Within-Runner Kinetic Responses to Heel and Forefoot Striking During Outdoor Running Speeds

unners habitually land on the heel or the forefoot during initial ground contact, and this pattern affects how ground reaction forces travel through the lower limbs. Previous literature that explores these different patterns utilises indoor laboratories and tests separate groups of runners. These controlled settings provide limited real-world insights and make it difficult to separate the effects of the foot strike pattern from pre-existing differences between runner groups. To explore strike-dependent loading in a realistic setting, this study used a within-participant design on an outdoor track to explore if trained runners were capable of performing both striking patterns. Thirteen nationally competitive male middle-distance runners ran one-minute repetitions at 12, 14, 16 and 18 km h⁻¹, switching between instructed heel and forefoot landing in identical zero-drop running shoes. Vertical ground reaction forces were measured using in-shoe sensors and analysed by multivariable linear mixed effects models.

outdoor track to explore if trained runners were capable of performing both striking patterns. Thirteen nationally competitive male middle-distance runners ran one-minute repetitions at 12, 14, 16 and 18 km h⁻¹, switching between instructed heel and forefoot landing in identical zero-drop running shoes. Vertical ground reaction forces were measured using in-shoe sensors and analysed by multivariable linear mixed effects models. Heel striking, compared to forefoot striking, increased peak impact force by 1110 ± 20 N, loading rate by 19000 ± 360 N·s⁻¹, and time to active peak by 0.005 ± 0.001 s (p < 0.001). Forefoot striking elevated active‑peak force by 340 ± 10 N and impulse by 0.024 ± 0.001 N·s (p < 0.001). Step compliance with the instructed pattern was successful when runners used their habitual forefoot strike (median 100 %, Interquartile range [IQR] IQR 100 - 100%). However, compliance dropped markedly when habitual heel runners attempted a forefoot style (median 52.5%, IQR 30.2–69.0%). This study clearly demonstrates that, while most athletes can effectively adhere to strike instructions, the consistent execution of a non-habitual movement pattern is not uniformly achieved across all individuals. Keywords Foot Strike, Gait, Running 1.Introduction A complete running cycle is comprised of initial contact, mid-stance, toe-off, and mid-swing [1]. The timing and size of these phases depend on strike pattern and speed [2, 3]. The shifts in these phases affect stride rate, swing duration, and ground-contact time [4-6]. Foot-strike patterns in runners are commonly classified as either forefoot or heel contact [7, 8]. Most runners consistently perform one strike pattern, with the vast majority (~95%) using a heel strike [9, 10]. However, this heel strike preference can change to more of a forefoot strike pattern as speed increases [11]. These strike patterns influence not only where contact occurs, but also how runners move through the entire gait cycle. Forefoot strikers typically show shorter contact times, higher stride rates, and shorter stride durations than heel strikers [4, 12]. As speed increases, both ground-reaction forces and loading rates rise correspondingly [13]. Footwear properties such as cushioning thickness and midsole stiffness can mitigate or

only where contact occurs, but also how runners move through the entire gait cycle. Forefoot strikers typically show shorter contact times, higher stride rates, and shorter stride durations than heel strikers [4, 12]. As speed increases, both ground-reaction forces and loading rates rise correspondingly [13]. Footwear properties such as cushioning thickness and midsole stiffness can mitigate or

1090 Within-Runner Kinetic Responses to Heel and Forefoot Striking During Outdoor Running Speeds amplify these loads [13-15]. Many studies have examined strike mechanics in various levels of runners, but several methodological gaps remain [9, 16, 17]. A large proportion of research relies on treadmill protocols, which may not replicate on-ground biomechanics because surface compliance and propulsion demands differ, altering strike behaviour and stride length [16]. Additionally, many investigations compared independent groups of heel and forefoot dominant runners, rather than examining both conditions within the same individuals [4, 9]. Between-subject designs can introduce variability linked to limb morphology, neuromuscular coordination, and training history, obscuring whether observed effects derive from strike pattern or inherent biomechanical diversity. Footwear variation adds another confounding layer: midsole cushioning, stiffness, and heel-to-toe drop strongly influence impact forces and stride mechanics [14, 15]. When shoe models are not standardised, isolating strike-specific effects becomes difficult. Despite the well-established influence of foot strike pattern on ground-reaction forces, stride characteristics, and joint loading, the interaction between foot strike and running speed remains poorly understood in outdoor environments where most training and competition take place [11, 18, 19]. This lack of clarity limits our understanding of how athletes adjust their mechanics in real-world performance settings and how these adjustments influence foot loading. This design was based on the assumption—supported by prior evidence demonstrating that trained runners exhibit neuromuscular adaptability to acute manipulations in footwear, running surfaces, and verbal instructions, enabling immediate modifications to their foot-strike patterns across varying running speeds [20, 21]. To advance this understanding, the present study employed a within-subject design, where runners alternated between heel and forefoot striking at different running speeds. It was hypothesised that runners would be able to execute both strike patterns consistently across the various running speeds. 2. Materials and Methods 2.1. Participants Thirteen male participants were recruited, and all were included in the final analysis. The inclusion criteria required participants to be free from injury (self- identified) for at least 3 months before testing to have experience in track running at speeds similar to those required in the data collection. The group was aged

running speeds. 2. Materials and Methods 2.1. Participants Thirteen male participants were recruited, and all were included in the final analysis. The inclusion criteria required participants to be free from injury (self- identified) for at least 3 months before testing to have experience in track running at speeds similar to those required in the data collection. The group was aged 25.2 ± 6.7 years, weighed 67.2 ± 5.3 kg, standing height 180.5 ± 5.9 cm and racing experience of 7.1 ± 2.9 years (Table 1). Written informed consent was provided by participants, and ethics approval was provided by the University Human Ethics Committee (OM1 24/61). A priori power analysis was conducted to estimate the required sample size for detecting differences in peak ground reaction force between foot strike patterns. Based on a large expected effect size (Cohen’s d = 1.62) derived from previous literature [22], with a significance level (α) set at 0.05 and statistical power (1 – β) of 0. 97, the analysis indicated a minimum sample size of 13 participants (G*Power, (version 3.1.9.7, Heinrich-Heine University, Düsseldorf, Germany). Table 1. Participant (n=13) Characteristics Participant Body mass(kg) Age (years) Standing height (cm) Preferred foot-strike Years running competitively 1 70.5 31 178 H 11 2 67.1 20 185 F 5 3 70.9 19 186 F 7 4 64.7 44 178 F 3 5 77.3 36 185 H 13 6 72.5 29 182 F 5 7 76.3 33 183 F 10 8 70.6 22 185 F 10 9 66.4 22 179 F 5 10 62.0 18 175 F 3 11 63.0 22 179 F 7 12 59.0 22 165 F 8 13 66.0 25 186 F 8 Mean (± SD) 68.2 ± 5.5 26.4 ± 7.7 180.5 ± 5.9 7.3 ± 3.1 Foot strike- heel (H) or forefoot (F)

International Journal of Human Movement and Sports Sciences 13(5): 1089-1096, 2025 1091 2.2. Measures Data was collected on an outdoor artificial 400 m track (Mondo Super X Performance surface, Mondo Worldwide, Gallo d'Alba, Italy). Environmental conditions during data collection were air temperature 13.8 ± 2.2 °C, humidity 59.6 ± 25.4%, and wind speed 7.2 ± 3 km∙h -1 . The experimental design consisted of a five-minute standardised warm-up at 10 km∙h -1 , after which participants were required to run for one minute equating to distances of 200, 233, 266, and 300 m at bicycle paced speeds of 12, 14, 16, and 18 km∙h -1 . These speeds represent submaximal to higher-intensity speeds commonly utilized during endurance training and competition, ranging from easy running (12 km·h⁻¹), moderate endurance training (14 km·h⁻¹), threshold-paced running (16 km·h⁻¹), to race pace conditions for recreational to sub-elite runners (18 km·h⁻¹). This incremental selection was intended to enhance the ecological validity of the study by reflecting typical real-world running intensities [23, 24]. Each trial was followed by a five-minute passive recovery period. Participants were instructed to adopt either a forefoot or heel strike pattern during each trial. For forefoot striking, they were instructed to initiate contact with the forefoot; for heel striking, to land on the heel and roll forward through the foot. The order of the foot strike and speeds were balanced and randomised using a Latin Square method [25]. All participants wore a standardised zero-drop running shoe (Torin 7, Altra Running, Logan, UT, USA) fitted with a validated force pressure insole (Loadsol® Pro, Novel GmbH, Munich, Germany) in both shoes, with kinetic data collected from the right foot insole [26]. Timing gates (TCi Timing System, Brower Timing Systems, Draper, UT, USA) were positioned at 200, 233, 266, and 300 m to match the distances covered in each speed condition. These were used to record the time taken to complete each trial and to calculate the actual average running speed post-trial. 2.3. Data Processing The raw force-time data from the insole was collected at 200 Hz and was pre-processed using a custom-made Matlab code

at 200, 233, 266, and 300 m to match the distances covered in each speed condition. These were used to record the time taken to complete each trial and to calculate the actual average running speed post-trial. 2.3. Data Processing The raw force-time data from the insole was collected at 200 Hz and was pre-processed using a custom-made Matlab code (version 9.2.0, Mathworks, Natick, MA, USA). Initially, the data was filtered using a ten‑point moving average to reduce baseline signal noise while preserving impact peaks and other stance‑phase variables. Visual inspection confirmed that the smoothed baseline stayed within the threshold and that key biomechanical features, such as the impact peak, remained unchanged. The data from the right foot was upsampled from 200 to 1000 Hz, by fitting a 4 th order spline, to permit greater precision in the identification of specific time points such as toe-off or initial contact. To confirm the accuracy of filtering and resampling, pilot analyses were conducted comparing raw force signals originally sampled at 1000 Hz with those down-sampled to 200 Hz, filtered via spline approximation, and resampled back to 1000 Hz. Differences in peak impact force, loading rates, and timing variables between original and processed signals were quantified. Preliminary analyses showed minimal discrepancies (<2%), verifying that the filtering and resampling procedure did not introduce significant errors or distort biomechanical calculations [27]. To ensure only steady-state velocity data were collected, the initial 10 strides and final 10 strides of each trial were excluded from analysis. Initial contact for both the forefoot and heel first strikes was defined as the first point when the vertical ground reaction force exceeded a threshold of 20 N [28]. Toe-off was identified when the vertical ground reaction force dropped below 30 N, marking the point where the forefoot left the ground [29]. Additionally, any steps that did not meet the predefined thresholds, including minimum force thresholds for impact detection, the requirement of a 0.4-second interval between consecutive steps, or peak impact forces that fell below the defined threshold, were classified as non-valid. Steps were also excluded if the time to active

N, marking the point where the forefoot left the ground [29]. Additionally, any steps that did not meet the predefined thresholds, including minimum force thresholds for impact detection, the requirement of a 0.4-second interval between consecutive steps, or peak impact forces that fell below the defined threshold, were classified as non-valid. Steps were also excluded if the time to active peak force did not fall within an expected range of 0.08 seconds after initial impact, or if anomalies such as missing or irregular force peaks were detected through automated filtering. Kinematic and kinetic variables were calculated from the vertical ground reaction force data. Stride duration (time to complete one gait cycle) was the primary kinematic measure analysed. The kinetic variables included peak impact force, defined as the first peak in the vertical ground reaction force following initial contact; if this first peak was absent in a forefoot strike, peak impact force was set at 13% of the stance phase [30]. Loading rate of peak impact was determined by taking the difference between the forces at 20% and 80% of the peak impact force and dividing by the corresponding time interval, representing the slope of the vertical ground reaction force from initial contact to impact peak [31]. Active peak force (the maximum force during propulsion) was the second peak for heel-strike runners or the highest force reading for forefoot strikers; in running, this corresponded to mid-stance, where the foot is directly beneath the centre of mass. Time to peak force was the duration from initial contact to active peak force, while impulse was calculated by integrating force overtime throughout ground contact. 2.4. Statistical Analysis A multivariable linear mixed effects model (lme4 package) was used to analyse the effects of foot strike pattern, running speed, and stride duration on the kinetic outcome measures. Participant identity was included as a random effect to account for individual variation. The outcome variables included peak impact force, loading rate of the impact peak, active peak force, impulse, and time to active peak force. Univariable analysis was used to identify

pattern, running speed, and stride duration on the kinetic outcome measures. Participant identity was included as a random effect to account for individual variation. The outcome variables included peak impact force, loading rate of the impact peak, active peak force, impulse, and time to active peak force. Univariable analysis was used to identify

1092 Within-Runner Kinetic Responses to Heel and Forefoot Striking During Outdoor Running Speeds variables associated with the outcome variables at p < 0.2 and was then included in the multivariable model. Foot strike pattern (as either forefoot or heel), running speed as a continuous variable, stride duration, and preferred strike pattern (y/n; y = yes, the runner was using their habitual strike pattern; n = no, the runner was using a non-habitual strike pattern) were included as fixed effects using a forward stepwise model selection approach. Fixed effects were introduced sequentially, and models were compared using analysis of variance tests and the Akaike Information Criterion to identify the model with the best fit. Model assumptions were evaluated by visually inspecting residual plots and testing for normality, homoscedasticity, and the presence of outliers. For each final model, estimates of fixed effects were reported with marginal and conditional R 2 values to quantify the proportion of variance explained by the fixed effects and the full model (including random effects) respectively. All statistical calculations were performed using R (version 4.3.3, R Foundation for Statistical Computing, Vienna, Austria) with statistical significance set at p < 0.05. 3.Results 3.1. Step Validity Median compliance with the instructed foot strike pattern remained high across all speeds and participant groups (Table 2, Figure 1). When forefoot-preferred runners used a forefoot strike, the median percentage of correct steps was 100% at all speeds, with an interquartile range (IQR) of 0%. When forefoot-preferred runners adopted a heel strike, compliance was lower, with median correct steps of 85.1% (IQR 77.1–93.1%). Table 2. Step Compliance by Strike Type Foot strike Preference n Median % Q1 % Q3% F F 44 100 100 100 H F 44 85.1 77.1 93.1 F H 8 52.5 30.2 69 H H 8 84.1 70.4 96.4 F indicates forefoot strike; H indicates heel strike. Data are reported as median percentage of correct steps (Q1, Q3). Analysis includes 13 runners (11 forefoot preferred, 2 heel preferred). n is the number of one-minute trials per condition. Compliance was highest when runners used their habitual strike: forefoot strikers, using a

52.5 30.2 69 H H 8 84.1 70.4 96.4 F indicates forefoot strike; H indicates heel strike. Data are reported as median percentage of correct steps (Q1, Q3). Analysis includes 13 runners (11 forefoot preferred, 2 heel preferred). n is the number of one-minute trials per condition. Compliance was highest when runners used their habitual strike: forefoot strikers, using a forefoot pattern, and achieved 100% correct steps (IQR 100–100%). In contrast, the heel strike, compliance dropped to 85.1% (IQR 16%), similar to heel strikers' natural heel strike (84.1%, IQR 70.4–96.4%). However, heel-preferred runners were far less consistent when attempting a forefoot strike, with median compliance falling to 52.5% (IQR 30.2–69.0%) This suggests that non-habitual forefoot striking is particularly difficult to adopt, especially for heel strikers. Figure 1. Scatter plot of speed and correct-step percentage for runners (n = 13) with either a forefoot (n = 11) or heel (n = 2) striking pattern.

International Journal of Human Movement and Sports Sciences 13(5): 1089-1096, 2025 1093 3.2. Linear Mixed Models Within the linear mixed effects models there was a significant effect of stride duration, running speed, and foot strike pattern for each kinetic outcome examined (Table 3,). Peak impact force and the loading rate of the impact peak both increased when stride duration and speed increased. Heel striking was associated with an increase in both peak impact force and the loading rate of the impact peak, whereas forefoot striking was associated with a reduction in these outcome measures. Active peak force increased with longer strides and faster speed. Heel striking reduced active peak force, whereas forefoot striking increased. Impulse overall increased with stride duration. Heel striking was associated with lower impulse, whereas forefoot striking led to higher values. Time to active peak force lengthened with longer stride duration and heel striking. This time was reduced with higher speeds and occurred earlier in forefoot striking. Table 3. Mixed-effects Model of Kinetic Outcomes Predictors Estimates CI p R 2 Peak Impact Force (N) (Intercept) −1.33 −1.59 – (−1.08) < 0.001 Stride Duration (s) 2.10 1.86 – 2.33 < 0.001 Strike Preference −0.30 −0.32 – (−0.28) < 0.001 Speed 0.07 0.06 – 0.07 < 0.001 Foot strike 1.11 1.09 – 1.12 < 0.001 Marginal R² / Conditional R² 0.719 / 0.884 Loading Rate of Impact Peak (N s⁻¹) (Intercept) −36.40 −46.84 – (−25.97) < 0.001 Stride Duration (s) 54.64 46.56 – 62.71 < 0.001 Strike Preference −12.11 −12.77 – (−11.45) < 0.001 Speed 2.24 2.13 – 2.35 < 0.001 Foot strike 19.02 18.34 – 19.70 < 0.001 Marginal R² / Conditional R² 0.438 / 0.829 Active Peak Force (N) (Intercept) −1.31 −1.53 – (−1.08) < 0.001 Stride Duration (s) 4.53 4.38 – 4.68 < 0.001 Strike Preference −0.03 −0.04 – (−0.02) < 0.001 Speed 0.08 0.08 – 0.08 < 0.001 Foot strike −0.34 −0.35 – (−0.33) < 0.001 Marginal R² / Conditional R² 0.362 / 0.878 Impulse (N˖s) (Intercept) −0.04 −0.07 – (−0.02) 0.001 Stride Duration (s) 0.59 0.58 – 0.61 < 0.001 Strike Preference −0.01 −0.01

The study investigates kinetic responses to different foot strike patterns in trained runners.