Impact loading in running shoes with cushioning column systems


This study evaluated the effects of running shoes–with two types of cushioning column systems–on impact force patterns during running. Kinematic and ground reaction force data were collected from 10 normal participants wearing shoes with the following cushions: 4-column multicellular urethane elastomer (Shoe 1), 4-column thermoplastic polyester elastomer (Shoe 2), and l-unit EVA foam (Shoe 3). Participants exhibited significantly lower impact force (p = .02) and loading rate (p = .005) with Shoe 2 (1.84 ± 0.24 BW; 45.6 ± 11.6 BW/s) compared to Shoe 1 (1.94 ± 0.18 BW; 57.9 ± 12.1 BW/s). Both cushioning column shoes showed impact force characteristics similar to those of a top-model running shoe (Shoe 3), and improved cushioning performance over shoes previously tested in similar conditions. Alterations in impact force patterns induced by lower limb alignment and running speed were negligible since participants did not differ in ankle position, knee position, or speed during all shod running trials. Ankle plantarflexion, however, was higher for barefoot running, indicating an apparent midfoot strike. Mechanical testing of each shoe during physiologic, cyclic loading demonstrated that Shoe 3 had the greatest stiffness, followed by Shoe 2 and Shoe 1. Shoe 1 was the least stiff of the two shoes with cushioning column systems, yet it displayed a significantly higher impact loading rate during running, possibly due to rearfoot motion alterations induced by the stiffer shoe. This study showed that even in similar shoe types, impact force and loading rate values could vary significantly with midsole cushion constructions. The findings of this study suggest that using these newer running shoes may be effective for runners who want optimal cushioning during running.

Key Words: impact force, loading rate, impact attenuation, spring-loaded shoes, rearfoot cushioning technology

During the impact phase of ground reaction force, the momentum from the decelerating limb rapidly changes as the foot collides with the ground, resulting in a transient force transmitted up the skeleton. In running, these forces can be up to three times body weight (Cavanagh & Lafortune, 1980). The repetitive transmission of these forces may contribute to degradation and overuse injuries (James, Bates, & Osternig, 1978). Athletic footwear with rearfoot cushioning systems has been proven to effectively attenuate impact forces experienced during barefoot heel-toe running (Clarke, Frederick, & Cooper, 1983), and consequently has been proposed as a mechanism for minimizing overuse injuries (Barnes & Smith, 1994).

In an effort to improve impact attenuation and durability, footwear manufacturers have adapted engineering concepts from other fields to design more effective rearfoot cushions. For example, the ShoxTM technology developed by Nike, Inc. (Beaverton, OR) incorporates a system of four spring-like columns made up of the same material found in jounce bumpers, which are shock absorbers used to cushion a car’s frame. This new cushioning technology was developed based on tuned running tracks, and it has demonstrated greater resiliency than other cushions used in popular running shoes (Valiant, Kilgore, Tawney, & McMahon, 2001). A similar cushioning technology was developed by Iso-Dynamics, Inc. (Cleveland, OH) using a resilient yet stiff elastomer for the midsole columns of a running shoe developed by DaDa Footwear (Los Angeles, CA).

While shoes with these cushioning columns may have undergone rigorous wear and biomechanical testing by their respective manufacturers, we found no evidence-based research in the scientific literature that investigated the effects of these advanced cushioning systems on impact forces or on running kinematics. With these new types of commercially available shoes, a biomechanical assessment may be beneficial for athletes who use such shoes in the hopes of minimizing injuries resulting from repeated impact loading. Therefore, the purposes of this study were to (a) evaluate the effects of two types of running shoes with advanced cushioning column systems on vertical ground reaction force patterns and materials characterization during running, and (b) compare them to those observed with a shoe constructed using a single rearfoot cushioning unit of viscoelastic foam.

Eight males and two females participated in the study after signing informed consent forms approved by the hospital’s institutional review board. All were screened with a musculoskeletal exam and were considered healthy, recreational runners (<10 miles per week), with no clinical signs indicating altered gait patterns. Their average height was 181 ± 4 cm and average body mass was 82 ± 4 kg.

Three commercially available running shoes were evaluated. Shoe 1 (Figure 1, top panel) was constructed with a set of four cushioning columns made of multicellular urethane elastomer arranged in an open configuration in the rear midsole (Nike, Inc.). Shoe 2 (Figure 1, bottom panel) had a similar cushioning column system, except that each column was built with a thermoplastic polyester elastomer molded into a hollow, bumper-like unit (Iso-Dynamics, Inc.). Shoe 3 was manufactured with a single midsole cushioning unit of proprietary ethyl vinyl acetate (EVA) (ASICS Tiger Corp., Kobe, Japan), considered to be a highly durable material (Whittle, 1999). The insole for Shoe 3, a top-model running shoe, was used for both Shoes 1 and 2.

To measure limb position at footstrike, we used 8 visible-red cameras (Motion Analysis Corp., Santa Rosa, CA) to capture coordinate data at a sampling rate of 120 Hz from a cluster-based marker set used to define the pelvis as well as bilateral thigh, shank, and foot segments. For the shod conditions, markers were placed on the heel counter and toe box of each shoe. Three floor-mounted force platforms (OR-6, AMTI, Watertown, MA), with a natural frequency of 450 Hz, were used to sample ground reaction force (GRF) data at 1,000 Hz. Analog signals were amplified using each platform’s associated signal conditioner (MCA-6, AMTI), connected via a 12-bit A/D card to a dual processor computer that was used to simultaneously capture GRF and marker data in real time.

Following a static calibration trial and a 5-min warm-up period, each runner repeatedly ran barefoot across a 12-meter runway at his or her preferred stride frequency, during which time motion and GRF data were collected. Three trials, involving complete right foot contact and without visual targeting, were collected for each pair of shoes, which were randomly selected by the lab technician.

GRF data were left unfiltered so as to preserve the high frequency components that may have indicated the presence of transient forces. Custom software was used to determine the following GRF parameters, as shown in Figure 2: Fz1, peak vertical GRF within 50 ms after footstrike (impact force); Rz1, loading rate of Fz1 calculated from the linear slope between footstrike and the onset of Fz1; Fz2, minimum vertical GRF; and Fz3, peak propulsive GRF. Using the Orthotrak software (Motion Analysis Corp.), we calculated sagittal knee and ankle kinematics after marker data were smoothed using a low-pass filter at a cutoff frequency of 12 Hz. Knee flexion angle and ankle plantar/dorsifiexion angle at footstrike were selected for subsequent analysis to address the effect of limb position on ground reaction forces.

Each parameter was statistically analyzed using a 4-level repeated-measures ANOVA with a post hoc Bonferroni-adjusted pairwise comparison test at a significance level of 0.05. To determine the repeatability of each participant’s running velocity, we computed the coefficient of variation, which is simply the variance-to-mean ratio, as described by Winter (1984).

In addition to biomechanical testing, a new shoe of each type was randomly chosen for cyclic testing using an MTS 858 Mini-Bionix servohydraulic testing machine (Eden Prairie, MN). The piston actuator of the machine (diameter = 2.5 cm) applied loads directly to the center of the rearfoot area in each shoe, simulating the primary weight-bearing area of the heel pad. The machine was tuned for high load, high frequency load control prior to testing, which involved a 5-Hz cyclic waveform operating between l0 N and 1,400 N of compressive load to approximate the impact loads of 2.5 times the average body weight of the runners. Displacement (mm) and load (N) were sampled at 100 Hz throughout each test using the TestWare 4.0C software manufactured by MTS. For each test, the last two of 100 cycles were used as data while previous cycles were used to precondition the system. Shoe stiffness was measured from the slope of a linear regression between 50 N and 1,400 N of the load-deflection curve.

The running velocity as well as the sagittal knee and ankle kinematics had no significant effect on the impact force parameters measured from the shod trials. Mean running velocity across all runners was 3.23 ± 0.02 m/s. No significant differences in knee flexion angle at footstrike were observed across all running sessions (p = .74). Conversely, runners exhibited significantly more ankle plantarflexion (p < .001) at footstrike during barefoot running compared to that of shod running, indicating an apparent midfoot strike. As a result, GRF data from the barefoot running trials were not analyzed. Likewise, data for one runner, who was observed to be a midfoot striker during all running sessions, were removed from the analysis. Ankle flexion at footstrike for the remaining 9 runners did not significantly differ across shod conditions (p = .70).

Differences in impact force parameters across all shoes could only be observed in magnitude (Fz1, p = .04) and loading rate (Rz1, p = .003) values, as shown in Table 1. No significant differences in the average minimum vertical (Fz2) or peak propulsive forces (Fz3) were found between shod conditions. Post hoc tests revealed that runners exhibited significantly lower impact forces (p = .02) and loading rates (p = .005) with Shoe 2 compared to Shoe 1, despite the fact that instrumented testing revealed Shoe 2 to have a higher stiffness (124 N mm-1) than Shoe 1 (92 N mm-1) (Figure 3). However, it is worth noting that the average impact force (Fz1) magnitudes were less than 2.00 BW for all shoe conditions, and the maximum difference in these values between all shoes was 0.10 BW. Pairwise comparison tests also revealed that GRF parameters for sessions performed in Shoe 3 were not statistically different (p >. 10) from those found in sessions with either Shoe 1 or 2, although Shoe 3 had the highest stiffness (137 N mm-1) of all shoes tested on the instrumented machine (Figure 3).

Due to the relatively recent launch of cushioning column shoes, this study represents the first comparative study on the cushioning performances of these new shoes. While there was a statistically significant difference in impact forces (Fz1) between both column-based shoes, the difference in average magnitude was within a 10th of the average body weight, and well within the range found in other popular running shoes that were previously tested using a similar ground reaction force analysis (Hennig, Milani, & Lafortune, 1993). Conversely, loading rates (Rz1) of running trials for both column-based shoes were considerably lower than those previously found with other shoes (Clarke et al., 1983; Hennig et al., 1993), particularly for the thermoplastic column-based shoe (Shoe 2). Such loading rate patterns represent the capacity of the cushion to reduce the rate at which the impact shock is transmitted to the lower extremity, and perhaps are better indicators of cushioning performance than peak impact forces. The fact that the loading rate of Shoe 2 was significantly less than that of Shoe 1 reinforces previous findings that changes in the foot/ground interface influence loading rates more than they do peak impact forces (Clarke et al., 1983; Lafortune, Hennig, & Lake, 1996). Hence, differences in midsole construction would appropriately affect impact loading rate, as interpreted by cushioning performance in this study.

Although Shoes 1 and 2 were similar in column configuration, the columns in Shoe 2 were constructed with a thermoplastic polyester elastomer material molded into a hollow bumper with a variable wall thickness increasing from top to bottom (Iso-Dynamics, Inc.). Shoe 1, on the other hand, was constructed with a system of four rearfoot cushions made up of highly resilient urethane foam with a higher radial thickness than that of Shoe 2 (Valiant et al., 2001). Oddly, instrumented cyclic testing conducted in the present study revealed Shoe 1 to be the softest shoe despite exhibiting relatively higher values of impact force peak and loading rate when compared to Shoe 2. This is in contrast to previous findings, which have shown that softer shoes normally have lower impact loading rates (Clarke et al., 1983; Lafortune et al., 1996). However, previous instrumented shoe testing was primarily performed using an impact tester (Cavanagh, 1980), which differs from the cyclic testing used in this study. It has generally been agreed that instrumented testing does not accurately approximate the in-vivo loading experienced during running (Cavanagh & Lafortune, 1980). Therefore, the load distribution at the midsole cushioning region during footstrike could be remarkably different from that found during instrumented testing, leading to the observed discrepancies in stiffness and impact force patterns.

Running velocity as well as knee and ankle plantar/dorsiflexion angles at footstrike were highly consistent across all runners during all sessions. Therefore, any differences observed in the GRF parameters were most likely caused by changes in the foot/ground interface (i.e., shoes) and not by gait adaptations induced by footwear. However, the cushioning column design may have induced changes in normal rearfoot motion, thus potentially affecting the impact forces observed (Perry & Lafortune, 1995). With these systems, only four cushioning units are found in the midsole as opposed to the thousands of closed air cells found in single-unit EVA cushions. Thus, column-based systems behave much like an independent suspension system, wherein each column is capable of providing a certain level of cushioning and collectively make up the overall stiffness of the shoe’s midsole. Hennig, Valiant, and Liu (1996) concluded that runners who wear harder shoes tend to alter their landing pattern to elicit lower impact forces. Unfortunately, due to the lack of a standardized foot model (Areblad, Nigg, Ekstrand, Olsson, & Ekstrom, 1990), differences in rearfoot motion were not specifically addressed in this study. The potential variations in rearfoot motion, combined with the differences in midsole properties between the two constructs, could explain the cushioning differences found between the two column-based shoes.

Impact forces during running are of considerable interest because of their potential contributions to overuse injuries (Barnes & Smith, 1994). The new rearfoot cushioning technology tested in this study was first introduced by Nike, Inc., as the ShoxTM technology, and its development was based on tunable running tracks designed to reduce impact forces (McMahon & Greene, 1979; Valiant et al., 2001). The present study showed that this new technology was in fact highly compliant and comparable to current cushioning systems. However, impact forces and loading rates could vary with midsole material properties as a result of alterations in rearfoot motion, and should be addressed in subsequent studies. Nevertheless, both cushioning column-based shoes attenuated impact to a level and at a rate similar to those of a top-model running shoe. For athletes wanting shoes with maximum midsole cushioning, this study has provided relevant information about impact loading on shoes with relatively new cushioning column systems.

The authors are with the Center for Human Performance, and Mahar is also with the Orthopedic Biomechanics Research Center, Children’s Hospital San Diego, 3020 Children’s Way, San Diego, CA 92123.

This study was supported in part by the David H. Sutherland Endowment Fund. The authors would like to thank Iso-Dynamics, Inc., and LL International, LLC, for their contributions, as well as Abigail Morales and Michael Litavish for their technical support.