Article Text

Download PDFPDF

Running in new and worn shoes: a comparison of three types of cushioning footwear
  1. P W Kong,
  2. N G Candelaria,
  3. D R Smith
  1. University of Texas at El Paso, El Paso, Texas, USA
  1. Correspondence to Dr Pui W Kong, Physical Education and Sports Science Academic Group, National Institute of Education, Nanyang Technological University, 1 Nanyang Walk, Singapore, 637616; venikong{at}


Objectives: In this study, the effect of shoe degradation on running biomechanics by comparing the kinetics and kinematics of running in new and worn shoes was investigated. Three types of footwear using different cushioning technologies were compared.

Design: Longitudinal study.

Setting: Pre- and post-tests on overground running at 4.5 m s−1 on a 20-m laboratory runway; performance measured using a force platform and a motion capture system.

Participants: 24 runners (14 men and 10 women)

Interventions: 200 miles of road running in the same pair of shoes. Within-group factor: shoe condition (new/worn); between-group factor: footwear type (air/gel/spring).

Main outcome measurements: Stance time was calculated from force data. External loads were measured by maximum vertical force and loading rate. Kinematic changes were indicated by sagittal plane angles of the torso, hip, knee and ankle at critical events during the stance phase.

Results: Stance time increased (p = 0.035) in worn shoes. The torso displayed less maximum forward lean (p<0.001) and less forward lean at toe-off (p<0.001), while the ankle displayed reduced maximum dorsiflexion (p = 0.013) and increased plantar flexion at toe-off (p<0.001) in worn shoes. No changes in the hip and knee angles. No between-group difference among the three footwear groups or condition by type interaction was found in any measured variables.

Conclusions: As shoe cushioning capability decreases, runners modify their patterns to maintain constant external loads. The adaptation strategies to shoe degradation were unaffected by different cushioning technologies, suggesting runners should choose shoes for reasons other than cushioning technology.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Running is a popular competitive and recreational activity with an estimated 30 million American participants.1 Although there are many positive health benefits associated with running, the incidence rate of running-related injuries has been reported to be as high as 59 per 1000 h of running.2 Despite the high injury incidence rate, the cause of running injuries remains unclear, probably because multiple factors are involved.1 One factor that has been associated with running injuries is footwear.3

Footwear may play an important role in preventing injuries by absorbing external shock due to ground impact.4 5 Nevertheless, the evidence is divided. One review6 reported that shock-absorbing insoles attenuated the number of stress fracture injuries, but others showed that the visco-elastic material of shoes were not effective in lowering the incidence of overuse7 and soft tissue8 injuries. Running in inappropriate footwear has been associated with bone fractures9 and plantar faciitis.10 Based on the mechanical properties of the footwear, one study concluded that high external loads may cause injuries, although the actual loads acting on the body during running were not measured.9 Thus, it is unclear whether footwear causes injuries due to high external load, since gait adaptation occurs to accommodate to different footwear.11

Shoe age may play a role in the influence of footwear on running injuries. A foam copolymer of ethylene and vinyl acetate (EVA) is commonly used in shoe midsoles to absorb the energy from ground impact. More recently, embedded air, gel, rubber, altered EVA and springs have also been used in the midsole. Irrespective of the technology, shoe cushioning capability decreases with mileage. One prospective study showed that injury was associated with running shoe age.3 It is unclear, however, how much mileage was accumulated over each training period, making the results difficult to interpret and determination of the effect of shoe age on injury risk inconclusive. Moreover, it is unknown if the type of technology used in the shoe midsole played a role in the deterioration of the shoes.

Running mechanics can be influenced by shoe midsole stiffness/geometry as demonstrated in studies where subjects wore several pairs of shoes each with a specific stiffness/geometry.12 13 14 15 16 However, changes in running mechanics when wearing different pairs of shoes may differ from those when wearing the same pair of shoes over time. Degradation of shoes/insoles from machine-simulated running has also been shown to differ from that during in vivo loading.17 18 Thus, longitudinal studies on runners wearing the same pair of shoes over time are necessary to examine the effect of shoe degradation on running mechanics. To date, no information regarding how changes in shoe properties over time influence running pattern is available, although longitudinal changes in mechanical properties of shoes17 and kinetic parameters5 19 with shoe degradation have been investigated. However, it is well known that mechanical tests do not predict shock during actual running because of neuromuscular adaptation,11 20 21 and there is a lack of kinematic data in previous longitudinal studies.

Thus, the purpose of the present study was (1) to compare the kinetics and kinematics of running in new and worn shoes using a longitudinal study design and (2) to investigate if cushioning technology (air/gel/spring) plays a role in influencing running biomechanics over time. It was hypothesised that running in worn shoes would not cause changes in external loads because of neuromuscular adaptation of the runners and that such adaptation would be reflected by changes in kinematics.



Thirty participants (15 men and 15 women) were recruited to the study, which was approved by the University of Texas at El Paso Institutional Review Board. All participants ran at least 20 miles/week for 2 years preceding the study and had not experienced any injuries of the lower extremities during this period. Written informed consent was obtained before data collection. Five men and five women were randomly assigned to one of three footwear groups—air (Nike Pegasus 2005), gel (ASICS GT-2100) or spring (Spira Volare II). Each participant was given one pair of assigned running shoes, which was distributed on the day of the pre-test. Participants were instructed to use the shoes only during data collections and for the completion of 200 miles of road running. Of the 30 participants, 24 completed the required mileage and returned to the laboratory for the post-test. Of the six participants who failed to complete the study, one was injured, one became pregnant and the others did not specify a reason. Descriptive characteristics of the 24 participants who completed the study are shown in table 1.

Table 1

Descriptive characteristics of 24 participants


Pre- and post-tests

On the pre-testing day, participants reported to the laboratory where their mass and height were measured. Participants completed a standardised 10 min warm up session that included quadriceps, hamstrings and triceps surae stretches along with a treadmill run in the assigned running shoes. Participants were encouraged to practise running at a speed of 4.5 m s−1 during the treadmill warm up in preparation for data collection. This speed represents 6 min/mile, which would result in a marathon of 2 h 37 min and is considered an average training pace for competitive distance runners.22 After the warm up, participants were asked to run along a 20-m runway at 4.5 m s−1, while synchronised kinetic and kinematic data were recorded. Each participant made five right followed by five left foot contacts with a force platform (Advanced Mechanical Technology, Model OR6-6-2000, Watertown, Massachusetts, USA) located in the middle of the runway. A successful contact was defined as the participant’s foot striking the force platform while running at a speed of 4.5 m s−1 (±1%) without altering their running technique. Speed was measured by a radar gun (Radar Sales, Plymouth, Minnesota, USA). At the end of the pre-testing, participants were instructed to complete 200 miles of road running in their assigned shoes. Treadmill running was not allowed. No instructions were given with respect to the exercise intensity, frequency of the running bouts or the duration in which to complete the 200 miles. Participants were left to self-determine their own training schedule. All participants were, however, supplied with a logbook in which they were instructed to record all mileage while running in the assigned shoes. The time in which participants returned to the laboratory for post-testing varied from 3 to 22 weeks (mean (SD) = 16.5 (5.0 weeks)). One reason for the large variation in completion time was that many participants alternated testing shoes with personal shoes during their training. Post-testing was identical to the pre-testing procedures.

Kinetic and kinematic data

Vertical ground reaction force (VGRF) was collected at 1200 Hz and low-pass filtered at 100 Hz16 23 using a fourth-order Butterworth filter in Matlab (The MathWorks, Natick, Massachusetts, USA). To obtain kinematic data, reflective markers were placed on both sides of the body at the following locations: acromion process of the scapula, greater trochanter of the femur, lateral epicondyle of the femur, lateral malleolus of the fibula, lateral portion of the calcaneus (over shoes) and the location of the fifth metatarsal (over shoes). The running trials were recorded at 60 Hz using a video camera (JVC-TK C1380) placed perpendicular to the plane of motion. Marker positions were digitised and low-pass filtered at 13 to 16 Hz17 using a Butterworth filter before calculating sagittal plane angles (table 2) using the Motus software (V.8.5, Vicon, Centennial, Colorado, USA). This software has been shown to produce accurate and reliable angular measurements.24 All angles were corrected to a reference upright standing position in the test shoes (zero°) recorded before each data collection to eliminate marker placement errors between the pre- and post-testing days. The angle data were interpolated and re-sampled to 1200 Hz to match with the VGRF data.

Table 2

Kinematic variables of interest

Variables of interest

From the VGRF data, touchdown and toe-off were identified, and stance time was calculated (fig 1). Instantaneous loading rate of the VGRF from touchdown to the first maximum value was calculated by differentiating the force–time history. Kinetic variables included maximum vertical active force (Fmax) and maximum instantaneous loading rate of the vertical force (Gmax), both normalised to the individual subject’s body weight (BW) (fig 1). These variables have been commonly used to indicate external shock.23 25 26 Kinematic variables analysed included each joint angle at touchdown and toe-off, together with maximum forward lean, hip flexion, knee flexion and ankle dorsiflexion. For each subject, a 10-trial mean value from both left and right sides was used for statistical analysis.

Figure 1

Three temporal and kinetic variables of interest obtained from the vertical ground reaction force (VGRF) data: (a) stance time from touchdown to toe-off, (b) maximum vertical active force (Fmax), and (c) maximum instantaneous loading rate from touchdown to the first peak vertical force (Gmax). (Note: Maximum vertical impact force was not analysed because of high inconsistency between and within subjects.)

Statistical analysis

Differences among results were analysed using statistical software (SPSS V.15.0, SPSS, Chicago, Illinois, USA). A one-way analysis of variance (ANOVA) was used to determine differences in anthropometric, temporal, kinetic and kinematic data among participants in the three footwear groups at baseline. A repeated measures ANOVA was used to analyse differences in each variable of interest—within-group factor: shoe condition (new/worn); between-group factor: footwear type (air/gel/spring). A 5% level of significance was adopted throughout, and all data were expressed as mean (SD).


At baseline, no differences were observed in physical characteristics or running mechanics among the participants in the three footwear groups when running in new shoes. Table 3 shows the temporal, kinetic and kinematic variables in both shoe conditions with data from all footwear groups combined. No difference among the three types of footwear or condition by type interaction was found in any variables.

Table 3

Temporal, kinetic and kinematic variables during running in new and worn shoes

With all three footwear groups combined, stance time increased (p = 0.035) when running in worn compared with new shoes. No difference was observed in the Fmax or Gmax between new and worn shoes. The torso displayed less maximum forward lean (p<0.001) and less forward lean at toe-off (p<0.001), while the ankle displayed reduced maximum dorsiflexion (p = 0.013) and increased plantar flexion at toe-off (p<0.001) in worn shoes. Typical torso and ankle angle time histories of one trial in new and worn shoes are compared in fig 2. No difference was found in any hip or knee angles.

Figure 2

Typical torso and ankle angle time histories of one trial during the stance phase of running in new (solid line) and worn (dotted line) shoes. Positive angles refer to forward torso lean and ankle plantar flexion from the reference standing position.


This is the first longitudinal study to describe the kinetics and kinematics of running in new and worn shoes and the first to compare how the degradation of different cushioning technologies influences running biomechanics. The three major findings are (1) worn shoes increase stance time; (2) worn shoes cause kinematic changes but do not influence force variables and (3) kinematic changes in response to shoe degradation are similar in different shoe cushioning technologies (air/gel/spring).

This study tested the new and worn shoe conditions on different days, raising the concern that difference between shoe conditions may be because of between-day variability. However, high repeatability of kinetic and kinematic variables have been demonstrated when subjects wear the same type of footwear 1 week apart.27 Also, the speed was strictly controlled within 1% deviation in the present study compared with the commonly used 5% to further reduce the potential influence of speed. Thus, we believe that the changes in the worn shoe condition are results of shoe degradation.

Stance time

In the present study, stance time was found to be longer (p = 0.035) in worn shoes irrespective of the type of cushioning technology. Stance time has been related to a change in shoe cushioning properties.28 Mechanical testing of the changes in shoe properties was not examined in the present study. It is reasonable, however, to expect some structural damage after 200 miles of road running since previous studies have shown a reduction in shoe shock absorption capability after 150 miles17 and 330 miles.5 Thus, the shoes in the present study may have become less “elastic” with mileage, resulting in a loss in shock attenuation capability. The literature is divided with respect to the effect of cushioning properties on stance time.28 29 30 31 One study found longer stance time in hard shoes in comparison with soft shoes, although the difference was not statistically significant.29 In contrast, another study reported that stance time was longer in “special soft” shoes compared with normal shoes.28 Longer stance time has also been observed in shod (soft) compared with barefoot (hard) running.30 31 There may be a non-linear relationship between stance time and shoe hardness with the optimum hardness remaining to be determined. While a relationship between shoe hardness and stance time cannot be drawn from the present study, it is believed that a change in stance time could be an indicator of shoe degradation. It has been shown that shoe hardness can influence energy expenditure in running28 and that shorter stance time is related to higher running economy, typically measured by the rate of sub-maximal oxygen consumption at a given speed.32 Our study showed that at the same running speed, stance time increased in worn shoes compared with new shoes. This may indicate a reduction in running economy when wearing worn shoes, which is detrimental to performance and could influence fatigue. Further investigation on running economy and shoe age will be needed to verify this speculation. Despite that statistical difference detected in stance time, the mean difference of 4 ms between new and worn shoes is small and therefore is unlikely to be functionally significant.

Gait adaptations

Constant external loads

To maintain an “optimal running condition”, it has been suggested that runners adapt their gait in response to changes in foot–ground interface stiffness.12 20 26 29 30 33 Such adaptations may be a strategy to maintain a constant vertical impulse and stance time,29 limit the local pressure underneath the heel,30 minimise metabolic cost,12 maintain26 or attenuate33 external impact forces, and operate within a “kinetic bandwidth”.20 The concept of maintaining constant stance time are in contradiction with the present study. Since Fmax or Gmax did not change as mileage increased in the present study, our results support the hypothesis that a slight modification in shoe mechanical characteristics does not lead to changes in external force variables because kinematic changes occurred to maintain constant external loads. This is in consensus with previous findings that shoe type does not cause changes in external shock as measured by force amplitudes,26 loading rate,26 tibial acceleration and time to peak acceleration.20 Our study did not measure oxygen consumption or plantar pressure and therefore cannot address issues related to metabolic cost or localised heel pressure.

Kinematic adaptation

The main kinematic changes in response to the mileage were observed at the ankle, which displayed reduced maximum dorsiflexion and increased plantar flexion at toe-off. At least one previous study has demonstrated that changes in shoe midsole hardness influence ankle kinematics.12 Ankle coordinative strategies in response to foot–ground interface stiffness are also observed in barefoot and shod running studies.30 33 The lack of differences in the ankle angle at touchdown between new and worn shoes was primarily because of the high inconsistency in running styles between and within runners. Another possible explanation could be that the human body regulates impact force passively during the initial stance phase without changes in muscle activation pattern and therefore no kinematic changes are necessary at or before touchdown in response to different shoe sole hardness.34

Shoe type

The change in stance time, a reflection of shoe degradation, did not differ among the three types of footwear tested. Although the degradation rate may differ among footwear types, the difference is too small to have a big influence since runners displayed similar kinematic adaptations in worn shoes regardless of footwear type. This implies that there is no clear advantage of incorporating a particular technology (air/gel/spring) in footwear with respect to shoe degradation, although it is possible that differences would have been observed with longer mileages. This is important as shoe life is of interest to runners and shoe manufacturers. General recommendations of shoe life range between 350 miles35 and 600 miles.36 While it is beyond the scope of the present study to determine an appropriate shoe life, our results demonstrate that changes in footwear properties after 200 miles of road running lead to kinematic adaptations. Whether such adaptations are beneficial or detrimental in relation to performance and/or injury prevention has yet to be explored.

Intrasubject and intersubject variability

One limitation of the present study is that not all participants were rearfoot strikers. Since midfoot/forefoot strikers do not display a double-peak VGRF pattern, we did not analyse the peak impact force that is commonly used to indicate external shock. More surprisingly was the observation that inconsistency in running style existed not only between runners but also within an individual. Among the 24 participants, 11 were consistent rearfoot strikers, and one was a consistent forefoot striker. Two participants showed a consistent midfoot/forefoot striking pattern during the pre-test but rearfoot pattern during the post-test; one showed exactly the opposite trend. The remaining nine participants displayed inconsistent striking patterns between legs and/or within the same leg. Intrasubject variability in running pattern has not been well documented in the literature. One study examined six foot strikes of 20 male runners and found that 17 were consistent rearfoot strikers while 3 displayed both rearfoot and midfoot striking patterns.25 This corresponds to 15% of the total sample size being inconsistent rearfoot strikers, compared with 50% in our study. It is uncertain whether the change in running style within individuals in the present study was related to footwear, mileage, “targeting” the force platform and/or unfamiliar pace. It is possible that variability is necessary to prevent overuse injuries37 by spreading forces across various tissues.38 Alternatively, it has been suggested that human intrasubject variability exceeds the variability introduced by different footwear.25 Irrespective of the reason for different striking patterns, such differences may have confounded the effect of other independent variables such as footwear and mileage. While analysing only consistent rearfoot strikers can eliminate this confounding factor, results obtained from such studies cannot be generalised to non-rearfoot strikers and inconsistent rearfoot strikers, which composed 54% of participants in the present study. The high intrasubject variability in running pattern warrants further investigation.


This study demonstrated that running in worn shoes caused increased stance time and kinematic adaptations but did not change force variables suggesting that as shoe cushioning decreases, runners modify their patterns to maintain constant external loads. Further, there was no difference in any measured variables among the three shoe cushioning technologies (air/gel/spring) tested. Runners may choose to purchase shoes for reasons other than cushioning technology.

What is already known on this topic

Running mechanics are influenced by footwear midsole stiffness/geometry. Shoe cushioning degrades over time leading to changes in mechanical properties, although mechanical tests do not predict shock during actual running. It is also known that shoe degradation from machine-simulated running differs from that during actual loading by runners.

What this study adds

This is the first longitudinal study to describe the kinetics and kinematics of running in new and worn shoes. This study shows that as shoe cushioning capability decreases, runners modify their patterns to maintain constant external loads. The adaptation strategies due to shoe degradation were unaffected by different cushioning technologies.


The authors would like to thank Dr Joe Tomaka for his advice on statistics and Dr Stephen Burns and Dr Chantal Vella for providing invaluable feedback on this manuscript.



  • Funding The present study was financially supported by Spira Footwear and the College of Health Sciences, University of Texas at El Paso.

  • Competing interests None.