Original Article

Proc IMechE Part P:

J Sports Engineering and Technology
1–14
Effect of mechanically aged minimalist Ó IMechE 2019
Article reuse guidelines:
and traditional footwear on female sagepub.com/journals-permissions
DOI: 10.1177/1754337118824001

running biomechanics journals.sagepub.com/home/pip

Nadine Lippa1,2, Jason Bonacci3, Paul K Collins4,

James W Rawlins2 and Trenton E Gould1

Abstract
The purpose of this study was to improve the understanding of human–material interactions by combining polymer engi-
neering and biomechanical approaches. The forefoot and heel of traditional shoes and minimalist running shoes were
degraded using a mechanical aging protocol to quantify (1) the effect of subject-specific degradation and (2) human biome-
chanical effects due to decreased material properties. Four recreational-level female participants ran in the shoes pre-
mechanical aging to determine the aging protocol input parameters and post-mechanical aging to evaluate the effect of
degradation on kinematics and kinetics. Initial biomechanics translated into different mechanical aging input parameters
among conditions: 500 greater number of impact cycles for minimalist shoe, 430 N higher peak force for forefoot, 75 kPa
greater peak stress for the heel, 3.1 and 13.7 kN/s greater loading rates for minimalist shoe and the heel, and recovery
time 220 ms greater for the heel. From mechanical aging, the shoe types and regions lost 1.2–1.8 mm thickness and 38%–
54% energy absorption overall, while drop decreased 0.6 mm for traditional shoe only. Samples degraded at different rates
depending on runner-specific input parameters. Human kinematics and kinetics were affected by both shoe type and aging.
Aging of the shoes decreased knee flexion velocity (1°/s; p = 0.01), decreased ankle dorsiflexion during stance (3°,
p = 0.01), and increased the vertical loading rate (4 BW/s, p = 0.01). The results support previous findings that different
footwear influence running biomechanics and concurrently advance footwear science to show running biomechanics are
also influenced by shoe degradation rates, such that unique and intuitive human–material interactions are apparent.

Keywords
Ethylene vinyl acetate foam, degradation, hysteresis, energy absorption, running shoes

Date received: 9 December 2017; accepted: 17 December 2018

Introduction rates, depending on the population studied.4–6,9–13

Significant engineering efforts have sought to advance
footwear technology through design or material devel-
opment of the midsole.1–3 Running shoe midsoles are
usually composed of ethylene vinyl acetate (EVA) foam
and function as the main energy damping component
of the shoe.1 While the majority of runners use tradi-
tional shoes (TS), the increase in popularity of barefoot
running led to a rise in use of minimalist shoes (MS).4
The main differences between traditional and minimal
shoes are the overall thickness of the midsole and the
‘‘drop,’’ or the difference in thickness between the heel
and forefoot. Modifications of the midsole cushion pro-
file in traditional and minimalist shoes have been associ-
ated with changes in running gait.4–8 For example, MS
have been suggested to promote shorter strides, a more
distal footstrike, and either higher or lower loading
However, the knowledge of the midsole degradation
rates in both traditional and minimalist shoes is sparse.
In general, mechanical aging (MA) of midsoles has
revealed that EVA foam thickness affects inherent
1
School of Kinesiology and Nutrition, The University of Southern
Mississippi, Hattiesburg, MS, USA
2
School of Polymer Science and Engineering, The University of Southern
Mississippi, Hattiesburg, MS, USA
3
School of Exercise & Nutrition Sciences, Deakin University, Waurn
Ponds, Australia
4
Mechanical Engineering, Deakin University, Waurn Ponds, Australia
Corresponding author:
Trenton E Gould, College of Education and Human Sciences, School of
Kinesiology and Nutrition, The University of Southern Mississippi, 118
College Drive, #5023, Hattiesburg, MS 39406, USA.
Email: trent.gould@usm.edu
2 Proc IMechE Part P: J Sports Engineering and Technology 00(0)
energy absorption capability, with thick, compliant
foam absorbing more energy overall, but degrading at a
faster rate than thinner, stiff foam.14 Additional work is
required to determine whether decreased drop,
decreased energy absorption, and altered running gait
occur in a more biofidelic degradation of the midsole
with realistic forces applied for each shoe-region condi-
tion. It may be anticipated that important interactions
among shoe type, midsole degradation, and runner bio-
mechanics may occur.4–7
Shoe degradation is known to occur15–22 and has
been associated with an increased risk for running
injury,23 yet changes in traditional and minimalist shoe
material properties with aging and its consequences on
human running biomechanics have not been reported.
One study reported that aged shoes affected running
kinematics (i.e. in worn shoes: less maximum lean and
less forward lean at toe-off for the torso; reduced maxi-
mum dorsiflexion and increased plantarflexion at toe-
off),24 but the degree of shoe degradation was
unknown. Conversely, footwear material studies have
reported that shoe degradation decreases midsole thick-
ness, compliance, and the shoe’s ability to absorb
energy,15–22 but have not examined how these changes
influence running biomechanics. For the running ath-
lete, a probable consequence of reduced thickness and
compliance is an increased loading rate,4,25 which has
been reported as a risk factor for running overuse inju-
ries.26,27 However, there is no literature that has
directly related thickness-dependent shoe energy
absorption to runner kinematics, kinetics, or work
done by the lower limb joints.
The ability to endorse or refute the hypothesis that
new shoes ‘‘perform’’ better than aged shoes requires
bridging polymer science and biomechanics, which tend
to occupy separate research silos. Consequently, previ-
ous footwear research falls into two categories: (1)
human biomechanics assessment without reporting
how properties of shoes changed over time4–7 or (2)
shoe MA under nonspecific conditions of an ‘‘average’’
runner excluding the interrelationship of degradation
and human biomechanics.15–21,28,29
Our aim was to determine how footwear degradation
affects knee and ankle running biomechanics as a func-
tion of footwear type. The following hypotheses were
tested: (1) MA input parameters (defined based on
lower limb running kinematics and kinetics) will differ
between minimalist and traditional shoes, (2) the rate of
footwear degradation will be affected by footwear-
specific kinematics and kinetics, and (3) kinematics and
kinetics will change as a function of polymer degrada-
tion and will differ between traditional and minimalist
shoes. The first hypothesis was substantiated by previ-
ous work,4–7 but several gaps still exist. This study is
the first to take the multidisciplinary approach of con-
currently relating footwear degradation and human
running biomechanics using established techniques and
contemporary footwear conditions.
Methods
Participants
Four female recreational runners with a mean (6SD)
age of 25.0 6 5.6 years, height of 1.69 6 0.07 m, and
body mass of 60.6 6 3.4 kg participated. The partici-
pants were heel strikers with US shoe sizes ranging
from 7 to 8.5, running 21.0 6 4.2 km per week at a typi-
cal pace of 5.7 6 0.4 min/km. Participants were
excluded if they had any illnesses that may affect bal-
ance, a lower limb injury in the past 6 months, or were
pregnant.
Participants were familiar with treadmill running
and were familiarized with the laboratory treadmill
before participation.
Experimental shoe conditions
Two pairs of New BalanceÒ running shoes were pur-
chased for each subject: (1) minimalist shoe model
WR10WW2 (MS) and (2) traditional shoe model
W880M13 (TS). To increase similarity between the two
shoe models (e.g. thickness, bending stiffness), the TS
shoes were slightly modified by (1) removing the
optional foam insole and (2) removing the hard plastic
Stability WebÒ. A 20-mm-diameter hole was cut in the
uppers of all shoes, directly above the metatarsal region
of the footbed, to facilitate accurate measurement of
the forefoot region thickness with customized calipers
throughout testing and to permit the compression pla-
ten access to the surface of the shoes during MA
(Figure 1).
Footwear MA protocol
The experimental protocol consisted of a first round
human participant testing (Pre-MA), followed by MA,
and then a second round of human participant testing
(Post-MA) (Figure 2). The shoes were mechanically
aged by a modified version of a previously developed
protocol23,30–32 using a MTS 858 Servo Hydraulic
mechanical testing unit (Eden Prairie, MN; PDIF tun-
ing factors: 36, 6.4, 0, and 0). Platens with 1 mm cham-
fer and 52 or 70 mm diameter applied forces to the heel
or forefoot shoe regions, respectively. Specific input
parameters of the MA protocols were based on decon-
voluted bimodal waveforms corresponding to the four
participants’ Pre-MA kinetics such that the heelstrike
peaks (F1) were applied to the heel regions and the
active peaks (F2) were applied to the forefoot of each
shoe. Therefore, four unique conditions were tested: TS
heel (n = 8), TS forefoot (n = 8), MS heel (n = 8), and
MS forefoot (n = 8). Test duration and number of
cycles depended on the self-determined speed and gait
of each subject in each shoe condition, such that MA
was personalized to each subject’s pre-aging gait. Stride
length (m/stride) was estimated by dividing the tread-
mill speed (m/s) by the stride frequency (strides/s). The
Lippa et al.
Figure 1. Test setup for (a) an example MS sample in compression and (b) the 20-mm-diameter hole cut into an example TS upper.
The inferior portion of the platen is not visible because it was inside the shoe in direct contact with the footbed.
Figure 2. Schematic representation of the study design. The
triangle vertices represent data collection and analysis phases,
while the sides indicate how the outcome of each phase affects
the next. Pre-MA human testing informed input variables for
mechanical aging. Over 42 km, MA caused footwear degradation,
which affected Post-MA human variables.
stride length and force-time profile were then used to
calculate the number of impacts needed to age the
shoes for 42 km. A marathon distance was chosen for
the study because measurable midsole degradation was
shown to occur within this distance,14 and it is a stan-
dard distance for an athlete to run continuously.
Human testing protocol
Participants attended the gait laboratory on two occa-
sions (Pre-MA and Post-MA) to run in each of the
shoe conditions (e.g. MS-Pre represented running in
the MS shoe on the first visit). Thirty-two retroreflec-
tive markers were attached to the pelvis and lower
limbs as per a previously validated musculoskeletal
model,33 and the same investigator performed the pla-
cement of retroreflective markers on every participant.
Data collection consisted of two 5-min running trials
wearing TS and MS footwear, with a 10-min rest
between trials. Previous research has identified that
4 min is sufficient time to optimize the foot–surface
3
interaction with a change in footwear or surface
hardness.34
The participants ran on an instrumented treadmill
(Bertec, Columbus, OH) at a self-selected speed deter-
mined prior to testing. A digital video camera captured
the sagittal view during running trials. Kinematic and
kinetic data were collected for a 20-s duration during
the final 30 s of the 5-min run. The shoe conditions were
counterbalanced between participants and sessions to
eliminate ordering effects.
Kinematics and kinetics
Retroreflective markers were attached bilaterally across
16 standard locations, fixed using double-sided adhe-
sive tape, and further secured with fixomull tape.35
Markers on the medial epicondyles of the knee and
medial malleolus were used only for calibration pur-
poses and were removed prior to running trials. All
analyses were completed on the right limb. Marker tra-
jectories and ground reaction forces (GRF) were
recorded at a sampling frequency of 250 and 1500 Hz,
respectively. A standard Vicon calibration procedure
was followed.33,36–39
Marker trajectory and GRF data were low-pass fil-
tered (20 Hz) with a cubic smoothing spline within
Vicon Nexus, then exported to Visual 3D software (C-
Motion Rockville, MD). A kinematic model33,37 was
defined within Visual 3D from the participant’s stand-
ing calibration trial. Positive values represented posi-
tions of knee flexion and ankle dorsiflexion. Negative
values represented positions of knee extension and
ankle plantarflexion. Joint kinetics were calculated
from filtered data using a standard Newton–Euler
inverse-dynamics approach. Three-dimensional joint
moments were expressed as internal moments normal-
ized to the product of body mass and height.40 Joint
stiffness was calculated as the change in joint moment
divided by the change in joint angle from initial contact
to the point of greatest joint flexion.5 Power was the
product of the net joint moment and corresponding
angular joint velocity. The negative work done during
4 Proc IMechE Part P: J Sports Engineering and Technology 00(0)
initial stance was computed for the knee and ankle
joints by integrating the absorption portion of the
power–time curve. Both power and work were normal-
ized to body mass.
Data analysis
The MA protocol input parameters calculated from
each subject’s bimodal force–time data were F1 and F2
peak forces (PF; N) and recovery time between impacts
(t; ms). From the input parameters and platen sizes,
loading rate (LR; BW/s) and peak stress (speak; kPa)
were also calculated. Throughout aging, the main
dependent variables from the MA output were thick-
ness, drop, net displacement (ND), energy absorbed
(EA), and percent energy absorbed (%EA). Drop was
the measured thickness difference between heel and
forefoot. ND was the displacement maximum sub-
tracted by the minimum for a single cycle. EA was the
integral of the force–displacement curve for a single
cycle (equation (1)). Percent energy absorbed was calcu-
lated by equation (2), where m was a measurement time
point corresponding to 0, 1, 3, 5, 10, 25, and 42 km
ð MA input parameters
Energy AbsorbedðEAÞ = Fdx ð1Þ
EAm
%EA = 3100% ð2Þ
EA0
Temporospatial stride characteristics, joint kine-
matics, moments, power, and work were extracted for
statistical analysis using a customized MATLAB
(MathWorks Inc., Natick, MA) program. The data for
each participant were normalized to the stride cycle
(0%–100%), averaged over the 20 s for each trial and
plotted over the stance phase. Dependent variables of
interest included the peak sagittal plane kinematics,
peak joint moments and power joint stiffness, and neg-
ative work done at the knee and ankle joints during the
first phase of stance (absorption). Differences between
experimental conditions were examined using factorial
repeated-measures analysis of variance (RM ANOVA).
Objective 1. To determine whether minimalist and tra-
ditional shoes promoted unique MA input parameters
(i.e. kinematics and kinetics), the MA input variables
were tested with five 2 (shoe type: TS, MS) 3 2 (region:
forefoot, heel) RM ANOVAs.
Objective 2. To determine whether the kinematics and
kinetics of each shoe type promoted unique footwear
degradation, five RM ANOVAs were used. Specifically,
for change in shoe thickness throughout testing, a 2
(shoe type: TS, MS) 3 2 (region: forefoot, heel) 3 6
(Time: Before/After baseline running test, Pre- and
Post-MA, Before/After follow-up running test) test was
used. For change in shoe drop between TS and MS, a 2
(shoe type: TS, MS) 3 6 (Time) test was used. For the
shoe material variables, three 2 (shoe type: TS, MS) 3
2 (region: forefoot, heel) 3 7 (distance: 0, 1, 3, 5, 10, 25,
42 km) tests were used. As a post hoc test, Bonferroni-
corrected pairwise comparisons were implemented on
the within-subjects time variable to find specific differ-
ences in thickness and drop.
Objective 3. To determine whether changes in kine-
matics and kinetics as a function of aging differed
between traditional and minimalist shoes, eight 2 (TS,
MS) 3 2 (Pre-MA, Post-MA) RM ANOVAs were cal-
culated on the aforementioned human kinetics (four
variables) and kinematics (four variables).
For significant post hoc findings, effect size was cal-
culated using Cohen’s d. The a level was set a priori at
0.05. For inferences about patterns of means across
conditions, within-subjects confidence intervals are gra-
phically presented on normalized scores for each sub-
ject using the Masson and Loftus procedure.41 Results
are reported as mean (6 1 SD) unless otherwise noted.
Results
All input variables are summarized in Table 1 and sig-
nificant statistics are summarized in Table 2. Each
marathon MA test lasted for 4.8 (0.1) h or 23,390 (406)
impact cycles. On average, MS underwent 500 more
impact cycles than TS. Peak forces were 430 N higher
for forefoot (F2 force) than heel (F1 force) (Figure 3).
Peak stress (speak) was 75 kPa greater for heel than fore-
foot. The loading rates were 3.1 and 13.7 kN/s greater
for MS than TS and heel than forefoot, respectively.
The shoe recovery times were 220 ms greater for heel
than forefoot. The resulting mechanical force–time pro-
files closely resembled that of the participants’ impulse
curves (Figure 3).
Footwear degradation
All MA output variables are listed in Table 3 and sig-
nificant statistics are summarized in Table 4. As a func-
tion of human and MA testing, midsole thickness
decreased overall, but there were significant stages of
degradation and recovery between trials (Figure 4(a)).
Post hoc tests on thickness revealed that (1) all condi-
tions decreased from Pre-MA to Post-MA and (2) all
conditions decreased from baseline to follow-up run-
ning tests (Figure 4(a)). Comparing across shoe-
thickness conditions, TS thickness was 8.4 mm greater
than MS and heel was 8.8 mm greater than forefoot.
There was an interaction effect, which indicated that
thickness decreased at a faster rate for heel (than fore-
foot) and TS (than MS). As a function of aging dis-
tance, drop decreased by 0–0.6 mm (Figure 4(b)).
However, post hoc comparison tests on drop revealed
that (1) MS drop did not significantly change from any
treatment; (2) from Pre-MA to Post-MA, TS drop
Lippa et al. 5

Table 1. Group mean (SD) of MA input variables across shoe and region variables derived from MA input calculated from subject-
specific running trials.

MA input variables MS forefoot MS heel TS forefoot TS heel

Cycles [#] 23,629 (364) 23,629 (364) 23,125 (284) 23,125 (284)
Peak force [N] 1318 (94) 909 (127) 1297 (111) 851 (104)
Peak s [kPa] 342 (24) 428 (60) 337 (29) 401 (49)
Loading rate [N/s] 9048 (1007) 25,443 (3552) 8660 (1196) 19,608 (3124)
Recovery time [ms] 437 (4) 658 (18) 443 (8) 657 (16)

MA: mechanical aging; MS: minimalist shoe; TS: traditional shoe.

Table 2. Statistical results for MA input variables across shoe and region variables derived from MA input calculated from subject-
specific running trials.

MA input variables Result DoF F p f

Cycles (#) MS . TS 1, 28 23.10 \ 0.001 0.91

Loading rate (N/s) MS . TS 1, 28 12.48 0.001 0.67
Peak force (N) Forefoot . heel 1, 28 121.39 \ 0.001 0.91
Peak s (kPa) Heel . forefoot 1, 28 24.03 \ 0.001 2.09
Loading rate (N/s) Heel . forefoot 1, 28 240.97 \ 0.001 2.94
Recovery time (ms) Heel . forefoot 1, 28 3517.98 \ 0.001 11.14

MA: mechanical aging; MS: minimalist shoe; TS: traditional shoe; DoF: degree of freedom; f: the standardized mean difference expressed as Cohen’s f.

Figure 3. Representative Pre-MA subject plots for TS and MS show similar bimodal ground reaction force response (a). Mechanical
testing input for all shoe-region conditions demonstrated differences between each condition (b). TS heel and forefoot regions
plotted over 1.5 cycles demonstrate different loading rates, peak forces, and recovery times afforded to each shoe region (c).

decreased 0.6 mm; and (3) there were no differences in
drop measured from baseline to follow-up running tests
(Figure 4(b)). Comparing across shoe-thickness condi-
tions, TS drop was 7.6 mm greater than MS. The overall
change in drop throughout testing (not Pre-MA vs Post-
MA) was equivalent for both shoe conditions (0.5 mm).
As a function of aging distance, ND and EA
decreased (Figure 5). Comparing across shoe-thickness
conditions, TS ND was 2.4 mm greater than MS and
heel was 1.6 mm greater than forefoot. There was an
interaction effect, which indicated that ND decreased
at a faster rate for MS and forefoot. EA also decreased
354 mJ as a function of aging distance (Figure 5).
Comparing across shoe-thickness conditions, TS EA
was 233 mJ greater than MS, but there was no differ-
ence between heel and forefoot. There was an interac-
tion effect, which indicated that EA decreased at a
faster rate for TS and forefoot. Hysteresis curves used
to calculate EA were visibly different among conditions
(Figure 6).
As a function of aging distance, %EA decreased by
44% (Figure 5). Comparing across shoe-thickness con-
ditions, %EA was 2.5% greater for TS (than MS) and
13% greater for heel (than forefoot). There was an
interaction effect which indicated that %EA decreased
at a faster rate for MS and forefoot.
6 Proc IMechE Part P: J Sports Engineering and Technology 00(0)

Table 3. Group mean (SD) of average behavior throughout aging and changes from 0 to 42 km (D) of dependent (i.e. output)
variables across shoe and shoe region variables derived from MA output.

MA output variables MS forefoot MS heel TS forefoot TS heel

Mean aging values

Thickness [mm] 11.1 (0.7) 16.1 (0.8) 15.7 (0.7) 28.3 (1.1)
Drop [mm]a 5.0 (0.5) 12.6 (0.9)
Net displacement [mm] 2.5 (0.4) 3.2 (0.6) 4 (0.4) 6.5 (0.9)
Energy absorbed [mJ] 447 (158) 469 (163) 648 (175) 734 (221)
% Energy absorbed [%] 57 (18) 73 (12) 64 (15) 70 (13)
Change from 0 to 42 km
D Thickness 1.2 (0.3) 1.2 (0.2) 1.2 (0.4) 1.8 (0.4)
D Dropa 0.0 (0.3) 0.6 (0.6)
D ND 1.1 (0.1) 0.5 (0.1) 1.1 (0.3) 0.2 (0.1)
D EA 422 (77) 243 (90) 487 (63) 418 (135)
D %EA 53.9 (1.5) 37.6 (8.1) 47.7 (1.0) 38.9 (3.7)

MA: mechanical aging; MS: minimalist shoe; TS: traditional shoe; ND: net displacement; EA: energy absorbed.
a
Drop is the difference in thickness between MS heel–MS forefoot and TS heel–TS forefoot.

Table 4. Statistical results of MA dependent (i.e. output) variables across shoe and shoe region variables derived from MA output.

MA output variables Result DoF F p f

Shoe effects
Thickness (mm) TS . MS 1, 28 1703.44 \ 0.01 7.91
Drop (mm) TS . MS 1, 14 618.66 \ 0.01 6.74
Net displacement (mm) TS . MS 1, 28 117.45 \ 0.01 2.28
Energy absorbed (mJ) TS . MS 1, 28 26.60 \ 0.01 4.88
% Energy absorbed (%) TS . MS 1, 28 8.26 0.01 1.14
Aging effects
Thickness (mm) Decrease 5, 140 136.65 \ 0.01 2.43
Drop (mm) Decrease 5, 70 4.73 \ 0.01 1.16
Net displacement (mm) Decrease 1, 28 522.30 \ 0.01 4.43
Energy absorbed (mJ) Decrease 1, 28 520.06 \ 0.01 4.43
% Energy absorbed (%) Decrease 1, 28 2298.75 \ 0.01 9.13
Region effects
Thickness (mm) Heel . forefoot 1, 28 1871.24 \ 0.01 8.16
Net displacement (mm) Heel . forefoot 1, 28 57.77 \ 0.01 1.75
% Energy absorbed (%) Heel . forefoot 1, 28 315.23 \ 0.01 3.49
Shoe–aging interactions Greater rate decrease
Thickness (mm) TS . MS 5, 140 5.65 \ 0.01 1.10
Net displacement (mm) MS . TS 1, 28 14.54 \ 0.01 1.23
Energy absorbed (mJ) TS . MS 1, 28 12.79 \ 0.01 1.21
% Energy absorbed (%) MS . TS 1, 28 2.31 0.04 1.04
Region–aging interactions Greater rate decrease
Thickness (mm) Heel . forefoot 5, 140 4.37 0.01 1.08
Net displacement (mm) Forefoot . heel 1, 28 154.57 \ 0.01 2.56
Energy absorbed (mJ) Forefoot . heel 1, 28 13.91 \ 0.01 1.22
% Energy absorbed (%) Forefoot . heel 1, 28 54.45 \ 0.01 1.71

MA: mechanical aging; MS: minimalist shoe; TS: traditional shoe; DoF: degree of freedom; f: the standardized mean difference expressed as Cohen’s f.

Human kinematics and kinetics Comparing MS and TS shoe conditions, participants

The averages of all statistically significant dependent
variables are summarized in Table 5 and significant sta-
tistics are summarized in Table 6. Because participants
selected treadmill speeds of either 2.4 (n = 2) or 2.5
(n = 2) m/s and ran at the same speed on both visits
(6.8 6 0.2 min/km pace), there were no differences in
mean running velocity. Sagittal plane knee and ankle
rotations, velocities, and joint moments are plotted in
Figure 7.
ran in TS with 22.5°/s greater peak knee extension velo-
city, 1.2 mJ/kg greater knee work, 0.04 BW lower peak
medial GRF, and a 20.5 BW/s lower vertical loading
rate (Figure 8). As a function of aging distance, ankle
dorsiflexion decreased 3°, knee flexion velocity
decreased by 1°, and vertical loading rate increased by
4.5 BW/s (Figure 9). There were shoe–aging interaction
effects for knee flexion, knee power, and knee work,
which indicated that TS was greater than MS before
Lippa et al. 7

Figure 4. Measured (a) thickness and (b) drop before and after each human (Pre-MA, Post-MA) and shoe (MA) testing phase.
Decreases in thickness and drop represent degradation, while increases indicate recovery. Error bars represent a 95% confidence
interval.

Figure 5. Plots of (a) net displacement, (b) energy absorbed, and (c) percent energy absorbed demonstrate that the dependent
variables decreased as a function of aging distance for each of the shoe-region conditions. Error bars represent a 95% confidence
interval.
8 Proc IMechE Part P: J Sports Engineering and Technology 00(0)

Figure 6. Hysteresis curves for (a) TS heel, (b) TS forefoot, (c) MS heel, and (d) MS forefoot. The first three cycles of aging (0 km)
and 10, 21, and 42 km are displayed. The slope of the loading portion and area of the hysteresis curves represent sample stiffness
and energy absorption, respectively. The X axis title and units for (a) and (b) are same as (c) and (d).

Table 5. Group mean (SD) of select peak kinematic and average or peak kinetic variables across shoe (TS, MS) and aging conditions
(pre-MA, post-MA) derived from human testing.

Human testing variables MSPre-MA MSPost-MA TSPre-MA TSPost-MA

Kinematic variables
Peak joint angle [°]
Knee flexion 39.6 (4) 39.7 (3.1) 41.8 (3.7) 39.8 (2.5)
Ankle dorsiflexion 19.9 (3) 16.5 (2) 20.4 (2) 17.9 (1.2)
Peak joint velocity [°/s]
Knee flexion 39.6 (4) 39.7 (3.1) 41.8 (3.7) 39.8 (2.5)
Knee extension 256 (17) 250 (19) 278 (10) 273 (22)
Kinetic variables
Peak medial GRF [BW] 0.17 (0.06) 0.17 (0.03) 0.13 (0.04) 0.13 (0.05)
Loading rate [BW/s] 53 (11) 60 (13) 35 (6) 37 (7)
Knee power [W/kg] 9.8 (3.0) 10.0 (1.4) 11.7 (2.6) 10.6 (0.9)
Knee work [mJ/kg] 7.4 (1.5) 7.6 (0.8) 9.0 (1.5) 8.4 (0.9)

MA: mechanical aging; MS: minimalist shoe; TS: traditional shoe; GRF: ground reaction force.

aging, but the shoe conditions were equivalent after Discussion

aging (Figure 10). There were no significant differences
Our goal was to determine how traditional and minim-
in the other kinematics and kinetics.
alist shoes degraded over the time-course equivalent to
Lippa et al. 9

Table 6. Statistical results of select peak kinematic and average or peak kinetic variables across shoe (TS, MS) and aging conditions
(pre-MA, post-MA) derived from human testing.

Human variables Result DoF F p f

Shoe effects
Knee extension velocity (°/s) TS . MS 1, 3 13.74 0.03 2.14
Peak medial GRF (BW) MS . TS 1, 3 12.57 0.04 2.04
Vertical loading rate (BW/s) MS . TS 1, 3 17.09 0.03 2.39
Knee work (mJ/kg) TS . MS 1, 3 15.84 0.03 2.30
Aging effects
Peak ankle dorsiflexion Decreased 1, 3 34.45 0.01 3.39
Knee flexion velocity (°/s) Decreased 1, 3 43.26 0.01 3.79
Vertical loading rate (BW/s) Increased 1, 3 27.88 0.01 3.05
Shoe–aging interactions
Peak knee flexiona TS Pre . all 1, 3 15.05 0.03 2.24
Knee power (W/kg)a TS Pre . all 1, 3 11.73 0.04 1.98
Knee work (mJ/kg)a TS Pre . all 1, 3 11.26 0.04 1.94

MA: mechanical aging; MS: minimalist shoe; TS: traditional shoe; GRF: ground reaction force; DoF: degree of freedom; f: the standardized mean
difference expressed as Cohen’s f.
a
Knee interactions were due to TS Pre condition.

Figure 7. Group mean plots of (a–c) sagittal plane knee and (d–f) ankle rotations, moments, and velocities for the four shoe–aging
conditions as a function of % stance. Peak knee flexion and dorsiflexion values were statistically evaluated for each variable. The X
axis titles and units for (a)–(c) are same as (d)–(f).

a marathon and how this degradation influenced warranted because, despite the attempt to mitigate run-
human running kinematics and kinetics at the knee and ning impact using footwear,30 injury rates have
ankle. The study of human–footwear interactions was remained consistent and no epidemiological study has
10 Proc IMechE Part P: J Sports Engineering and Technology 00(0)

Figure 8. Profile plots of peak ankle and knee angles and velocities demonstrate that (a) ankle dorsiflexion decreased with aging,
(b) there was a significant shoe–aging interaction for knee flexion, (c) knee extension velocity was greater for TS than MS, and (d)
knee flexion velocity decreased with aging. Error bars represent a 95% confidence interval.

Figure 9. Profile plots of relevant kinetics variables from the force-instrumented treadmill demonstrate that (a) vertical loading
rate was greater for MS than TS and increased with aging and (b) peak medial GRF was greater for MS than TS. Error bars represent
a 95% confidence interval.

sufficiently demonstrated that shoes can reduce the rate
of overuse running injury.3 Therefore, the first step to
understanding the role of shoes and injury was to
understand concomitant changes in footwear material
properties and athlete running biomechanics. Overall,
it was expected that degradation and biomechanics
would interact in an interdependent way to produce
consistent performance and running differences
between traditional and minimalist shoes.
MA input parameters
The MA input parameters were not estimates or repre-
sentative impulses,15–22 but were actual unique impulses
measured for each participant during the first running
session (Figure 3). Because each of the unaged shoe
conditions demonstrated different kinematics and
kinetics,4–7 the MA input parameters also varied as a
function of both shoe type and shoe region (Tables 1
and 2). Certain shoe–region combinations yielded more
aggressive input parameters, such as (1) the forefoot
had a higher peak force and shorter recovery time than
the heel, (2) the heel had greater stress and a higher
loading rate than forefoot, and (3) there was a greater
number of cycles and higher loading rate in the MS
compared to the TS. Heel and forefoot appeared to
have ‘‘degradation tradeoffs’’ which prevented any pre-
diction of which region would deteriorate more rapidly.
Conversely, MS was expected to deteriorate more rap-
idly than TS because all MS input parameters were
more aggressive (Tables 1 and 2). While the partici-
pants were expected to run differently in each shoe
type, this study improved understanding of the conse-
quences of footwear selection: differences in shoe type
Lippa et al.
Figure 10. Profile plots of (a–c) stiffness, power, and work for knee flexion and (d–f) ankle dorsiflexion in the landing phase of
stance. There were shoe–aging interactions for (b) knee power and (c) knee work. There were no differences in knee stiffness or
ankle variables. Error bars represent a 95% confidence interval.
affected participant kinematics, which in turn resulted
in unique footwear degradation patterns.
Similarly, input parameters were expected to affect
the level of material degradation because degradation
by cyclic fatigue has been shown to be quite sensitive to
selected input parameters.22,42 Degradation has been
shown to proceed more rapidly due to several factors
related to running kinematics and kinetics, including
(1) decreasing the recovery time between impacts42 or
(2) increasing the number of cycles, peak force, engi-
neering stress, or loading rate.42,43 However, under-
standing the origin and degree of footwear degradation
was improved by also quantifying the subject-driven
input parameters (i.e. specific variables directly related
to kinematics and kinetics), which helped explain why
degradation generally occurred more rapidly for the
MS shoe and forefoot region, compared to TS and heel,
respectively.
11
Footwear degradation
Footwear degradation was exhibited by decreases in
thickness, ND, EA, and %EA with aging distance
(Figures 4 and 5) and indicated that the cell walls
underwent bending, breaking, and permanent plastic
deformation,1,44 characteristic of cell flattening, foam
densification, and reduced performance.18,21,29 Material
stiffness also increased with aging, in agreement with
previous literature.25,45–47 The loading region of the
hysteresis curves did not show a yield point (Figure 6),
which supported previous reports that the microscopic
cell face buckling characteristic of polymer foams1,48–50
was masked by rubber outsole deformation.51
Certain differences in compressive behavior for each
shoe-region combination were attributed to initial mid-
sole foam thickness.14 For example, TS and heel under-
went more displacement and absorbed more energy
12
than MS and forefoot on average because the thicker
foams were more compliant (Figure 6), which resulted
in a less aggressive loading rate, greater deformation,
and larger hysteresis area (Figures 3, 5 and 6).
Conversely, other degradation patterns were largely
influenced by participants’ kinematics and kinetics and
were counter to what one would expect based on thick-
ness alone. From a materials perspective, thicker foams
were expected14 to undergo more rapid deterioration
due to (1) greater compliance, (2) a greater viscous
response,1,21,52 (3) greater strain recovery rates needed
to return to 0 mm displacement, and (4) lower foam
relative density.48,53,54 However, there were several
instances where thinner foams degraded faster than
thicker foams (Table 4), which was due to the unique
running biomechanics in each shoe condition.
Participants ran with higher peak forces, less recovery
time on the forefoot region, a greater stride frequency
(i.e. number of cycles), and a higher loading rate in MS
shoes. These conditions resulted in the forefoot and
MS degrading more rapidly than the heel and TS,
respectively. Results confirmed that the difference in
TS and MS degradation rates were due to the input
parameters (based on human kinematics and kinetics)
rather than thickness or inherent material properties.
Although ND, EA, and %EA decreased more rap-
idly for thinner-foamed shoes/regions, thickness
decreased more for thicker heel than forefoot, which
resulted in an overall decrease in drop.14 A change in
drop due to MA was only evident for TS and was small
(D = 0.6 mm). A similar comparison of thickness
immediately before and after human participants ran in
the shoes suggested that drop did not significantly
change.
In summary, some results were expected due to
material differences, while other findings deviated from
previous reports due to the incorporation of real human
data rather than uniform prescribed input parameters.
As expected, thicker foams were more compliant and
underwent more displacement, such that more energy
was absorbed and in certain ways more degradation
occurred. However, thinner foams degraded more rap-
idly not because of material differences, but because
they were subjected to more aggressive input para-
meters due to how participants ran in the shoes.
Human kinematics and kinetics
Kinematics and kinetics were affected by wearing tradi-
tional versus minimalist shoes, which is consistent with
the literature.4–6, 9–13 Although a heelstrike gait was
observed for both MS and TS conditions, knee exten-
sion velocity was significantly higher for TS and non-
significant trends were observed (i.e. lower stride
frequency, higher dorsiflexion) which suggested that
the participants potentially ran with longer strides in
the TS.12,13 Vertical loading rate and peak medial GRF
were also higher when participants wore MS. Both fac-
tors have been suggested to increase injury risk.55,56
Proc IMechE Part P: J Sports Engineering and Technology 00(0)
The current results suggested that participants did
not substantially adapt their running patterns to a
more distal footstrike in MS. A substantial change in
footstrike pattern between the TS and MS was not
expected because all participants demonstrated a heel-
strike pattern during the pre-test running trial and were
not instructed to change their natural, habitual foot-
strike during testing. Inspection of the vertical GRF
profiles showed that all running tests in both shoes
demonstrated both a passive impact peak and an active
propulsion peak, indicative of a heelstrike pattern.
Average vertical loading rates were, however, greater in
the MS compared to the TS.6 The thinner MS heel
foam was stiffer (Figure 6) and provided less compres-
sion for deceleration during heelstrike. Previous studies
also found that participants wearing MS maintained a
heelstrike and exhibited only slight (if any) kinematics
changes such that impulse curves were bimodal and
loading rates and peak forces increased.2,5,7,8 While
these differences in TS and MS were expected based on
previous literature on minimalist footwear,2,5,7,8 the
effects of footwear degradation warrant further
discussion.
Knee flexion velocity decreased with aging and sev-
eral aging–shoe interactions were observed: (1) peak
knee flexion, (2) knee power, and (3) total work at the
knee in the absorption phase of stance. The results are
supported by previous literature that reported cush-
ioned shoes to elicit higher peak knee flexion compared
to non-cushioned shoes.11,57 Furthermore, the current
results demonstrated that reducing cushioning via MA
reduced peak knee flexion velocity, power, and work.
Because participants maintained a heelstrike gait in all
conditions, we posit that the graded heel of TS-Pre
caused participants to run with greater peak knee flex-
ion and ankle dorsiflexion than when running in an
MS. After aging, the overall midsole thickness, energy
absorption, and drop of the shoe decreased. As a result,
TS-Post mid-stance knee flexion and dorsiflexion
decreased, which consequently affected knee power and
total work.
Interpreting the findings presented herein, one can
calculate several interesting relationships for the habi-
tually shod heelstrike runners as a function of footwear
type. For the MS, a 243 mJ (38%) decrease in energy in
the shoe’s heel and 422 mJ (54%) in the forefoot led to
(1) 3.4° decrease in peak ankle dorsiflexion and (2)
7.0 BW/s increased loading rate. For the TS, a 418 mJ
(39%) decrease in energy in the shoe’s heel and 487 mJ
(48%) in the forefoot led to (1) 2.5° decrease in peak
ankle dorsiflexion, (2) 2°/s decrease in peak knee flex-
ion velocity, and (3) 2.0 BW/s increased loading rate.
Limitations
A limitation of this study was that the outsole geometry
and shoe design were different between shoe condi-
tions. Apart from customizing footwear samples, which
would defeat the applicability of this work to
Lippa et al.
commercial running shoes and injury, a thorough
review of available options revealed these two condi-
tions to be the best approximation of similarity.
A second limitation was that the actual shoe thick-
ness could not be measured during human testing and
needed to be estimated. Because kinematic and kinetic
data were captured in the last 30 s of the subject run-
ning trials (last 30% of total running time), the best
thickness estimates were after baseline and follow-up
running test because thickness was measured immedi-
ately after the participants removed their shoes.
However, thickness and drop were likely dynamic
properties during each running session.
A third limitation was that only knee and ankle
sagittal plane kinematics and kinetics were reported.
Therefore, the intuitive assumption that human-
generated energy absorption would increase in response
to shoe wear was an oversimplification. The mechanical
work of different body segments, movements in frontal
and transverse planes, and metabolic cost must be con-
sidered in future approaches.
Conclusion
In summary, aging the footwear decreased the midsole
thickness, increased the material stiffness, and reduced
the energy absorption of the shoe. Thinner foams, such
as that in the forefoot and in the MS, degraded more
rapidly than thicker foams. These changes in footwear
properties were associated with a decrease in ankle dor-
siflexion during running, which may have been a
response to the reduced attenuation of impact forces.
Despite the changes in kinematics, greater vertical load-
ing rates were observed when running in the aged
shoes.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest
with respect to the research, authorship, and/or publi-
cation of this article.
Funding
The author(s) disclosed receipt of the following finan-
cial support for the research, authorship, and/or publi-
cation of this article: Doctoral student support was
provided by the Australian Government via the
Endeavour Fellowship [grant number 4623-2015]. The
Applied Impact Biomechanics Lab in the School of
Kinesiology and Nutrition is supported by the
Department of Defense and U.S. Army Award
#NSRDEC-W911QY-15-C-0038.
ORCID iD
Trenton E Gould https://orcid.org/0000-0002-5905-
2206
13
References
1. Mills N. Polymer foams handbook. London: Butterworth-
Heinemann, 2007.
2. Heidenfelder J, Sterzing T and Milani TL. Systematically
modified crash-pad reduces impact shock in running
shoes. Foot Sci 2010; 2: 85–91.
3. Richards CE, Magin PJ and Callister R. Is your prescrip-
tion of distance running shoes evidence-based? Br J
Sports Med 2009; 43: 159–162.
4. Moore IS, Pitt W, Nunns M, et al. Effects of a seven-
week minimalist footwear transition programme on foot-
strike modality, pressure variables and loading rates.
Foot Sci 2014; 7: 17–29.
5. Hamill J, Russell EM, Gruber AH, et al. Impact charac-
teristics in shod and barefoot running. Foot Sci 2011; 3:
33–40.
6. Lieberman DE, Venkadesan M, Werbel WA, et al. Foot
strike patterns and collision forces in habitually barefoot
versus shod runners. Nature 2010; 463: 531–535.
7. Horvais N and Samozino P. Effect of midsole geometry
on foot-strike pattern and running kinematics. Foot Sci
2013; 5: 81–89.
8. Deneweth JM, McGinnis R, Zernicke R, et al. Individ-
ual-specific determinants of successful adaptation to min-
imal and maximal running shoes. Foot Sci 2015; 7: S97–
S99.
9. Moore IS, Jones A and Dixon S. The pursuit of
improved running performance: can changes in cushion-
ing and somatosensory feedback influence running econ-
omy and injury risk? Foot Sci 2014; 6: 1–11.
10. Divert C, Mornieux G, Baur H, et al. Mechanical com-
parison of barefoot and shod running. Int J Sports Med
2005; 26: 593–598.
11. Bonacci J, Vicenzino B, Spratford W, et al. Take your
shoes off to reduce patellofemoral joint stress during run-
ning. Br J Sports Med 2014; 48: 425–428.
12. Squadrone R and Gallozzi C. Biomechanical and physio-
logical comparison of barefoot and two shod conditions
in experienced barefoot runners. J Sports Med Phys Fit
2009; 49: 6–13.
13. Franz JR, Wierzbinski CM and Kram R. Metabolic cost
of running barefoot versus shod: is lighter better. Med Sci
Sports Exerc 2012; 44: 1519–1525.
14. Lippa NM, Collins PK, Bonacci J, et al. Mechanical age-
ing performance of minimalist and traditional footwear
foams. Foot Sci 2017; 9: 9–20.
15. Cook SD, Kester MA and Brunet ME. Shock absorption
characteristics of running shoes. Am J Sports Med 1985;
13: 248–253.
16. Brückner K, Odenwald S, Schwanitz S, et al. Polyur-
ethane-foam midsoles in running shoes—impact energy
and damping. Procedia Eng 2010; 2: 2789–2793.
17. Schwanitz S, Moser S and Odenwald S. Comparison of
test methods to quantify shock attenuating properties of
athletic footwear. Procedia Eng 2010; 2: 2805–2810.
18. Schwanitz S and Odenwald S. Long-term cushioning
properties of running shoes. In: Brisson P (ed.) The
engineering of sport 7. New York: Springer, 2008,
pp.95–100.
19. Wang L, Xian Li J, Hong Y, et al. Changes in heel cush-
ioning characteristics of running shoes with running mile-
age. Foot Sci 2010; 2: 141–147.
14
20. Dib M, Smith J, Bernhardt K, et al. Effect of environ-
mental temperature on shock absorption properties of
running shoes. Clin J Sport Med 2005; 15: 172–176.
21. Sun P-C, Wei H-W, Chen C-H, et al. Effects of varying
material properties on the load deformation characteris-
tics of heel cushions. Med Eng Phys 2008; 30: 687–692.
22. Lippa N, Hall E, Piland S, et al. Mechanical ageing pro-
tocol selection affects macroscopic performance and
molecular level properties of Ethylene Vinyl Acetate
(EVA) running shoe midsole foam. Procedia Eng 2014;
72: 285–291.
23. Taunton JE, Ryan MB, Clement DB, et al. A prospective
study of running injuries: the Vancouver sun run ‘‘in
training’’ clinics. Br J Sports Med 2003; 37: 239–244.
24. Kong PW, Candelaria NG and Smith DR. Running in
new and worn shoes: a comparison of three types of cush-
ioning footwear. Br J Sports Med 2009; 43: 745–749.
25. O’Leary K, Vorpahl KA and Heiderscheit B. Effect of
cushioned insoles on impact forces during running. J Am
Podiatr Med Assoc 2008; 98: 36–41.
26. Hreljac A. Impact and overuse injuries in runners. Med
Sci Sports Exerc 2004; 36: 845–849.
27. Hreljac A, Marshall RN and Hume PA. Evaluation of
lower extremity overuse injury potential in runners. Med
Sci Sports Exerc 2000; 32: 1635–1641.
28. Chambon N, Sevrez V, Ly QH, et al. Aging of running
shoes and its effect on mechanical and biomechanical
variables: implications for runners. J Sports Sci 2014; 32:
1013–1022.
29. Verdejo R and Mills NJ. Simulating the effects of long
distance running on shoe midsole foam. Polym Test 2004;
23: 567–574.
30. Shorten M. Running shoe design: protection and perfor-
mance. In: Pedoe DT (ed.) Marathon medicine. London:
The Royal Society of Medicine Press, 2000, pp.159–169.
31. Taunton JE, Ryan MB, Clement DB, et al. A retrospec-
tive case-control analysis of 2002 running injuries. Br J
Sports Med 2002; 36: 95–101.
32. Gardner LI, Dziados JE, Jones BH, et al. Prevention of
lower extremity stress fractures: a controlled trial of a
shock absorbent insole. Am J Public Health 1988; 78:
1563–1567.
33. McLean SG, Su A and van den Bogert AJ. Development
and validation of a 3-D model to predict knee joint load-
ing during dynamic movement. J Biomech Eng 2003; 125:
864–874.
34. Divert C, Baur H, Mornieux G, et al. Stiffness adapta-
tions in shod running. J Appl Biomech 2005; 21: 311–321.
35. Leardini A, Chiari L, Croce UD, et al. Human movement
analysis using stereophotogrammetry: part 3. Soft tissue
artifact assessment and compensation. Gait Posture 2005;
21: 212–225.
36. McLean SG, Borotikar B and Lucey SM. Lower limb
muscle pre-motor time measures during a choice reaction
task associate with knee abduction loads during dynamic
single leg landings. Clin Biomech 2010; 25: 563–569.
37. McLean SG and Samorezov JE. Fatigue-induced ACL
injury risk stems from a degradation in central control.
Med Sci Sports Exerc 2009; 41: 1661–1672.
38. Bell AL, Pedersen DR and Brand RA. A comparison of
the accuracy of several hip center location prediction
methods. J Biomech 1990; 23: 617–621.
Proc IMechE Part P: J Sports Engineering and Technology 00(0)
39. van denBogert AJ, Smith GD and Nigg BM. In vivo
determination of the anatomical axes of the ankle joint
complex: an optimization approach. J Biomech 1994; 27:
1477–1488.
40. Moisio KC, Sumner DR, Shott S, et al. Normalization of
joint moments during gait: a comparison of two tech-
niques. J Biomech 2003; 36: 599–603.
41. Masson M and Loftus G. Using confidence intervals for
graphically based interpretation. Can J Exp Psychol 2003;
57: 203–220.
42. Lippa NM, Krzeminski DE, Piland SG, et al. Biofidelic
mechanical ageing of ethylene vinyl acetate running foot-
wear midsole foam. Proc IMechE, Part P: J Sports Engi-
neering and Technology 2016; 231: 287–297.
43. Ouellet S, Cronin D and Worswick M. Compressive
response of polymeric foams under quasi-static, medium
and high strain rate conditions. Polym Test 2006; 25:
731–743.
44. Mills NJ and Rodriguez-Perez MA. Modelling the gas-
loss creep mechanism in EVA foam from running shoes.
Cell Polym 2001; 20: 79–100.
45. Mills C, Yeadon MR and Pain MT. Modifying landing
mat material properties may decrease peak contact forces
but increase forefoot forces in gymnastics landings.
Sports Biomech 2010; 9: 153–164.
46. Even-Tzur N, Weisz E, Hirsch-Falk Y, et al. Role of
EVA viscoelastic properties in the protective performance
of a sport shoe: computational studies. Biomed Mater
Eng 2006; 16: 289–299.
47. Benanti M, Andena L, Briatico-Vangosa F, et al. Viscoe-
lastic behavior of athletics track surfaces in relation to
their force reduction. Polym Test 2013; 32: 52–59.
48. Gibson LJ and Ashby MF. Cellular solids. 2nd ed. Cam-
bridge: Cambridge University Press, 1997.
49. Ashby MF. The mechanical properties of cellular solids.
Metall Trans A 1983; 14A: 1755–1769.
50. Mills NJ and Zhu HX. The high strain compression of
closed-cell polymer foams. J Mech Phys Solids 1999; 47:
669–695.
51. Silva RM, Rodrigues JL, Pinto VV, et al. Evaluation of
shock absorption properties of rubber materials regard-
ing footwear applications. Polym Test 2009; 28: 642–647.
52. Joubert C, Michel A, Choplin L, et al. Influence of the
crosslink network structure on stress-relaxation behavior:
viscoelastic modeling of the compression set experiment.
J Polym Sci, Part B: Polym Phys 2003; 41: 1779–1790.
53. Han CD and Yoo HJ. Studies on structural foam pro-
cessing. IV. Bubble growth during mold filling. Polym
Eng Sci 1981; 21: 518–533.
54. Liao R, Yu W and Zhou C. Rheological control in foam-
ing polymeric materials: II. Semi-crystalline polymers.
Polymer 2010; 51: 6334–6345.
55. Cheung RT, Wong MY and Ng GY. Effects of motion
control footwear on running: a systematic review. J
Sports Sci 2011; 29: 1311–1319.
56. Daoud AI, Geissler GJ, Wang F, et al. Foot strike and
injury rates in endurance runners: a retrospective study.
Med Sci Sports Exerc 2012; 44: 1325–1334.
57. De Wit B, De Clercq D and Aerts P. Biomechanical anal-
ysis of the stance phase during barefoot and shod run-
ning. J Biomech 2000; 33: 269–278.