Abstract
iomechanical strategies of running gait were compared among healthy and recently injured pediatric and adult runners (N = 207). Spatiotemporal, kinematic, and kinetic parameters (ground reaction force [GRF], vertical average loading rate [VALR]) and leg stiffness (Kvert) were obtained during running on an instrumented treadmill with simulta- neous 3D-motion capture. Significant age X injury interactions existed for cadence, peak GRF, and peak joint angles in stance. Cadence was fastest in healthy adults and 2–3% lower in other groups (p= 0.049). Injured adults exhibited higher variance in stance and swing time, whereas injured pediatric runners had lower variance in these measures (p< 0.05). Peak GRF was highest in non-injured adults (2.6–2.7 BW) and lowest in injured adults (2.4 BW;p< 0.05). VALRs (BW/s) were higher among pediatric groups, irrespective of injury(p< 0.05). The interaction for ankle dorsiflexion/plantarflexion moment was sig- nificant (p= 0.05). Healthy pediatric runners produced more plantarflexion than all other groups (p= 0.026). Pelvis rotation was highest in healthy pediatric runners and lowest in healthy adults (17.3 ◦ versus 12.0 ◦ ;p= 0.036). Pediatric runners did not leverage force- dampening strategies, but reduced gait
pediatric groups, irrespective of injury(p< 0.05). The interaction for ankle dorsiflexion/plantarflexion moment was sig- nificant (p= 0.05). Healthy pediatric runners produced more plantarflexion than all other groups (p= 0.026). Pelvis rotation was highest in healthy pediatric runners and lowest in healthy adults (17.3 ◦ versus 12.0 ◦ ;p= 0.036). Pediatric runners did not leverage force- dampening strategies, but reduced gait cycle time variance and controlled pelvic rotation. Injured adults had lower GRF and longer stance time, indicating a shift toward force mitigation during stance. Age-specific rehabilitation and gait retraining approaches may be warranted. Keywords:running; injury; adolescent; pediatric; gait; loading rate; kinematics 1. Introduction Over the last decade, participation in running among the pediatric population has steadily increased by an estimated 10%. During the 2023–2024 academic year, 423,350 U.S. high school students participated in cross-country, and 1.13 million participated in track and field [1]. According to the 2022 USA Global Runner Survey, adults aged 25–54 years comprise 48% of the overall running population [2]. Industry survey data indicate growing participation in running among the older adult demographic, with a 59% global growth in running club participation among adults [3]. Despite the popularity of running, there are several inherent risks of musculoskeletal injury, including biomechanical patterns and musculoskeletal loading. Several studies have determined the proportion of injuries with incidence rates ranging from 3.2% to as high as 84.9% depending on the study sample [4–6]. Bioengineering2025,12, 937
Bioengineering2025,12, 937 2 of 14 To ensure that participation rates remain high among pediatric and adult runners, it is imperative to understand the age differences in biomechanical strategies when runners are injured and when they are healthy. Pediatric patients are not ‘small adults’ and are likely to have different running responses after an injury. A strong understanding of age-related mechanics is essential for clinicians and therapists to guide the return-to-run process after injury, enabling them to: (1) inform age-specific clinical approaches to gait retraining by tailoring loading rate reduction strategies and movement cueing techniques for younger versus older athletes, and (2) refine the goals and content of rehabilitation programs in distance runners with different injuries by age. Characterization studies reveal differences in running biomechanical parameters be- tween pediatric and adult runners, which may be due to the ongoing development of motor control and coordination [7]. Compared to adults and adolescents in later maturational stages, younger adolescents undergoing developmental changes exhibit shorter stride lengths and higher cadence [8]. In addition, pre-pubertal runners demonstrate higher variability with specific metrics such as hip flexion, hip adduction, hip internal rotation, and knee flexion than post-pubertal runners [9]. As young runners mature, some inves- tigators have documented a reduction in cadence [10] and an increase in peak vertical ground reaction forces (vGRF) [11] which are related to lower extremity overuse running injuries [12]. Pediatric runners control forward momentum during gait termination by adjusting step frequency rather than joint moment adaptations, a strategy less commonly seen in adults [13]. Spatiotemporal gait parameters also differ significantly between chil- dren and adults. Differences such as variable stride length and width, as well as labile velocities, can alter a runner’s center of mass [13]. Cesar et al. [13] found that pediatric runners utilized less stable running mechanics to build momentum during the initial start of a run. Adolescents exhibit asymmetry in their running gait at different speeds, which may contribute to an increased risk of musculoskeletal imbalances over time [14]. Adult runners exhibit refined and stable gait mechanics, characterized by greater efficiency in energy transfer [15]. This stability has
that pediatric runners utilized less stable running mechanics to build momentum during the initial start of a run. Adolescents exhibit asymmetry in their running gait at different speeds, which may contribute to an increased risk of musculoskeletal imbalances over time [14]. Adult runners exhibit refined and stable gait mechanics, characterized by greater efficiency in energy transfer [15]. This stability has been attributed to well-developed neuromuscular coordination and more consistent activation of lower limb musculature [7]. In our recent age comparison of natural running biomechanical asymmetries, we found that runners <18 years produced greater, more variable sagittal hip and knee excursions, greater knee adduction/abduction variability, and ankle joint plantarflexion moment variability than adults [16]. Maturation-related changes in knee and hip strength can also influence running biomechanics; specifically, younger runners may have less effective, dynamically strong hip abductors and knee extensors, potentially predisposing them to altered gait mechanics and increasing their susceptibility to injury [17]. While previous works have characterized running biomechanics in different age populations with different injuries, there is no direct comparison of these features according to age and injury status. As such, clinicians often extrapolate data from injured adults to develop interventions and cues for gait retraining in children and adolescents. To determine whether biomechanical strategies of running motion differ between injured and healthy non-injured pediatric and adult runners, a comprehensive analysis of temporospatial, kinetic, and kinematic features is needed. Therefore, the purpose of this investigation was to compare running gait biomechanical parameters between healthy and recently injured pediatric and adult runners. We hypothesized that pediatric injured runners (<18 yrs) would produce higher GRFs, load rates, and greater lower extremity joint excursions during a typical gait cycle compared to injured adults (25–35 yrs).
Bioengineering2025,12, 937 3 of 14 2. Materials and Methods 2.1. Study Design This was a cross-sectional study of the running gait in pediatric and adult runners without a running-related injury. This study and its procedures followed the guidelines for the Declaration of Helsinki’s Protection of Human Subjects. The study was approved by the University of Florida Institutional Review Board (study #202500580). The manuscript follows the recommended format for observational research described by the statement in Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) [18]. The Study Flow Diagram is presented in Figure. Figure 1.STROBE study flow diagram for observational studies. 2.2. Setting The Exercise Medicine and Functional Fitness Laboratory is located within a qua- ternary health care facility and provides gait services among several performance-based and injury prevention options. After physician referral or self-referral into the laboratory, the team offers 3D running motion analysis, functional movement assessment screen to identify kinetic chain deficits, training program recommendations, gait retraining with cueing, counseling on return-to-run programs, shoe wear selection, and therapeutic exercise to runners of all ages and experience levels (amateur to professional athletes). 2.3. Participants Participants were a convenience sample obtained from our departmental research databank (Study #202101632) from January 2014 to 5 April 2025. Inclusion criteria were: (1) aged <18 years or 25–35 years; (2) injury free for over 6 months (“non-injured”) or injured within the preceding six months (“recently injured”); and (3) previous experience of more than 60 min of using treadmills as part of training, which is considered ample for treadmill accommodation [19]. Exclusion criteria were: (1) aged 19–24.9 years or older than 35 years; (2) major traumatic musculoskeletal injury (i.e., anterior cruciate ligament rupture); and (3) other pre-existing neurological or orthopedic conditions that interfered with normal gait (i.e., scoliosis, sciatic pain, ataxias, cerebral palsy) or a previous history of orthopedic trauma with resultant persistent motion aberrations. Runners were categorized
Bioengineering2025,12, 937 4 of 14 into four groups by age group (<18 years or 25–35 years) and injury status (not injured, recently injured). Exclusion of the ages 19–24 was to ensure that we assessed runners with stable mechanics who were in distinct age groups and not transitioning from adolescence to adulthood, and to avoid the systemic age changes in running form beyond the age of 35 years. A total of 207 runners were included in this analysis. 2.4. Participant Characteristics Characteristics were collected from a comprehensive intake form based on our pub- lished recommendations for runner assessment [20]. Sections included characteristics and medical history, presence and type of injury if present, training history (volume, type, cross- training activities, strengthening exercises), shoe wear characteristics (weight, heel-to-toe drop, heel height), and orthotics if applicable. Injury types present in this analysis included recent history of lower extremity stress fractures (healed and cleared to run), tendinopathies (Achilles, flexor hallucis longus, iliotibial band, extensor hallucis longus), plantar fasciitis, and hip labral pain or tear. The current training status for competition was obtained (yes, no). Each runner used their regular, preferred training shoes for the testing to minimize any acute effects of changing footwear on biomechanical responses. 2.5. Data Sources and Measurements Data were acquired from a comprehensive health history intake and biomechanical running gait analyses. Initial Procedures and Instrumentation for Running Analysis. Motion during running at self-selected speed was captured using a high-speed, seven-camera 3D optical motion analysis system (Motion Analysis Corp., Santa Rosa, CA, USA). Motion data sampled at 200 Hz were synchronized with force plate data collected from an instrumented treadmill (AMTI, Watertown, MA, USA) at 1200 Hz [21,22]. The method of Kadaba et al. [23] was used to apply 33 retroreflective markers on anatomical landmarks, as shown in Figure 24]. Markers were placed onto acromion processes, triceps (midway between elbow and shoul- der), lateral elbow condyles, radial forearms midway between elbow and wrist), dorsal wrists, posterior superior iliac spine, anterior superior iliac spine, anterior bilaterally on the thigh, medial and lateral condyles of the femur, tibial tuberosity, medial and lateral malleoli, calcaneus,
on anatomical landmarks, as shown in Figure 24]. Markers were placed onto acromion processes, triceps (midway between elbow and shoul- der), lateral elbow condyles, radial forearms midway between elbow and wrist), dorsal wrists, posterior superior iliac spine, anterior superior iliac spine, anterior bilaterally on the thigh, medial and lateral condyles of the femur, tibial tuberosity, medial and lateral malleoli, calcaneus, lateral to the head of the fifth metatarsal, and medial to the base of the hallux. One offset marker was placed on the right scapular inferior angle. Prior to data collection, a static calibration trial was collected to develop the computer model of each runner in a neutral anatomical position using commercial software (Cortex version 9.2, Motion Analysis Corp., Santa Rosa, CA, USA). The optical motion cameras captured the static positions of the retroreflective mark- ers. Each marker was identified by the software in the baseline skeletal model using the known marker set. Body segment lengths and joint centers were established based on the anatomical marker placement by the Cortex software version 9.2. Data Collection during Running and Post-Processing. Runners ran at a self-selected velocity, defined as the typical pace used for long-distance running. After acclimating to the treadmill for eight minutes to ensure dynamic stabilization of kinematics, slow-motion videos were captured for reference in the sagittal and frontal planes (240 Hz). Between minutes 9 and 10, a 10-s sample of data was captured, with an average of 12–14 strides. Joint angles at initial foot contact for the ankle, knee, hip, and pelvis were determined from the software tracking of the retroreflective markers in space in each frame for each trial. Using the foot markers, the ankle angle at foot strike was calculated using the foot segment and the ground at initial contact relative to the natural angle during standing [25]. The reference point for joint angles was established at 90 ◦ as vertical for the knee and hip, and
at initial contact relative to the natural angle during standing [25]. The reference point for joint angles was established at 90 ◦ as vertical for the knee and hip, and
Bioengineering2025,12, 937 5 of 14 0 ◦ as horizontal for the ankle angle. The pelvis segment was generated from a digitization of the anterior and posterior superior iliac spine markers, and the pelvis anterior inclination was expressed relative to the horizontal as 0 ◦ of anterior tilt. Figure 2.Anatomical location of the skin-based retroreflective markers. Black circle = marker. Several standard spatiotemporal, spatial, and kinematic variables were determined. Bone models were developed for each runner to establish the individual center of mass (COM) location using commercially available software (Visual3D, version.17.1 C-Motion, Inc., Ger- mantown, MD, USA). [21,26] Marker data were filtered at 9 Hz with a fourth-order, low-pass Butterworth filter. Studies in gait and running analysis recommend cutoff frequencies in the 6–12 Hz range based on residual analysis and signal-to-noise optimization. A 9 Hz cutoff is a well-balanced choice, effectively capturing running biomechanics while minimizing soft tissue artifacts. Bone models were generated using the methods ofde Leva et al. [24] . Gait cycle time is presented in percent (0% = initial foot contact,100% = samefoot contact post-swing phase). Cadence (steps/min) and the vertical displacement of the COM (the difference between the minimal and maximal vertical height of the COM during a gait cycle) were calculated. The distance between two successive placements of the same foot was defined as the stride length. The medial-lateral distance between the proximal end position of the foot at the foot strike and the end position of the foot at the next contralateral foot strike was calculated as the stride width. Stance time was defined as the duration that each foot remained in contact with the treadmill. The medial-lateral range of motion (ROM) excursion of the COM was calculated as the shift in the medial-lateral positions of the estimated COM during an average gait cycle [27]. The foot strike type was determined by the angle between the foot segment and the horizontal ground at foot contact and was visually confirmed with high-speed videos. Runners were categorized into rearfoot and non-rearfoot strikers. Joint excursions of the ankle, knee, and hip represented the angular excursion of the joint
of the estimated COM during an average gait cycle [27]. The foot strike type was determined by the angle between the foot segment and the horizontal ground at foot contact and was visually confirmed with high-speed videos. Runners were categorized into rearfoot and non-rearfoot strikers. Joint excursions of the ankle, knee, and hip represented the angular excursion of the joint in the sagittal, frontal, and transverse planes during an average gait cycle for the ankle, knee, hip, and pelvis. The pelvis was developed from the anterior and posterior superior iliac spine markers, and the anterior inclination was expressed relative to the horizontal as 0 ◦ of anterior tilt.
Bioengineering2025,12, 937 6 of 14 Force data were collected from the instrumented treadmill at a frequency of 1200 Hz. A threshold of 20 N GRF was used to set the initial foot contact and toe-off. GRF data were low-pass filtered with a cutoff frequency of 40 Hz using a 4th-order Butterworth filter. Processed, filtered treadmill data were integrated by Cortex software with the motion data, and three-dimensional kinetics were determined via full inverse dynamics calculations in Visual 3D. The peak GRF, the vertical average loading rate (VALR), and vertical impulses were normalized to body weight. The vertical component was chosen because it is the most widely studied aspect of loading in comparative literature. VALR was calculated using the slope of the∆F/∆t of the most linear portion of the force curve, where∆F is the change in vertical force and∆t is the change in time (between 20–80% of the first rise to the peak of the vertical GRF [28]) or vertical GRF at 13% of stance in case the initial peak was absent. Vertical stiffness was estimated using the following: Kvert= Fmax/∆y, where Fmaxis the peak vertical force and∆y is the maximum displacement of the COM. Leg vertical stiffness, or Kvert, describes the interaction of the load placed on the leg and the resulting limb and central nervous system’s response to attenuate that load. The maximum joint moments in the sagittal plane (flexion/extension) and frontal planes (inversion/eversion, adduction/abduction) were determined for the ankle, knee, and hip. Joint moments were normalized to body mass multiplied by leg length. The preprocessed filtered treadmill data were combined with the motion data. Three-dimensional kinetics were determined via full inverse dynamics calculations implemented in Visual 3D. 2.6. Statistical Analyses Statistical analyses were performed using SPSS version 29.0 (IBM, Armonk, New York, NY, USA). Normality of the data (skewness and kurtosis) was assessed using Kolmogorov- Smirnov tests. Descriptive statistics were calculated on all study variables and demo- graphics. The assumptions of normality were tested on demographic, anthropometric, and training history continuous variables to determine whether differences existed between age brackets. Chi-square tests (χ 2) were used to determine
(IBM, Armonk, New York, NY, USA). Normality of the data (skewness and kurtosis) was assessed using Kolmogorov- Smirnov tests. Descriptive statistics were calculated on all study variables and demo- graphics. The assumptions of normality were tested on demographic, anthropometric, and training history continuous variables to determine whether differences existed between age brackets. Chi-square tests (χ 2) were used to determine if there were differences in cate- gorical variables among the four age groups. Univariate analyses of covariance (ANCOVA) were applied to test group differences for biomechanical variables, where the between- group factors were age (<18 years, 19–35 years) and injury status (non-injured, injured). Based on published evidence that running velocity and sex can affect the biomechanics of running [29], these variables were entered as covariates. The eta squared (η 2) values were provided to show the effect sizes for continuous variables; values of 0.01, 0.06, and 0.14 represented negligible to small, medium, and large effects [25]. Phi values (Φ) were determined as effect sizes for categorical variables. Statistical significance was established in advance atp< 0.05. 3. Results 3.1. Characteristics Participant characteristics are provided in Table. The healthy adult group consisted of more females than the other groups (bothp< 0.05). A lower proportion of injured adults were currently competing compared to the remaining groups (p< 0.001). Compared to adults, a lower proportion of both pediatric groups participated in yoga/Pilates and strength training, and overall, the pediatric group wore shoes with a larger heel-to-toe drop (allp< 0.05). The non-injured pediatric group had the highest proportion of rearfoot strikers (p= 0.042). The injured pediatric group was characterized by a higher proportion of bone injuries and a lower proportion of soft tissue injuries compared to the adult injured group (p< 0.05). The injured adult group participated less in competition around the time of injury (p= 0.023).
Description
This study analyzes running gait differences between healthy and injured pediatric and adult runners.