The effect of upper panel stiffness on biomechanical performance in athletic footwear

Abstract Athletic footwear can improve performance through targeted design features to improve biomechanical mechanisms of performance. The alteration of athletic shoe uppers can impact athletic performance and biomechanics in both males and females; moreover, the impact may differ between males and females due to anatomical differences between sexes. Therefore, assessment of sex-specific impacts of footwear modification on athletic performance is essential when determining how to modify footwear. The purpose of this research was to determine the impact of upper panel stiffness on biomechanical performance during agility tasks and determine if the optimal upper stiffness depends on the sex of the athlete. It was hypothesized that stiffening the upper panel, quantified through uniaxial tension testing, would result in decreased contact time and increased force production during jumping and change of direction tasks. Forty participants (20 males, 20 females) were recruited and completed eight lateral skater jumps, eight counter movement jumps, three triangle drop step drills and six anterior-posterior drills in three footwear conditions that varied in upper panel stiffness. 3D motion capture and embedded force plates were used to measure biomechanical outcomes that were compared between stiffness conditions using a linear mixed effect model. Upper panel stiffness significantly affected contact time during the lateral skater jump with the stiffest shoe resulting in the shortest contact time compared to the two other conditions (p = .02–.046, η2 = 0.18). No sex-related interactions were found for any outcome variable. Taken together, upper panel stiffness is important to consider when developing both male and female footwear.


Introduction
Movements that require agility are common in sports such as basketball, soccer and tennis. Athletic performance can be assessed using biomechanical metrics in tasks that involve rapid changes of direction and power production as well as sport-specific tasks (Falch et al., 2020;Harrison et al., 2021;Philpott et al., 2021;Pryhoda et al., 2021). Modifying footwear is one way to improve performance in these sports (Sun et al., 2020). To this end, athletic footwear brands have invested in research to determine how shoes could improve athletic performance for the last four decades (Nigg, 1988;Nike, Inc, 2013). While most of their research has focussed on differences in the midsole of the shoe (Nigg, 1988;Nike, Inc, 2013), it is imperative to consider other aspects of footwear design such as the upper construction given its large contact area with the foot (Subramanium et al., 2021;Pryhoda et al., 2021). Changing the characteristics of the shoe upper can alter the way athletes perform when completing agility tasks. The upper of footwear has been studied in two primary ways: firstly, by varying the stiffness of the upper material (Subramanium et al., 2021) and secondly by altering the geometry of the closure . Increasing the stiffness, or increasing force required to stretch a material, of the upper decreases the amount of ankle work performed during lateral changes of direction , while modifying the geometry and closing mechanism of a shoe upper can decrease contact time, increase the peak concentric ground reaction force and decrease eccentric work during agility movements . Pryhoda et al. examined geometrical and structural differences in shoe uppers, however, they did not evaluate material differences between footwear conditions. Quantifying the impact of various upper materializations of the topperforming geometry from Pryhoda et al. combines these two lines of research with the goal of providing insights regarding material and design selection to increase athletic performance.
Athletic performance during agility movements can be parameterized with a set of biomechanical variables. Pryhoda et al. examined six biomechanical variables related to athletic performance . The first variable, contact time, is an important performance indicator given that tasks such as a lateral skater jump is intended to train athletes to move as quickly and powerfully as possible. Peak ground reaction force, ankle moment and ankle power during the concentric phase of movement can provide insight as to how much energy is produced during the push-off phase of the movement (Judge et al., 1996). Greater eccentric rate of force development is associated with better jump performance and can be used to assess braking force (Merino-Muñoz et al., 2020), while eccentric and concentric ankle work are important indicators of locomotor performance (Nigg, 1988;Stefanyshyn & Nigg, 1998). The purpose of these biomechanical variables is to completely parameterize the change in direction; an athlete may change direction more quickly at the trade-off of doing so with less force. By evaluating the six biomechanical variables we can attempt to fully describe the performance of the movement.
Sex-specific footwear design is critical due to foot structure differences between men and women (Wunderlich & Cavanagh, 2001). Because differences in foot structure exists between sexes, it is likely that altering properties of the upper and fit will impact males and females differently (Wunderlich & Cavanagh, 2001). These differences could have a substantial impact on athletic performance as well as sex-specific injury risk. Previous work has demonstrated that females experience a higher rate of ankle sprains (Lievers et al., 2020) and plantar fasciitis compared to males (Scher et al., 2009). Plantar heel pain is associated with discomfort in footwear fit and it is possible that because footwear has been more consistently designed for males, females are more prone to these injuries (Sullivan et al., 2015). Therefore, it is important to understand how modifications in footwear can impact performance and injury risk for both sexes (Peterson & Renstrom, 2016). This study examined the effect that sex had on the biomechanical performance variables used to assess performance and whether these differences are influenced by the stiffness of the shoe upper.
The aim of this study was to determine if stiffening the shoe upper using alternative configurations provided biomechanical benefits associated with improved performance. We hypothesized that increasing the upper panel stiffness would enable athletes to decrease contact time, increase peak concentric ground reaction forces, increase peak ankle moments and powers during the concentric phase of movement, increase work at the ankle during the concentric phase of movement, decrease work at the ankle during the eccentric phase of movement and increase the eccentric rate of force development when completing agility tasks.

Materials and methods
Individuals that participated in running and sports that involve cutting (soccer, basketball, football, rugby, tennis) at least three times a week for more than 30 min at a time were recruited for the study and enrolled if they met inclusion criteria. The Virginia Tech Institutional Review Board approved this investigation and all participants signed informed consent prior to study initiation. All data were collected in the Granata Biomechanics Lab at Virginia Tech.

Participants
Twenty male and 20 female athletes between 18 and 35 years old, that exercise at least three times a week for 30 min and participated in running and cutting sports from the university and surrounding community participated in this study. The participants were limited to individuals that wore a men's size 10 or a women's size 8 athletic shoe due to footwear availability.
A statistical power analysis was completed in RStudio (RStudio Inc, Boston, MA, USA) using the superpower (Lakens & Caldwell, 2021) and tidyverse (Wickham et al., 2019) packages with data from Pryhoda et al. in which they determined the differences in contact time between different footwear conditions Pryhoda et al., 2021). A sample size of 19 was required to statistically power the study to 80%. Therefore, 40 individuals were recruited to increase the statistical power of this experiment to explore differences in the other performance metrics and sex-related responses.

Footwear
During data collection, participants wore Adidas FUTURENATURAL (Adidas AG, Herzogenaurach, Germany) shoes with modified uppers with one of three panel materials when performing each task. One pair of each shoe was used across all trials, and shoes were available in either a men's size 10 or women's size 8. The footwear was used in a counterbalanced order so that no shoe condition was used first or last every time. The Adidas FUTURENATURAL shoes were modified by BOA Technology (BOA Technology Inc, Denver, CO, USA) with a tri-panel design (see Figure 1).

Materials testing
The three footwear conditions differed in upper panel stiffness: Blue (A), Grey (B) and Yellow (C). Materials testing was conducted on the sample panels provided by BOA Technology.
The stiffness of the materials was quantified as the change in force divided by the change in displacement with a higher value indicating a stiffer material. Three samples of each material were tensile tested to measure stiffness; additionally, the thickness, width and lengths were measured for each sample. The samples were then put in a screw-driven tensile testing machine (Instron 4204, Instron, Norwood, MA) one at a time, the clamps were hand tightened, and a piece of sandpaper was folded and clamped on each end of the samples to ensure the material did not slip within the clamps. The tensile test was performed at room temperature and was run at 12 mm/min, and the data were collected at 20 Hz. The samples were stretched to 65 mm and then taken out of the clamps. The force measurements were taken at 1.5 mm and 7.5 mm displacement from the start. The averages stiffness of the three samples of each material were determined and reported as the material stiffness.

Testing procedure
Following enrolment, the participants had a series of measurements taken including height and weight and specific measurements of their feet. The Foot Posture Measurement System (Dickerson & Queen, 2021) was used to measure foot characteristics including foot length, truncate foot length, navicular height, dorsum height and foot width. Once the measurements were collected, the participants tried on a specified pair of prototyped shoes and were instructed to walk around in them and warm up for 5 min to ensure that the shoe fit was comfortable. During this warmup, the tasks were demonstrated to the participants and each participant was asked to practice the tasks to ensure that they could complete the task comfortably. The participants were able to adjust the tightness of the shoe by torquing the dial to their preferred tightness during this warmup period but were asked to not adjust the shoe after the warmup period was over and data collection began. This represents an ecologically valid way the participants would wear the footwear in a normal life scenario, and we have no a priori way of determining how tight the participant should make the footwear (Honert et al., 2023). Following the warmup, participants were outfitted with 41 retroreflective markers (25 individual markers and four clusters of four markers). The markers were placed on the lower extremities only and were placed as shown in Figure 2. A 10-camera three-dimensional motion capture system (Qualisys, Gothenburg, Sweden) recorded at 120 Hz and four time-synced force plates (Bertec, Columbus, Ohio) recorded at 1920 Hz were used to collect biomechanical data.
Following marker placement, a static trial was collected for a minimum of 1 s. Following the static trial, the medial knee and ankle markers were removed. The participants then completed the four agility tasks which included eight lateral skater jumps (LSJ), eight counter movement jumps (CMJ), three triangle drop-step drills (TDS) and six anterior-posterior (AP) drills (Figures 3 and 4)   in a random order. These tasks are the same four tasks completed in the Pryhoda et al. study and further explanation can be found in the supplemental materials .
The participants were given 5-min to warm up and learn the different tasks in the first pair of shoes before data collection began. To prevent fatigue, a 2-min rest period was provided between each task. The testing was conducted using a counter-balanced order, where none of the shoe conditions were always used first. A qualitative survey was administered at the end of the session to gauge participants' preferences between the footwear conditions. The participants were asked to rank each shoe from worst to best in terms of how they believed they performed in each shoe in the agility tasks.

Data analysis
Three-dimensional motion capture data were processed in Qualisys Track Manager (Qualisys, Gothenburg, Sweden). The marker position data and force plate data were exported, and additional processing was completed using Visual3D (C-Motion, Inc, Germantown, MD). Marker data were filtered using a low pass Butterworth filter with a 7 Hz frequency cut off and force plate data were filtered using a low pass Butterworth filter with a 100 Hz frequency cut-off (Christina et al., 2001;Crenna et al., 2021;Renner et al., 2018;Stefanyshyn & Nigg, 1997). Internal joint moments and powers were calculated in Visual3D using cardan sequence X-Y-Z. All metrics that included force were normalized to body mass of the individual. The marker positions, ground reaction forces, and joint moments and powers were exported and a custom MATLAB (MathWorks, Natick, MA) script was created to calculate the outcome measures for all the movements.

Statistical analysis
Statistical analysis was completed in RStudio (RStudio Inc, Boston, MA, USA). using the lmerTest package   ( Kuznetsova et al., 2017) for Linear Mixed Models, to determine the differences between the three footwear conditions and the participant's sex. The emmeans package (Lenth, 2022) for post-hoc Dunnett's Test was used to determine the differences in the outcome measures between footwear conditions when differences were present, and the eta squared (g 2 ) effect sizes were determined using the effectsize package (Ben-Shachar et al., 2020). Ground contact time, negative and positive ankle work, peak concentric ground reaction force, plantar flexion (foot movement in which the toes and foot flex towards the sole) moment, ankle inversion moment, plantar flexion power, ankle inversion power and jump height for the counter movement jump were the primary outcome measures. Each outcome variable was entered into a mixed effects model with random intercepts and random slopes (Wilkinson et al., 2023). This model provides a numerical difference for each variable between configurations for each group while enabling each subject's data to be appropriately fit by the model.

Demographic results
The mean and standard deviations of the participant demographics are presented in Table 1. Males and females differed in height, weight, foot length and foot width. Three participants' (three males) marker-based data were removed because of misaligned static trials. The Grey (B) shoe was ranked the best shoe by 17 of the 40 participants. Twelve participants selected the Yellow (C) shoe as the best and 11 selected Blue (A) as the best. There were multiple differences seen between males and females in performance  variables (Tables 5 and 6, Supplemental Table). In the counter movement jump, males jumped higher than females. In the lateral skater jump and triangle drop step drill, males had higher peak frontal plane moment compared to females. Females produced more concentric work in the Anterior-Posterior drill compared to males.

Materials testing results
The Blue (A) panel was the least stiff material and was less than half of the stiffness of the highest stiffness panel, Yellow (C). The Grey (B) panel material was 3.75 N/mm higher than the Blue (A) panel and 4.29 N/mm less than the Yellow (C) panel. Material testing results are presented in Table 2. The force/displacement results are visualized in Figure 5.

Performance results
In the lateral skater jump task, the Grey (B) and Blue (A) conditions had higher contact times (p ¼ .020, 0.046) compared to the Yellow (C) condition. The eta squared effect size for contact time was large (g 2 ¼ 0.18). Eta squared effect sizes are considered small when g 2 ¼ 0.01, medium when g 2 ¼ 0.06 and large when g 2 ¼ 0.14 (Cohen, 1988). No differences were seen in the linear mixed model analysis between footwear conditions for any other outcome measures in any movement (p > .050); however medium effect sizes were seen for multiple outcome measures across the different agility tasks (CMJ: peak concentric ground reaction force (GRF), negative ankle work, peak eversion ankle moment, peak eversion ankle power, LSJ: peak concentric GRF, peak ankle inversion moment, negative ankle work, Triangle: peak inversion ankle moment, AP: positive ankle work) (Tables 3-6, Supplemental Table). These results suggest that there is a non-linear relationship between upper panel stiffness and lateral quickness. The highest stiffness shoe, Yellow (C), performed the best while the middle stiffness shoe, Grey (B), performed the worst for the lateral skater jump task (Table 4).

Discussion
We investigated how stiffening of the upper panel material in a shoe with an alternatively configured upper closure impacts biomechanical performance in agility tasks. The participants increased their lateral quickness in the shoe condition with the highest stiffness which supports the theory that stiffness affects lateral quickness. Interestingly, the middle stiffness condition resulted in the slowest lateral quickness. This suggests that the stiffness of the upper panel material could have a non-linear relationship with lateral quickness in this geometry upper configuration. The non-linearity seen in this study could suggest that there are multiple ideal stiffness levels that allow for improved  performance. Increasing or decreasing stiffness does not warrant automatic improved or decreased performance. The performance improvements were only seen in the lateral skater jump which contradicts the hypothesis that altering stiffness will impact biomechanical performance metrics in all the agility tasks. Group effects were consistent with individual trends on average when looking at the lateral skater jump results ( Figure 6). Subjects were slowest in the grey configuration and fastest in the yellow configuration. Twenty-eight individuals were fastest in the yellow condition relative to the grey, 25 were faster in the blue condition relative to the grey and 27 were faster in the yellow condition relative to the blue condition. The partial-pooling, mixed-effects model used in this study considers the individual data and learns from the entire population to produce the final estimated results. Observing individualized data points will help in the adoption of biomechanical data in footwear development. Individualized performance can be altered by a multitude of factors and one step towards improved   footwear is the use of datasets containing subject-specific performance differences.
Previous literature has shown that stiffening the upper of an athletic shoe reduces negative ankle work during touchdown and push-off phases. In the study by Subramanium et. al, only two conditions were tested, and the shoe had a typical u-throat design. Exploration of the impact that a wider range of upper stiffnesses could allow for better optimization of footwear design for athletic tasks. It is also possible that other material properties not  ). Each individual plot shows the individual intercepts and slopes estimated from the statistical model. The type of the line represents the estimated difference in contact time between that condition and the blue condition. A positive slope indicates that subject was slower in the condition of the shape of the line while a negative slope indicates they were faster. For example, Subject 13 shows a faster time in the yellow condition relative to blue and a slower time in grey relative to blue. explored in this study resulted in the differences seen between the footwear conditions. There were several sex-associated differences in performance, however, we did not identify any interaction effects of footwear between performance and sex. The female participants were shorter and lighter than males. The males jumped higher and produced more frontal plane moments in lateral jumping and in the triangle drop step change of direction movement. However, the females produced more concentric ankle work during the AP drill. Regarding performance, producing more positive work at the ankle is beneficial, but also means a greater chance of injury. The lack of interaction effects between footwear and sex suggests that increasing stiffness improves performance in lateral quickness in both men's and women's footwear. Decreasing stiffness increases performance in power production in the ankle during anterior-posterior movements in men and women.
There was not a consensus on the best performing shoe between the qualitative survey and the biomechanical performance metrics. The difference in perception of the bestperforming shoe is unknown but could be caused by aesthetics.

Limitations
There were limitations to this study that should be considered when interpreting the results. The only biomechanical metrics observed were at the ankle joint. Other lower extremity joints could experience changes in moments, powers or work to compensate for the changes in footwear. Also, because this study focussed on biomechanical variables, performance changes caused by neuromuscular input could not be determined. This was minimized by counterbalancing the order of shoe configurations; however, participants could still look at and feel the difference in the shoes on their feet before performing the tasks. The sample size determined for this study was based on only one variable of interest and an increase in sample size would have increased the power of this study.

Future work
Additional material properties such as energy return (tan delta) and torsional properties of the upper should be explored in further studies to understand more about the impact that changing the upper panel material has on biomechanical performance metrics. Wider ranges of stiffnesses should be researched as well to determine if higher and lower stiffness levels alter performance. Variations in the location of stiffness should be explored in future studies to further understand how footwear upper design impacts athletic performance. The impact that upper panel stiffness has on different athletic activities such as running should be explored as well.