strides 6–15); the 10 strides nearest to 33% and 66% of HIRF duration; and at the end of the run (i.e., strides 15–6 from the end). The first five strides and the last five strides of the HIRF were disregarded to avoid any biomechanical interference associated with mounting and dismounting the treadmill at full speed. Postrun blood lactate was measured at 1-min intervals after the HIRF until it reached a peak as an index of the anaerobic metabolism and metabolic disturbance.

Submaximal step test and V˙O2max test

A submaximal step test on the treadmill was used to assess the running velocity at LTP. Ten minutes were given for warm-up at a self-selected speed. The submaximal step test started at 11 km·h−1 for females and 12 km·h−1 for males and comprised 3-min stages with 1 km·h−1 increments and ≤30 s between increments to take capillary blood lactate samples from the earlobe while the runner straddled the treadmill. These samples were analyzed immediately for lactate concentration (Biosen C-line; EKF, Cardiff, UK). The test was terminated when blood lactate exceeded 4 mmol·L−1 or LTP could be identified as an increase of blood lactate concentration of >1.0 mmol·L−1 from the previous stage. Before the V˙O2max test, participants had another 5 min for warm-up at a self-selected speed. The V˙O2max test was started 2 km·h−1 lower than the final stage of the submaximal test. The running speed increased by 1 km·h−1 every minute until voluntary exhaustion. Verbal encouragement was given throughout the whole run.

Pulmonary gas exchange

Respiratory gas exchange was measured breath by breath during the submaximal step test and V˙O2max test via an automated open circuit metabolic cart (Vyntus; CareFusion, Basingstoke, UK). Participants breathed through a low-dead space mask (V2 Mask; Hans Rudolph Inc., Shawnee, KS). Before the test, both gas sensors were calibrated with gases of known concentration (16% O2, 5% CO2) and ambient air. A 3-L syringe was used to calibrate ventilatory volume (CareFusion, Yorba Linda, CA). Oxygen consumption was averaged over 30 s, and V˙O2max was expressed as the highest 30-s period in milliliters per kilogram per minute (26).

Isometric dynamometry for maximum voluntary contractions

MVIC were performed with four custom-built dynamometers: calibrated S-beam strain gauges (Force Logic, Reading, UK) for the assessment of knee extension, knee flexion, and hip extension, and a force plate for the plantarflexion measurements (Kistler, Hook, UK). During all MVIC, participants were firmly secured with extensive straps to minimize extraneous movement. Joint angles were selected to be within the range of motion (ROM) during running and close to the angle of peak torque. Angles are expressed relative to anatomical position = 0° with positive rotations defined as ankle dorsiflexion, knee extension, and hip flexion. Plantarflexion dynamometry involved participants seated with their trunk upright (vertical), the hip flexed at ~80° (i.e., the leg was oriented ~10° below horizontal), the knee fully extended, and the ankle joint at 0° with the foot flat on the force plate, which was perpendicular to the leg and reclined at 10° from the vertical. The seat was precisely positioned and secured with inextensible straps such that minimal plantarflexion movement was possible even during MVIC. For knee flexion, the participant lay prone with hip and knee angles of 40° and −30°, respectively. For knee extension, the runner was seated with hip and knee angles of 60° and −65°, respectively. For both knee extension and flexion, an ankle cuff, which was in series with a calibrated S-beam strain gauge positioned perpendicular to the tibia, was attached 5 cm proximal to the medial malleolus. For hip extension, the participants lay with their upper body on a horizontal bench in a prone position and the uninvolved leg supported. The knee was fully extended, and the hip flexed at 45°. A thigh brace and strap that were in series with the strain gauge were attached just proximal to the knee of the involved leg and positioned perpendicular to the thigh.

Measurements (force and sEMG) with each dynamometer involved the same sequence of warm-up contractions (50% [3×], 75% [3×], and 90% [1×] of maximum perceived effort), followed by three MVIC. For the MVIC, participants were instructed to extend or flex the relevant joint as hard as possible for 3 s and rest for 1 min between MVIC (27). Rest breaks between legs were at least 2 min and breaks between muscle groups a minimum of 5 min. During each series of MVIC, a force–time curve with a horizontal cursor indicating peak force was used for biofeedback, and verbal encouragement was given for each trial. The force signal was amplified (×1000), sampled at 2000 Hz with an analog–digital converter (Micro 1401-3; CED, Cambridge, UK), and displayed and recorded with Spike 2 software (CED).


sEMG recordings from the following eight muscles of both legs (i.e., 16 sensors in all) were recorded with wireless sensors (Trigno Wireless EMG; Delsys, Natick, MA): gluteus maximus (GM), rectus femoris (RF), vastus lateralis (VL), vastus medialis (VM), lateral hamstrings (LH), medial hamstrings (MH), gastrocnemius lateralis (LG), and soleus (S). Segment length for thigh was defined as the distance between the greater trochanter and the lateral epicondyle of the knee, and the segment length for the shank was defined as the distance between the lateral epicondyle of the knee and the lateral malleolus. Muscles were palpated and outlined to position the sensor over the center (mediolateral) of the superficial muscle belly, as well as at specific distances along the length of the segment: 55%, 60%, and 35% of thigh length from superior border of the patella for RF, VL, and VM, respectively; 45% of thigh length from popliteal fossa for LH and MH; 65% of shank length from the medial knee joint space for S; and 85% of shank length from the calcaneus for LG. For GM, 50% of the distance between the second sacral vertebrae to the greater trochanter was used. These locations were used to avoid the innervation zones and to maximize signal fidelity (28). The skin of the participants at these sites was shaved, abraded, and cleaned with alcohol before the sensors were attached with adhesive interfaces and secured with tape. EMG signals were amplified at source (×300; 20- to 450-Hz bandwidth) before further amplification (overall effective gain, ×909) and sampled at 2000 Hz with Vicon Nexus.

Motion capture

Twelve infrared cameras were positioned around the treadmill (Vicon, Oxford, UK), and the system was calibrated with a 5-marker reflective wand. Dynamic calibration was done for 2000 refinement frames before the wand was placed in the middle of the treadmill to define the origin and global coordinate system. Forty-seven retroreflective markers were attached to the participant on specific bony landmarks (29) by a single assessor. In addition, clusters of four markers were attached to the thigh and the shank of each leg. Four markers placed on the treadmill load cells were used to define the plane of the treadmill surface and the coordinate system of the force plate relative to the global coordinate system. Before the HIRF, a static trial was collected for 3 s with the treadmill level while participants were standing upright, feet shoulder width apart, facing forward with their arms outstretched anteriorly. All motion data were captured at 250 Hz. During the HIRF, data were recorded for 15-s intervals, with 15-s gaps, beginning at the start of the run, until 120–150 s into the run when data were recorded continuously until exhaustion.


Forces in mediolateral, anterior–posterior, and vertical directions were recorded with a force plate using four force transducers (MC3A; AMTI, Watertown, MA) embedded in the treadmill. Synchronized force was recorded at 1000 Hz within the Vicon motion analysis system. Afterward, the recorded voltages were processed in Matlab, multiplying the raw voltages with the inverted sensitivity matrix of the force plate and summing forces from the four transducers (MathWorks, Cambridge, UK). Finally, GRF, center of pressure, and free moment were calculated.

Data processing

Reconstructed 3D retroreflected marker locations were labeled, and gaps of up to 30 frames were filled using cubic spline interpolation to avoid errors in the trajectories because of the interpolation of large gaps (Vicon). For larger gaps, marker trajectories of three visible markers on the same rigid segment were used to predict the missing marker. Marker and force plate data were low-pass filtered with a second-order Butterworth filter with a cutoff frequency of 12 Hz based on residual analysis of marker positions. Using the same cutoff frequency for motion and force data has previously been shown to provide more accurate joint moments using inverse dynamics (30,31).

A model of the human body was developed using Visual3D (C-Motion, Germantown, MD) comprising 15 rigid segments linked by joints with three rotational, but no translational, degrees of freedom. Joint angles were determined by fitting the model to marker trajectories using inverse kinematics, and lower limb joint moments were calculated by inverse dynamics. IC and toe-off (TO) for each ground contact were defined as rising above/falling below a threshold of 40 N of vertical GRF. Steps and strides were defined from successive IC of contralateral legs and the same leg, respectively.

Step frequency, GCT (IC to TO of the same leg), flight time (TO to IC of the contralateral leg), and swing time (TO to IC of the same leg) were calculated. Step length was measured by dividing treadmill speed by step frequency. Sagittal plane lower limb joint angles at discrete points, such as peaks and IC and TO, were determined during stance and swing. Ankle dorsiflexion, knee extension, and hip flexion were defined as positive. Joint power was derived by multiplying each of the joint moments by the joint angular velocity. Power time histories in the direction of each sagittal plane moment (flexion/extension) about each joint (ankle, knee, and hip) were then numerically integrated to calculate positive and negative work during stance and swing phases. All variables were averaged across both legs for 10 strides. Kinematic and kinetic data from two participants were discarded because of retroreflective marker loss during the run.

For the analysis of muscle activation, filtered sEMG recordings at each of the four time points of the HIRF were analyzed over 10 strides for root-mean-square (RMS) amplitude throughout five different phases of the stride cycle (Fig. 1): braking (from IC to zero-crossing of horizontal GRF), propulsion (from the end of the braking phase to TO), early swing (from end of propulsion to IC of contralateral leg), contralateral stance (mid swing; from end of early swing to TO of contralateral leg), and late swing (from end of contralateral stance to IC of ipsilateral leg). sEMG during MVIC was analyzed with a 500-ms RMS epoch (250 ms either side of maximum voluntary force; EMGMVF) and used to normalize sEMG measurements during the HIRF. Specifically, normalized sEMG was calculated as the ratio between RMS amplitude during the phases of running divided by EMGMVF. Normalized sEMG data were averaged across the left and right legs. Poor quality recordings that did not show clear phases of activation or had implausibly large amplitudes were discarded. If recordings from only one leg were affected, signals from the opposite leg were retained and used for the analysis.

Schematic showing the division of the EMG data during one gait cycle into five phases, based on GRF data. GRF data are illustrative, whereas EMG data were recorded from one participant. EMG phases were as follows: braking from IC to zero anterior–posterior GRF, propulsion ending with TO, early swing ending with IC of the contralateral foot, contralateral stance ending with TO of the contralateral foot, and late swing ending with IC of the ipsilateral foot.


During pilot testing, we assessed the reliability of our key kinetic and kinematic variables for three repeated measurements with our standard period of 10 strides (i.e., three consecutive periods of 10 strides) at a sustainable running speed of 16 km·h−1 with 10 male runners. A summary of the reliability of some key variables is presented to provide context to the size of changes with fatigue in the current study (see Table, Supplemental Digital Content 1, which shows reliability measures, The three blocks of 10 strides were compared using the intraclass correlation coefficient using a two-way fixed model as recommended previously (32), and the minimal detectable change (MDC) was calculated (32). The significance from the repeated-measures ANOVA is reported to indicate that there were no differences between the three periods of 10 strides. The within-subject coefficient of variation (CVws) was calculated using the RMS approach (33). All variables showed strong agreement with an intraclass correlation coefficient ≥0.96. The CVws was lower than 3% for all variables shown in the supplementary material apart from ankle angle at IC that showed a CVws 19.41%, likely because of the low mean value at this instant (~1.7°).

Statistical analysis

Individual data were plotted for all participants over time as a visual check for outliers and only EMG data plots revealed substantial outliers. Individual EMG time histories that were large outliers (three SD away from the mean) were excluded from subsequent statistical analysis. A mixed ANOVA confirmed that there were no interaction effects between time and sex for all variables, and thus male and female data were subsequently considered together. Shapiro–Wilks test was used to assess normality. Almost all of the kinematic and kinetic variables were normally distributed; thus, parametric tests were performed on all variables. A one-way repeated-measures ANOVA was used to establish within-participant differences between the four measured time points during the run for kinematic, kinetic, and EMG variables. If the assumption of sphericity was violated, values were corrected using the Greenhouse–Geisser correction. In case of a significant main effect of time, Bonferroni post hoc tests were used to identify pairwise differences between two specific time points. Partial eta squared (η2) was calculated for main time effects as the ratio of the variance explained by the effect of time divided by the total variance with 0.01 considered a small effect size (ES), 0.06 a medium ES, and 0.14 a large ES (34). All statistical analysis procedures were performed with IBM SPSS Statistics Version 21 (IBM Corp., New York, NY). Data are presented as mean ± SD, and differences were considered to be significant if P < 0.05.


The athletes speed at LTP was 16.0 ± 0.9 km·h−1, and V˙O2max was 68.5 ± 6.0 mL·kg−1·min−1 for all participants. Participants ran the HIRF at 20.6 ± 1.5 km·h−1 (i.e., on average 4.6 km·h−1 above the speed at LTP and thus demonstrably unsustainable for all individuals), leading to exhaustion and termination of the HIRF in 180.8 ± 42.1 s with peak post blood lactate concentrations of 12.6 ± 2.6 mmol·L−1. Exact measurement time points for the 10 analyzed strides intended to be at 33% and 66% of the run were at 34.9% ± 3.1% and 66.0% ± 2.1% of the HIRF, respectively.

Stride parameters

GCT showed a modest increase from the start to 66% of the HIRF, with an overall increase of 0.006 s (ANOVA P < 0.001; ES = 0.58; Table 1). Concomitantly, flight time decreased during the HIRF by 0.005 s (ANOVA P = 0.035; ES = 0.20), but without any post hoc differences between specific time points. Step length, step frequency, and swing time did not change throughout the HIRF (ANOVA P = 0.449, ES = 0.05; P = 0.41, ES = 0.06; and P = 0.422, ES = 0.06, respectively).

TABLE 1 - Stride parameters and sagittal joint angles at the ankle, knee, and hip at the start, 33%, 66%, and end of the HIRF.
Variable Start 33% 66% End ƞƞ 2
Stride parameters
 Step length (m) 1.75 ± 0.16 1.76 ± 0.16 1.76 ± 0.16 1.75 ± 0.16 0.05
 Step rate (steps per minute) 197 ± 10 195 ± 11 195 ± 10 196 ± 10 0.06
 Step time (s) 0.306 ± 0.016 0.309 ± 0.017 0.308 ± 0.016 0.307 ± 0.016 0.06
 GCT (s)** 0.149 ± 0.013 0.153 ± 0.014a 0.155 ± 0.015 a,b 0.155 ± 0.015 a 0.58
 Flight time (s)* 0.157 ± 0.012 0.156 ± 0.012 0.154 ± 0.012 0.152 ± 0.013 0.20
 Swing time (s) 0.463 ± 0.025 0.464 ± 0.026 0.462 ± 0.024 0.460 ± 0.024 0.06
 COM minimum height stance (m)** 0.933 ± 0.064 0.932 ± 0.065 0.930 ± 0.063 0.925 ± 0.064 a,b,c 0.42
Joint angles
 Peak trunk forward lean (°) −6.1 ± 4.8 −6.2 ± 5.5 −6.4 ± 5.3 −7.3 ± 5.4 0.13
 Angle at IC (°)* −2.1 ± 6.8 −1.0 ± 7.2 −0.5 ± 7.2b −0.1 ± 6.5 0.26
 Angle at TO (°) −11.8 ± 5.5 −12.2 ± 5.5 −11.5 ± 5.2 −11.2 ± 5.3 0.08
 Peak angle (°) 24.2 ± 3.5 23.9 ± 3.8 24.5 ± 3.8 24.6 ± 3.7 0.06
 ROM stance (°) 36.1 ± 4.4 36.3 ± 5.2 35.7 ± 5.2 35.3 ± 4.9 0.09
 Angle at IC (°)** −24.2 ± 2.8 −25.1 ± 2.3 −25.6 ± 2.1a −26.3 ± 2.5 a,b,c 0.43
 Angle at TO (°) −17.9 ± 3.5 −17.1 ± 3.6 −17.6 ± 3.7 −17.8 ± 3.6 0.10
 Peak flexion angle stance (°)** −45.2 ± 4.5 −46.0 ± 4.7 −47.0 ± 4.6 a −47.4 ± 4.6 a 0.31
 Peak flexion angle swing (°) −132 ± 8 −132 ± 7 −131 ± 7 −131 ± 8 0.04
 ROM stance (°) 21.0 ± 3.2 20.9 ± 3.5 21.0 ± 3.9 20.6 ± 3.8 0.04
 ROM swing (°) 118 ± 10 119 ± 9 118 ± 9 118 ± 9 0.03
 Angle at IC (°)** 41.1 ± 6.6 42.8 ± 5.7 a 43.6 ± 5.5 a 44.7 ± 5.7 a,b,c 0.53
 Angle at TO (°)** −3.2 ± 5.5 −2.5 ± 5.1 −2.3 ± 5.1 −1.4 ± 4.6 0.25
 ROM hip stance (°)* 44.4 ± 5.1 45.4 ± 4.5 46.0 ± 4.7 46.3 ± 4.6 0.24
 ROM hip swing (°)** 81.9 ± 5.6 84.7 ± 5.5 a 85.5 ± 5.7 a,b 86.8 ± 6.5 a,b,c 0.59
Values are displayed as mean ± SD (n = 16).
Significant repeated-measures ANOVA main effect of time: *P < 0.05; **P < 0.01.
Significant Bonferroni post hoc tests (P < 0.05): adifferent from start of the run; bdifferent from 33% of the run; cdifferent from 66% of the run.
COM, center of mass.


Throughout the HIRF (i.e., significantly with each successive time point), ankle, knee, and hip were less extended at IC (e.g., from the start to the end of the HIRF: ankle 1.9° more dorsiflexed ANOVA P = 0.022, ES = 0.26; knee 2.1° more flexed ANOVA P < 0.001, ES = 0.43; hip 3.6° more flexed P < 0.001, ES = 0.53; Table 1, Fig. 2). Peak knee flexion angle during stance increased from the start to 66% and the end of the HIRF (ANOVA P = 0.008, ES = 0.31). Sagittal plane hip joint ROM increased during swing throughout the HIRF (ANOVA P < 0.0001, ES = 0.59). The hip was also more flexed at TO, and the hip ROM during stance increased during the HIRF (ANOVA P = 0.004, ES = 0.25, and P = 0.022, ES = 0.24, respectively). However, post hoc testing did not reveal any differences between specific time points in the hip joint ROM during stance and hip angle at TO.

Ankle joint angle, moment, and power throughout the gait cycle at the start 33%, 66%, and end of the HIRF. Data are presented as group mean ± SD (n = 16) shown as colored lines and shaded areas, based on the average of 10 strides of each runner at each time point. Vertical lines represent TO.

Peak joint angles during swing all occurred later at the end of the HIRF than the start. The instances of peak hip extension and flexion during swing increased from 33.2% ± 0.04% to 36.3% ± 0.05% (ANOVA P < 0.001, ES = 0.51) and from 78.4% ± 0.01% to 79.8% ± 0.02% of the stride (ANOVA P < 0.001, ES = 0.59), respectively. From the start to the end of the HIRF, the swing timing of peak knee flexion and extension during late swing increased from 60.5% ± 0.01% to 61.8% ± 0.01% and from 94.2% ± 0.1% to 95.2% ± 0.1% of the stride cycle (both ANOVA P < 0.001, ES ≤ 0.63), respectively. Finally, the point of peak ankle dorsiflexion before foot contact increased from 83.6% ± 0.08% to 88.3% ± 0.06% of the stride cycle from the start to the end of the HIRF (ANOVA P = 0.002, ES = 0.40).


Peak vertical GRF progressively decreased throughout the HIRF, such that from the start to the end, it decreased by 0.19 N·kg−1 (−6.1%; ANOVA P < 0.001, ES = 0.69; Table 2) and vertical impulse decreased by 3.9 Ns (−4.1%; ANOVA P = 0.004, ES = 0.30). Peak ankle plantarflexion moment also decreased progressively throughout the HIRF by 0.4 N·m·kg−1 (−9.0%; ANOVA P < 0.001, ES = 0.8; Fig. 2). By the end of the HIRF, compared with the start, positive and negative plantarflexion work done during stance displayed more substantial decreases of 0.19 J·kg−1 (−13.9%; ANOVA P < 0.001, ES = 0.69; Fig. 3) and 0.18 J·kg−1 (−14.7%; ANOVA P < 0.001, ES = 0.71), respectively. Peak knee extension moment during stance increased from the start to the end of the HIRF by 0.24 N·m·kg−1 (+10.3%; ANOVA P < 0.001, ES = 0.44) and was accompanied by larger increases in the positive and negative knee extension work done during stance, 0.09 J·kg−1 (+33.3%; ANOVA P = 0.001, ES = 0.57; Fig. 3), and 0.07 J·kg−1 (+25.0%; ANOVA P < 0.001, ES = 0.50), respectively. Several indices of hip kinetics during swing showed main effects of time and large ES with monotonic increases (All ANOVA P ≤ 0.04, ES = 0.23–0.33), but there were no post hoc differences between the start and the end of the HIRF (peak hip flexion and extension moment swing, negative hip flexion and extension work swing), or a tendency for a difference between start and end (positive hip extension work swing, P = 0.092).

TABLE 2 - Kinetic parameters at the start, 33%, 66%, and end of the HIRF.
Variable Start 33% 66% End ƞ 2
 Vertical peak GRF (N·BW−1)** 3.09 ± 0.21 3.03 ± 0.21 2.97 ± 0.20 a,b 2.90 ± 0.20 a,b,c 0.69
 Vertical impulse (N·s)** 94.8 ± 16.3 94.1 ± 15.7 92.6 ± 15.1 90.9 ± 14.5 a,b 0.30
 Peak braking GRF (N·BW−1) −0.49 ± 0.06 −0.50 ± 0.07 −0.50 ± 0.07 −0.49 ± 0.07 0.05
 Peak propulsion GRF (N·BW−1) 0.52 ± 0.05 0.52 ± 0.05 0.51 ± 0.05 0.51 ± 0.05 0.17
 Horizontal braking impulse (N·s) −16.0 ± 3.27 −16.3 ± 3.31 −16.3 ± 3.23 −16.4 ± 3.08 0.10
 Horizontal propulsion impulse (N·s) 14.4 ± 3.07 14.6 ± 2.98 14.6 ± 2.87 14.6 ± 2.71 0.19
 Peak plantarflexion moment stance (N·m·kg−1)** −4.45 ± 0.64 −4.29 ± 0.62 a −4.16 ± 0.59 a,b −4.05 ± 0.58 a,b,c 0.80
 Negative plantarflexion work stance (J·kg−1)** −1.22 ± 0.27 −1.13 ± 0.26 a −1.09 ± 0.24 a,b −1.04 ± 0.22 a,b,c 0.71
 Positive plantarflexion work stance (J·kg−1)** 1.37 ± 0.27 1.28 ± 0.27 a 1.22 ± 0.25 a,b 1.18 ± 0.26 a,b 0.69
 Peak ext. moment stance (N·m·kg−1)** 2.33 ± 0.62 2.46 ± 0.69 2.53 ± 0.68 a 2.57 ± 0.67a 0.44
 Peak ext. moment swing (N·m·kg−1) 0.64 ± 0.15 0.65 ± 0.14 0.65 ± 0.15 0.65 ± 0.17 0.04
 Peak flex. moment swing (N·m·kg−1)** −1.50 ± 0.22 −1.54 ± 0.22 −1.55 ± 0.21 −1.57 ± 0.24 0.31
 Negative ext. work stance (J·kg−1)** −0.28 ± 0.09 −0.32 ± 0.11a −0.33 ± 0.12 a −0.35 ± 0.11 a,b 0.50
 Positive ext. work stance (J·kg−1)** 0.27 ± 0.11 0.31 ± 0.13a 0.34 ± 0.14 a,b 0.36 ± 0.15 a,b 0.57
 Negative ext. work swing (J·kg−1) −0.79 ± 0.18 −0.80 ± 0.17 −0.80 ± 0.18 −0.80 ± 0.19 0.08
 Negative flex. work swing (J·kg−1) −1.20 ± 0.17 −1.21 ± 0.16 −1.21 ± 0.16 −1.23 ± 0.18 0.08
 Peak ext. moment stance (N·m·kg−1) −2.47 ± 0.53 −2.55 ± 0.50 −2.63 ± 0.46 −2.65 ± 0.47 0.17
 Peak flex. moment swing (N·m·kg−1)** 2.94 ± 0.48 3.06 ± 0.48 3.06 ± 0.49a 3.09 ± 0.52 0.33
 Peak ext. moment swing (N·m·kg−1)* −2.87 ± 0.38 −2.94 ± 0.39 −2.97 ± 0.37 −3.04 ± 0.40 0.28
 Positive ext. work stance (J·kg−1) 0.32 ± 0.10 0.31 ± 0.07 0.29 ± 0.07 0.29 ± 0.08 0.12
 Negative flex. work stance (J·kg−1) −0.34 ± 0.10 −0.36 ± 0.10 −0.37 ± 0.10 −0.35 ± 0.10 0.07
 Negative flex. work swing (J·kg−1)** −0.13 ± 0.06 −0.16 ± 0.07 −0.18 ± 0.07 a −0.21 ± 0.10 0.31
 Positive flex. work swing (J·kg−1) 0.84 ± 0.16 0.80 ± 0.16 0.81 ± 0.15 0.83 ± 0.15 0.11
 Negative ext. work swing (J·kg−1)* −0.17 ± 0.07 −0.19 ± 0.08 −0.19 ± 0.08 −0.20 ± 0.10 0.23
 Positive ext. work swing (J·kg−1)* 1.30 ± 0.31 1.39 ± 0.31 1.40 ± 0.34 1.44 ± 0.37 0.29
Values are displayed as mean ± SD (n = 16).
Significant repeated-measures ANOVA main effect of time: *P < 0.05; **P < 0.01.
Significant Bonferroni post hoc tests (P < 0.05): adifferent from start of the run; bdifferent from 33% of the run; cdifferent from 66% of the run.
ext., extension; flex., flexion.

Changes in positive work during the stance phase of the HIRF, for ankle plantarflexion, knee extension, and hip extension. Measurements at later time points (33%, 66%, and end) are each compared with the start of the HIRF. Data are presented as mean change ± SD.

Muscle activations

The majority of the changes in muscle activation occurred during phases with relatively low levels of activation (<35% EMGMVF; Fig. 4). This included increases in activation from the start/33% to the end of the HIRF for two muscles during propulsion (VM and VL +64.5%–78.0%; both ANOVA P ≤ 0.009, ES = 0.25–0.31), four muscles during the early swing phase (VM, VL, GM, S + 36.1%–50.7%; all ANOVA P ≤ 0.015, ES = 0.25–0.36), three muscles during contralateral stance/mid swing of the ipsilateral leg (all ANOVA; VM +48.7%, P = 0.017, ES = 0.30; LH +27.2%, P = 0.008, ES = 0.27; and RF +75.8%, P = 0.003, ES = 0.52), and one muscle during late swing (RF +10.2%; ANOVA P = 0.007, ES = 0.33). The only changes in high levels of activation (>50% EMGMVF) throughout, and at the end of, the HIRF were for the hip extensors during late swing (LH +24.5%, ANOVA P < 0.001, ES = 0.68, MH +26.4%, ANOVA P = 0.001, ES = 0.44; and GM +37.6%, ANOVA P = 0.001, ES = 0.36) and quadriceps muscles during late swing (VL +16.5%, ANOVA P < 0.001, ES = 0.39; and VM +10.8%, ANOVA P = 0.002, ES = 0.35). Finally, activation during the braking phase of stance did not change between the start and the end of the HIRF for any muscle.

RMS amplitude for all eight muscles during the five different phases of the stride cycle at the start 33%, 66%, and end of the HIRF. Data are presented as mean ± SD (n = 10–18) based on 10 strides of each runner at each time point. aSignificantly different from start of the run (P < 0.05). bSignificantly different from 33% of the run (P < 0.05). cSignificantly different from 66% of the run (P < 0.05)).


Due to the unknown nature of the kinematic, kinetic, and muscle activation changes with unsustainable high-intensity running, the aim of the study was to identify the neuromechanical changes during the HIRF. In accordance with the study hypotheses, during the HIRF (i.e., end vs start), we found a subtle increase in GCT (+4%) and decrease in flight time (+3%), as well as a decrease in peak vertical GRF (−6%) concomitant with a decrease in plantarflexion peak moment (−9%) and positive and negative work (−14% and −15%, respectively). These changes in plantarflexion kinetics likely indicate the fatigue of the plantarflexor muscle group. By contrast, positive knee extension work done increased by 33%. Kinematically, ankle, knee, and hip joints were more flexed at IC at the end of the HIRF than the beginning, which may in part be a consequence of the progressively later timing of peak hip flexion (78.4% to 79.8% of stride cycle), knee extension (94.2% to 95.2% of stride cycle), and ankle dorsiflexion (83.6% to 88.3% of stride cycle) during late swing reducing the time to prepare for IC. Hip extensor RMS EMG during this late swing phase was also elevated at the end of the HIRF compared with the beginning (late swing LH, MH, GM +21%–37%), although there was no change in plantarflexor EMG during any phase of the stride cycle.

The neuromechanical data at the start of the HIRF within the current study were broadly similar to previous research at similar or somewhat slower running speeds, including stride parameters (e.g., GCT 145 vs 140 ms [35], step length 1.75 vs 1.71 m [20]), joint kinematics (36), and peak vertical GRF (3.09 vs ~3.3 N·kg−1 [37]). The peak extensor moments during stance in the current study were similar to previous reports for the knee (2.57 vs 2.74 N·m·kg−1 [37]) and hip (2.65 vs 2.95 N·m·kg−1 [6]) and slightly larger for the ankle (4.45 vs 3.55–3.77 N·m·kg−1 overground [6] vs 4.01 N·m·kg−1 on a treadmill at 3.8 m·s−1 [38]). EMG data also showed broadly similar activation patterns to those found in other running studies (39). Thus, the values generated within the current study were in the expected range, and this gives confidence in the veracity of the experiment.

Stride parameters

In this study, the changes in stride parameters were relatively modest, when considering the MDC, with GCT increasing by 0.006 s while flight time decreased by 0.005 s, giving runners slightly less time to swing their leg and prepare for ground contact. However, the decrease in flight time was slightly below, and the change in GCT was exactly the same as the MDC. The increase in GCT with fatigue has been widely reported after prolonged running at lower speeds (e.g., 80%–95% of V˙O2max [4,15,16,40]). The relatively modest changes in stride parameters we have observed despite pronounced fatigue and ultimately task failure may reflect the highly constrained spatiotemporal nature of the task of high-intensity running at a fixed speed, i.e., with little flexibility available in successfully completing the task. Nonetheless, with the increase in GCT in the current study, we found a concomitant decrease in peak vertical GRF by 0.19 N·kg−1 (−6.1%) and vertical impulse by 3.9 Ns (−4.1%), but no change in horizontal impulse. Despite the reduced ability to apply high forces to the ground with fatigue, it seems likely that runners try to maintain total vertical impulse during stance phase (4), and this necessitates a longer contact time. Nevertheless, the decrease in vertical impulse that does occur would appear to explain the reduction in flight time.

Joint angles

There were subtle kinematic changes at the end compared with the start of the HIRF with all three major lower limb joints being more flexed at IC (ankle 1.9°, knee 2.1°, and hip 3.6°). The increase in knee flexion angle at IC appears to persist throughout stance leading to a more flexed knee during midstance (i.e., peak knee flexion 2.2°) at the end of the HIRF. These more flexed knee angles (at IC and midstance) are consistent with previous research showing increased flexion during more prolonged running to fatigue (13–16). The increased knee flexion angle during stance is likely associated with the decreased peak vertical GRF, which has been shown to be related to “Groucho” running characterized by low center of mass height and greater knee flexion (41). However, higher knee flexion during stance is potentially detrimental to performance because it is associated with less economic running (42).

The increase in knee flexion at IC could be an attempt to obtain the same horizontal landing position of the foot in relation to the hip joint and reduce braking forces as a consequence of increased hip flexion at IC (+3.6°). Higher knee flexion at IC also facilitates more effective shock attenuation (i.e., reduced acceleration of axial segments [43]), and thus this strategy of increased knee flexion may reduce the effect and the risk of injury when fatigued. In addition, the decrease in flight time and the later occurrence of key instances throughout the gait cycle (e.g., peak hip extension and peak knee flexion), especially during late swing phase (e.g., later peaks in ankle dorsiflexion [83.6% to 88.3%], knee extension [94.2% to 95.2%], and hip flexion [78.4% to 79.8%]), likely provide less time to prepare for ground contact and extend the knee fully before IC.

Joint moments and work

From the start to the end of the HIRF, plantarflexor moment during stance decreased by 0.4 N·m·kg−1 (−9.0%), and positive (−0.19 J·kg−1, −13.9%) and negative (−0.18 J·kg−1, −14.7%) plantarflexion work during stance was also reduced. All these changes were more than three times the MDC within a run. These reductions in ankle kinetic outputs in the direction of plantarflexion were presumably a consequence of the fatigue of the plantarflexor muscle group because of forceful and repeated stretch-shortening cycle (eccentric–concentric) contractions during the HIRF. An alternative perhaps less likely possibility is a substantial increase in dorsiflexor coactivation, reducing net plantarflexion moments and work. Unfortunately, because of the limited number of EMG sensors in this study, we were unable to assess dorsiflexor activation; however, a previous study found no changes in tibialis anterior activation with fatigue during a run to exhaustion at vV˙O2max (23). In the current study, the magnitude of positive and negative plantarflexion work was very similar throughout the HIRF and showed similar changes (start to end, −0.19 and −0.18 J·kg−1). During the stretch-shortening cycle contractions performed by the plantarflexors during high-intensity running, fatigue would be expected to impair muscle force production during the initial eccentric phase of contraction and, thus, also lead to reduced tension and energy stored by the Achilles tendon. As energy stored by the Achilles tendon during the eccentric phase of contraction has been shown to contribute more than 50% of the subsequent positive (concentric) plantarflexion work (41), this might explain the apparent coupling and similar decrements in negative and positive work.

Moreover, the plantarflexors have been found to be operating at a higher proportion of their maximum muscle forces than the knee extensors during unfatigued running and sprinting (21), and to be responsible for around 75% of the total vertical support impulse over a wide range of running speeds (3.5–9.0 m·s−1 [20]). Similarly, in the current study, the plantarflexors produced ~70% of the total positive work generated by the joints of the leg during stance at the beginning of the HIRF, and thus much more than the knee (~14%) and hip (~16%) muscles. The reductions in ankle plantarflexion moment and positive and negative work that we have observed during the HIRF are consistent with this disproportionately high contribution of the plantarflexors to running. The decreased plantarflexion kinetics and the presumed fatigue of the plantarflexors are also likely the main reason for the observed reduction in peak vertical GRF. The change in positive work at the ankle, knee, and hip between time points showed relatively large SD likely because of additive errors across two time points and the magnitude of change data being much smaller than the values at any specific time point. Nonetheless, in contrast to the ankle, positive knee extension work increased by 0.09 J·kg−1, and thus a shift of positive work from the ankle to the knee occurred with fatigue. Furthermore, although the plantarflexor moment decreased by 0.4 N·m·kg−1, the knee extensor moment increased by 0.24 N·m·kg−1 during stance. In addition, it is also likely that positive hip extension work during swing increased as a consequence of fatigue (start–end +10%; large ES, ~4-fold MDC), which was consistent with our finding of increased hip extensor (MH, LH, and GM) EMG during the late swing phase. However, this was not conclusive as the post hoc test showed only a tendency for a difference (P = 0.09). A shift of work from distal to proximal muscle groups was also shown in a study investigating fatigue during treadmill running at 10 km pace (19), and we have found the same mechanism seems to occur at faster running speeds.

Muscle activations

The HIRF led to a consistent increase in hip extensor (LH, MH, and GM) activation during late swing from an already high level of activation, >70% EMGMVF at the start. The hip extensors play an important role in accelerating the leg at the end of swing, particularly at high running speeds (44). In the current study, flight times decreased concomitantly with an increased hip ROM, whereas importantly peak hip flexion happened later in the gait cycle. These results indicate that less time is available to reverse the hip flexion angular momentum and begin to extend the hip and prepare for IC. The mechanical changes at the end of the HIRF seem to necessitate greater hip extensor activation. Consistent with this observation, it has previously been shown that hamstrings absorb more energy during late swing with shorter flight times as they are required to accelerate the leg more quickly (20). RF activation increased during early swing and contralateral stance. Because peak hip extension after TO occurred later in the gait cycle and flight time decreased, hip flexion needs to happen more quickly to prepare for the next ground contact. RF flexes the hip in the swing phase, contributing to swinging the leg faster (45). Simultaneously, the hip flexion moment during swing increased, indicating that RF activity increased to facilitate quicker hip flexion.

There were also increases in muscle activation in periods of relatively low activation (<30% EMGMVF) during propulsion, early swing and midswing (contralateral stance) phases, but not the braking phase of running. The muscle activation of the plantarflexors, assessed with EMG, did not increase despite decreases in plantarflexion moment and positive and negative work. An increased muscle activation because of the recruitment of additional motor units has been found to occur with fatigue (22), assuming initially submaximum activation. The activation of the plantarflexors was highest during the braking phase of running (LG >150% EMGMVF; S > 200% EMGMVF), showing the highest normalized values for any muscle during this study, and was also high during the propulsion phase (LG ~80% EMGMVF; S ~ 100% EMGMVF). Therefore, it is likely that plantarflexion activation was maximal, or near maximal, at the start of the HIRF, limiting the scope for increases in activation even as the force-generating capacity of these muscles declined. Another study investigating changes in muscle activation in running during an incremental step test also found no changes in plantarflexor muscle activation (46), potentially supporting near maximal activation at moderate to high speeds. Alternatively, the high velocity and the high impact nature of high-intensity running may result in a relatively low signal-to-noise ratio of EMG recordings, especially for the plantarflexors that are close to the ground, and thus mask the effect of fatigue. However, the reduction in work done and moment produced at the ankle suggested that the plantarflexors are the main limitation in high-intensity running.


There are several limitations of the study methods. Performing the running trials on an instrumented treadmill enabled neuromechanics to be averaged over 10 strides (20-ft GCs) at each measured time point, facilitating more reliable measurements than would be possible during only one or two steps of overground running. Although the kinematics and kinetics of overground and treadmill running mainly follow the same movement pattern (38), there are some subtle differences, such as greater plantarflexion of the foot at IC (47) and higher peak ankle moments (38) on the treadmill. These authors concluded though that overground and treadmill running are comparable if the treadmill is sufficiently stiff (38). The treadmill used in the current study was a specialist instrumented scientific treadmill, qualitatively stiffer than many commercial/fitness treadmills. More important for this study is the effect of fatigue on overground and treadmill running. Although there are limited data comparing the fatigue effects of treadmill and overground running, existing data suggest that there is no difference in the effect of fatigue on peak plantar pressure between treadmill and overground running (48). This indicates that the difference in plantarflexion moment with fatigue is also likely to be the same in overground and treadmill running. The level of fatigue at the end of the HIRF may have been less severe than that experienced during competitive races because of the experiment being conducted outside the competitive season, lower motivation than in a competitive situation, and insufficient confidence to run to complete exhaustion on the treadmill. However, on average, the runners completed the HIRF at 4.6 km·h−1 above their speed at LTP, and thus the speed was demonstrably unsustainable for all individuals and resulted in mean post blood lactate concentrations of 12.6 ± 2.6 mmol·L−1, which are similar to blood lactate values after a 1500-m race (49), indicating substantial anaerobic metabolism and acute skeletal muscle fatigue. In the current study, because of the limited number of EMG sensors, not all the possible lower limb muscles could be measured. However, given our findings show reductions in plantarflexion kinetics, future studies should assess muscle activation from all three parts of the triceps surae as well as the dorsiflexors to more fully assess if changes in activation are involved in the reductions in plantarflexion kinetics we have observed.


Runners spent more time on the ground at the end of a HIRF, likely to counteract their reduced ability to apply high vertical peak forces, and hence impulses, to the ground. It appears that the ankle plantarflexors fatigued, with a reduction in peak plantarflexion moment and positive work, and in compensation work was redistributed to the knee joint. Improving the fatigue resistance of the plantarflexors to high-intensity stretch-shortening cycle activity might help athletes to maintain high running speeds for longer.

This work was in part financially supported by the English Institute of Sport.

None of the authors have any conflict of interest to disclose. The results of the present study do not constitute endorsement by the American College of Sports Medicine. The authors declare that the results of the present study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation.


1. Duffield R, Dawson B, Goodman C. Energy system contribution to 100-m and 200-m track running events. J Sci Med Sport. 2004;7(3):302–13.
2. Monod H, Scherrer J. The work capacity of a synergic muscular group. Ergonomics. 1965;8(3):329–38.
3. Enoka RM, Duchateau J. Muscle fatigue: what, why and how it influences muscle function. J Physiol. 2008;586(1):11–23.
4. Rabita G, Slawinski J, Girard O, Bignet F, Hausswirth C. Spring-mass behavior during exhaustive run at constant velocity in elite triathletes. Med Sci Sports Exerc. 2011;43(4):685–92.
5. Morin JB, Tomazin K, Edouard P, Millet GY. Changes in running mechanics and spring-mass behavior induced by a mountain ultra-marathon race. J Biomech. 2011;44(6):1104–7.
6. Schache AG, Blanch PD, Dorn TW, Brown NAT, Rosemond D, Pandy MG. Effect of running speed on lower limb joint kinetics. Med Sci Sports Exerc. 2011;43(7):1260–71.
7. Black MI, Jones AM, Blackwell JR, et al. Muscle metabolic and neuromuscular determinants of fatigue during cycling in different exercise intensity domains. J Appl Physiol. 2017;122(3):446–59.
8. Girard O, Millet GP, Micallef JP, Racinais S. Alteration in neuromuscular function after a 5 km running time trial. Eur J Appl Physiol. 2012;112(6):2323–30.
9. Nummela AT, Heath KA, Paavolainen LM, et al. Fatigue during a 5-km running time trial. Int J Sports Med. 2008;29(9):738–45.
10. Reenalda J, Maartens E, Homan L, Buurke J. Continuous three dimensional analysis of running mechanics during a marathon by means of inertial magnetic measurement units to objectify changes in running mechanics. J Biomech. 2016;49(14):3362–7.
11. Nilsson J, Thorstensson A, Halberstma J. Changes in leg movements and muscle activity with speed of locomotion and mode of progression in humans. Acta Physiol Scand. 1985;123(4):457–75.
12. Elliott BC, Roberts AD. A biomechanical evaluation of the role of fatigue in middle-distance running. Can J Appl Sport Sci. 1980;5(4):203–7.
13. Mizrahi J, Verbitsky O, Isakov E, Daily D. Effect of fatigue on leg kinematics and impact acceleration in long distance running. Hum Mov Sci. 2000;19(2):139–51.
14. Derrick TR, Dereu D, Mclean SP. Impacts and kinematic adjustments during an exhaustive run. Med Sci Sport Exerc. 2002;34(6):998–1002.
15. Dutto DJ, Smith GA. Changes in spring-mass characteristics during treadmill running to exhaustion. Med Sci Sports Exerc. 2002;34(8):1324–31.
16. Fourchet F, Girard O, Kelly L, Horobeanu C, Millet GP. Changes in leg spring behaviour, plantar loading and foot mobility magnitude induced by an exhaustive treadmill run in adolescent middle-distance runners. J Sci Med Sport. 2015;18(2):199–203.
17. Hunter I, Smith GA. Preferred and optimal stride frequency, stiffness and economy: changes with fatigue during a 1-h high-intensity run. Eur J Appl Physiol. 2007;100(6):653–61.
18. Kadono H, Ae M, Suzuki Y, Shibayama K. Effects of fatigue on leg kinetics during all-out 600 m running. Int J Sport Heal Sci. 2013;11:54–61.
19. Sanno M, Willwacher S, Epro G, Brüggemann G-P. Positive work contribution shifts from distal to proximal joints during a prolonged run. Med Sci Sport Exerc. 2018;50(12):2507–17.
20. Dorn TW, Schache AG, Pandy MG. Muscular strategy shift in human running: dependence of running speed on hip and ankle muscle performance. J Exp Biol. 2012;215(13):2347–7.
21. Kulmala JP, Korhonen MT, Ruggiero L, et al. Walking and running require greater effort from the ankle than the knee extensor muscles. Med Sci Sports Exerc. 2016;48(11):2181–9.
22. Enoka RM, Stuart DG. Neurobiology of muscle fatigue. J Appl Physiol. 1992;72(5):1631–48.
23. Rabita G, Couturier A, Dorel S, Hausswirth C, Le Meur Y. Changes in spring-mass behavior and muscle activity during an exhaustive run at VO2max. J Biomech. 2013;46(12):2011–7.
24. Jones AM, Doust JH. A 1% treadmill grade most accurately reflects the energetic cost of outdoor running. J Sports Sci. 1996;14(4):321–7.
25. Yeadon MR. The simulation of aerial movement—II. A mathematical inertia model of the human body. J Biomech. 1990;23(1):67–74.
26. Sperlich PF, Holmberg HC, Reed JL, Zinner C, Mester J, Sperlich B. Individual versus standardized running protocols in the determination of VO2max. J Sport Sci Med. 2015;14(2):386.
27. Balshaw TG, Massey GJ, Maden-Wilkinson TM, et al. Changes in agonist neural drive, hypertrophy and pre-training strength all contribute to the individual strength gains after resistance training. Eur J Appl Physiol. 2017;117(4):631–40.
28. Rainoldi A, Melchiorri G, Caruso I. A method for positioning electrodes during surface EMG recordings in lower limb muscles. J Neurosci Methods. 2004;134(1):37–43.
29. Worthington PJ, King MA, Ranson CA. Relationships between fast bowling technique and ball release speed in cricket. J Appl Biomech. 2013;29(1):78–84.
30. Bezodis NE, Salo AIT, Trewartha G. Excessive fluctuations in knee joint moments during early stance in sprinting are caused by digital filtering procedures. Gait Posture. 2013;38(4):653–7.
31. Kristianslund E, Krosshaug T, Van den Bogert AJ. Effect of low pass filtering on joint moments from inverse dynamics: implications for injury prevention. J Biomech. 2012;45(4):666–71.
32. Weir JP. Quantifying test-retest reliability using the intraclass correlation coefficient and the SEM. J Strength Cond Res. 2005;19(1):231–40.
33. Hyslop NP, White WH. Estimating precision using duplicate measurements. J Air Waste Manag Assoc. 2009;59(9):1032–9.
34. Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. Hillsdale (NJ): Lawrence Erlbaum Associates Inc.; 1988. pp. 286–7.
35. Cunninghan R, Hunter I, Seeley M, Feland B. Variations in running technique between female sprinters, middle, and distance runners. Int J Exerc. 2013;6(1):43–51.
36. Cavanagh PR. The biomechanics of lower extremity action in distance running. Foot Ankle Int. 1987;7(4):197–217.
37. Arampatzis A, Brüggemann GP, Metzler V. The effect of speed on leg stiffness and joint kinetics in human running. J Biomech. 1999;32(12):1349–53.
38. Riley PO, Dicharry J, Franz J, Della Croce U, Wilder RP, Kerrigan DC. A kinematics and kinetic comparison of overground and treadmill running. Med Sci Sports Exerc. 2008;40(6):1093–100.
39. Howard RM, Conway R, Harrison AJ. Muscle activity in sprinting: a review. Sport Biomech. 2018;17(1):1–17.
40. Clansey AC, Hanlon M, Wallace ES, Lake MJ. Effects of fatigue on running mechanics associated with tibial stress fracture risk. Med Sci Sports Exerc. 2012;44(10):1917–23.
41. McMahon TA, Valiant G, Frederick EC. Groucho running. J Appl Physiol. 1987;62(6):2326–37.
42. Folland JP, Allen SJ, Black MI, Handsaker JC, Forrester SE. Running technique is an important component of running economy and performance. Med Sci Sports Exerc. 2017;49(7):1412–23.
43. Derrick TR. The effects of knee contact angle on impact forces and accelerations. Med Sci Sports Exerc. 2004;36(5):832–7.
44. Schache AG, Dorn TW, Williams GP, Brown NAT, Pandy MG. Lower-limb muscular strategies for increasing running speed. J Orthop Sport Phys Ther. 2014;44(10):813–24.
45. Yeow CH. Hamstrings and quadriceps muscle contributions to energy generation and dissipation at the knee joint during stance, swing and flight phases of level running. Knee. 2013;20(2):100–5.
46. Hanon C, Thépaut-Mathieu C, Vandewalle H. Determination of muscular fatigue in elite runners. Eur J Appl Physiol. 2005;94(1–2):118–25.
47. Wank V, Frick U, Schmidtbleicher D. Kinematics and electromyography of lower limb muscles in overground and treadmill running. Int J Sports Med. 1998;19(7):455–61.
48. García-Pérez JA, Pérez-Soriano P, Llana S, Martínez-Nova A, Sánchez-Zuriaga D. Effect of overground vs treadmill running on plantar pressure: influence of fatigue. Gait Posture. 2013;38(4):929–33.
49. Duffield R, Dawson B, Goodman C. Energy system contribution to 1500- and 3000-metre track running. J Sports Sci. 2005;23(10):993–1002.


Supplemental Digital Content

Copyright © 2021 by the American College of Sports Medicine