Secondary Logo

Share this article on:

Impact shock and attenuation during in-line skating


Medicine & Science in Sports & Exercise: August 1997 - Volume 29 - Issue 8 - p 1069-1075
Applied Sciences: Biodynamics

Although impact and shock attenuation associated with foot contact during running has been studied extensively, much less is known about these phenomena during in-line skating (ILS). The purpose of this study was to describe these impact characteristics for ILS and to test the hypothesis that there is lower impact shock during ILS than in running at preferred velocities. Subjects ran and skated on a treadmill at preferred velocity, with low-mass accelerometers attached to both the distal tibia and head. Tibia and head acceleration data during stance were used to calculate peak acceleration (PA), peak frequency(PF), and median frequency (MedF). Impact attenuation (IA) between the tibia and head was quantified by comparison of PA values and by calculating a transfer function between the head and tibia frequency spectra. PA, PF, and MedF values were significantly lower in ILS than in running for both tibial and head data, whereas IA was similar between the two movements. The ILS condition exhibited almost no power in the frequency range between 10 and 20 Hz, which has been associated with the foot impacting the ground. It is concluded that in-line skating results in less impact shock to the body with each foot contact, and may be a useful exercise modality for those wishing to reduce impact shock during aerobic training.

Department of Exercise Science, University of Massachusetts, Amherst, MA 01003

Submitted for publication September 1996.

Accepted for publication May 1997.

The authors would like to thank Rollerblade, Inc. for financial support of the project, Dr. Patty S. Freedson for the use of her laboratory and treadmill, and Ed Melanson for subject recruitment.

Address for correspondence: Graham E. Caldwell, Ph.D., Biomechanics Laboratory, Department of Exercise Science, Totman 105, University of Massachusetts, Amherst, MA 01003. E-mail:

Walking and running are well known as exercises that can improve cardiovascular endurance, strengthen muscles, and reduce body fat. An alternative activity that has grown in popularity over the past few years is in-line skating (ILS). The ILS motion appears to have greater range of motion at some joints, with a smooth transition between contralateral skates and therefore may offer benefits over more traditional aerobic exercises. Among these benefits may be biomechanical advantages such as reduced joint loading. However, there are few studies on ILS found in the literature, and those have focused on physiological responses (8,9). The biomechanical aspects of this movement have yet to be revealed and could provide insight to benefits or disadvantages in comparison to walking or running.

One aspect of walking and running which has been well-documented is the impact shock transmitted through the body when the foot collides with the ground. This event occurs roughly 6,000 times in the course of a 10-km running race (16). The shock wave is transmitted across the structures of the lower extremity and upward through the spine to the skull. These impact characteristics during walking and running have been well studied in the biomechanics literature(1,4,6,14,17). One positive aspect of these impacts is their role in providing loading for the bones of the lower extremities, necessary for bone remodeling. However, repetitive impacts and the resulting shock waves have also been implicated in degenerative diseases of the knees, hips, and spine(15). Analysis of impacts and the propagated shock wave have also been used to determine characteristics of pathologies in the lower extremity (18).

In running, peak impact accelerations measured on the distal tibia near the ankle have ranged from 4 to 10 g (4,6,17). Variations in these impact values have been attributed to the mass of the accelerometer, the method of attachment, running velocity, and accelerometer design(5,16). As the shock wave associated with footstrike moves upward through the body, it is attenuated by both active and passive mechanisms. The attenuation during running and walking has been cited between 70 and 80%, so that peak accelerations registered at the head range from 1 to 3 g (4,14,17). Limiting the shock experienced by the head may be important to limit disturbances of the visual and vestibular sensory organs(4,11,12).

Impact shock has also been characterized using frequency analyses. Fast Fourier transformation (FFT) techniques have been performed on time domain signals to determine the frequency content of the acceleration signal(4,14). In running, the frequency spectrum displays two peaks: a low frequency “active” peak at about 4 to 8 Hz, associated with the movement of the body during stance, and a second“impact” peak at 12 to 20 Hz (14). Differences in frequency content between tibial and head accelerations show gains in spectral power to the head below 5 Hz and reductions to the head at frequencies above this value (4,14).

With the dramatic increase in the popularity of in-line skating, the issue of impact shock and attenuation in ILS has become important. Therefore, the purpose of this study was to examine impact shock and the attenuation characteristics during in-line skating. To that end we measured impact accelerations at the tibia and head during ILS and compared these results with data collected from the same subjects while running. Because of the absence of a flight phase and the smooth transition from swing to stance phase in ILS, it was hypothesized that the impact shock would be lower during in-line skating than in running.

Back to Top | Article Outline


Eleven subjects (mean ± SD age: 21 ± 2 yr.; height: 163± 7 cm.; mass: 62 ± 8 kg.) were recruited from the university community (4 males, 7 females). Each subject read and signed a physical activity readiness questionnaire (PAR-Q), and informed consent documents before testing. All subjects were trained in-line skaters who had successfully completed an 11-wk training study (8). The sample size(N = 11) was established using the Cohen case 4 method, using variability estimates from a previous study on running impact(4).

To eliminate surface irregularities and changes in progression velocity, and to provide a consistent environment for data collection, all testing was completed on an extra wide (≅2 m) motorized treadmill that would accommodate both running and skating movements. Subjects were given an acclimatization period for both running and skating conditions before testing. Despite the warm-up period, skating on the treadmill at level grade created balance problems and unnatural skating mechanics for the subjects. Changing the grade to 2% uphill produced a marked improvement in both balance and skating mechanics. Subjects reported that the 2% grade felt similar to skating on level ground outdoors. This self-assessment regarding skating mechanics was confirmed by experienced skaters observing the trials.

Subjects performed trials in each of three conditions: 1) skating at a 2% grade, 2) running at a 2% grade, and 3) running at a 0% grade. During each condition, subjects skated or ran for 5 min at their preferred velocity and stride frequency. Preferred progression velocities were determined before testing by increasing the treadmill velocity until subjects confirmed that the speed was comparable to their own training pace. The preferred velocities found in the warm-up period were used in the appropriate conditions during the subsequent testing. Acceleration data were measured with 1.7-g uniaxial accelerometers (model 353B17, PCB Piezotronics Inc., Depew, NY) attached to small (2.1-g) aluminum mounting brackets to the anteromedial side of the distal portion of the tibia and to the forehead, using elastic strapping tightened to the limit of subject tolerance. The frequency range of the accelerometers was 0.35 Hz to 30 kHz, with a resonant frequency greater than 70 kHz and a resolution of 0.01 g. Accelerometer signals were amplified and sampled at 1000 Hz with a 12-bit A/D converter. During each 5-min trial, data were collected for 5 s at minutes 3 and 4. Each of the two data collection trials contained data from several consecutive strides.

Pilot work revealed high frequency vibration of the accelerometer during skating due to the skate wheels rolling over the rough treadmill surface. Also, a metal fastener (zipper) that connected the two ends of the treadmill mat caused an impulse response when struck by the skate wheels (seeFig. 1, top panel). Since these responses were not part of the normal impact of the skate on the treadmill, they were considered noise. This noise was examined by collecting accelerometer data from a skater standing on the treadmill while the treadmill belt was moving at the preferred skating velocity. Frequency analysis of the noise revealed content within a 40- to 80-Hz range, with peak power at roughly 50 Hz(Fig. 1, middle panel). Because this is well above the frequency range associated with the impact in running, an 8th order, cascaded, 40-Hz low-pass Butterworth filter was used to remove the noise from the accelerometer signal in all trials for each condition(Fig. 1, bottom panel).

The impact shock in each trial was quantified by measuring the peak acceleration (PA) that occurred just after foot contact in each stance phase. The attenuation of the impact shock wave between the tibia and head was calculated in the time domain using these peak impact accelerations. The attenuation was calculated as: Equation

where IA is the impact attenuation, PAhead is the peak head acceleration, and PAtibia is the peak tibial acceleration. Attenuation was also examined in the frequency domain using a transfer function technique(4,14). The time series for each stance period was used to generate a power spectral density (PSD) function using FFT techniques. The transfer function was calculated as: Equation

where T is the gain or reduction of signal in decibels (dB), and PSDhead, and PSDtibia are the head and tibial PSD curves, respectively.

The time domain variables output for statistical analysis were PAhead, PAtibia, and IA. Also, the peak (PF) and median (MedF) frequencies were calculated from the PSD curves of both the tibia and head. The main effect of condition for these dependent variables was tested using a within-subject, repeated-measures ANOVA. The transfer functions (T) for each condition were assessed qualitatively, and those for running were compared with prior studies.

Back to Top | Article Outline


The preferred velocities chosen by the subjects differed between skating and running. The mean value for the skating condition was 7.4 mph (SD ± 0.9), whereas for the running conditions the mean velocity was 6.0 ± 0.7 mph. The progression velocities in the 2% running condition were set equal to the preferred 0% running condition. This was done to ascertain the effect of the change in grade on the impact accelerations. The higher skating velocity is noteworthy because Clarke et al. (1) demonstrated that peak leg shock values in running increased by 34% for each 1-ms-1 increase in progression velocity.

Typical examples of tibial accelerometer signals during the stance phase are shown for both skating and running in Figure 2. Tibial peak accelerations (PAtibia) occurred just after foot contact for both activities. Note that the time of support lasted much longer for ILS(≅1000 ms) than for running (≅300 ms). The mean PAtibia value for the skating condition was 2.02 ± 0.63 g, whereas in both running conditions the mean was above 4.00 g (running 0% = 4.25± 1.43 g, running 2% = 4.22 ± 1.60 g). Repeated-measures ANOVA results showed a significant effect of condition(P ≤ 0.0001; F = 15.71; df = 2) and Tukey'spost-hoc analysis found that the peak impact for skating was significantly less than for running in either the 0% or 2% conditions. For this and all other dependent variables, there was no significant difference between the two running conditions. For clarity, only the 2% grade results will be discussed for the other dependent variables. Table 1 displays a summary of all dependent variables and the results of statistical tests between conditions.

Frequency analysis of the tibial accelerometer signals for running displayed peaks at 6-8 Hz and 12-14 Hz with magnitudes of ≅0.1g 2·Hz-1 (Fig. 3). The PSD for skating showed a single peak at 1-3 Hz with a lesser magnitude of 0.025g 2·Hz-1. The mean PF at the tibia for skating(1.5 ± 0.7 Hz) was less than that found for running (8.8 ± 3.1 Hz). The ANOVA and Tukey's post-hoc test found this difference to be significant (P ≤ 0.0001; F = 42.3; df = 2). The mean MedF found from the PSD curves was 7.2 ± 3.1 Hz for skating and 12.5 ± 2.1 Hz for running. Again, MedF from the skating condition was found to be significantly lower than in running (P ≤ 0.0001;F = 20.9; df = 2).

Figure 4 presents an exemplar time series of stance phase head acceleration data for the skating and running conditions. The mean PAhead for skating was 0.45 ± 0.17 g, whereas for running the mean value was 0.91 ± 0.30 g. Thus, for both the tibial and head accelerations, the skating impact peak was roughly half the magnitude of the impact registered during the running conditions. Statistical analysis of the PAhead indicated that the peak in skating was significantly lower than in the running conditions (P ≤ 0.0001;F = 42.30; df = 2).

As with the tibia, the PSD functions for the head (Fig. 5) were nearly identical for the two running conditions, with a single peak of 0.16 g 2·Hz-1 at ≅3-4 Hz. The second peak found in the tibial PSD at 8-14 Hz has been almost completely attenuated in the head accelerations. The skating condition displays a single peak at 1-3 Hz with a small magnitude of 0.025 g 2·Hz-1. The mean PF at the head was 1.4 ± 0.6 Hz for skating and 3.2 ± 0.5 Hz for the running condition. Statistical analysis indicated that the head PF for skating was significantly lower than the running conditions (P≤ 0.0001; F = 60.78; df = 2). For MedF in the head acceleration, the skating condition (2.1 ± 0.6 Hz) displayed lower values than in running(3.6 ± 0.4 Hz); this difference was also significant (P ≤ 0.0001; F = 61.02; df = 2).

Mean attenuation values (IA) calculated from the peak accelerations at the tibia and head for skating and running ranged from 73 to 76% and were not found to be statistically different. In the frequency domain, attenuation was apparent in the shift of peak and median frequencies between the tibia and head (Fig. 6). MedF for tibial shock shifted to lower frequencies at the head in both skating and running, whereas PF shifted downward in the running trials only. Another indication of attenuation can be seen in the transfer function plots (Fig. 7), which illustrate differences in spectral power between tibia and head at specific sites across the frequency spectrum. The running conditions display nearly identical curves with gains as high as 22 dB below 5 Hz and with attenuation at all frequencies above 5 Hz. In the range of impact frequencies from 10 to 20 Hz there is considerable attenuation, as high as -40 dB. The skating condition shows attenuation at all frequencies, with a maximum of -35 dB occurring at ≅10 Hz.

Back to Top | Article Outline


The results of this study indicate that in-line skating creates impacts of lesser magnitude than running at preferred progression velocities. The ILS condition produced peak accelerations that were roughly half of those found during running, whether measured at the tibia or the head. In both ILS and running, the body attenuated the impact measured at the tibia before it reached the head. Frequency analysis indicated that the skating condition produced accelerations with less power across the entire frequency spectrum when compared with running, particularly in the 10-20 Hz range associated with the foot striking the treadmill bed. The transfer function used to examine attenuation at discrete frequencies showed that both the running and ILS conditions attenuated power between the tibia and head in this 10-20 Hz impact range. Based on these results, the hypothesis that ILS involves less impact shock than running at preferred velocities is accepted.

Although there are no previous data on impacts during in-line skating, our running results are comparable to those found in other studies. The PAtibia values of 4 g in running correspond with those found in some studies (4) but are lower than the 10-18g reported in others (14). These differences among studies may result from several factors, including compliance of the treadmill bed and/or shoe. Other factors such as device mass and the attachment method are known to influence the acceleration signal due to the effect of resonant oscillations (3). The accelerometer and bracket mass in the present study (3.8 g) was lower than the 4.4-g accelerometers used in some other studies(14,16). Further, the tight elastic strapping used in the present case produced no discernible changes in mechanics nor pain during movement, yet prevented excess motion of the accelerometer and underlying tissue. Low device mass and secure attachment have been shown to reduce the possibility of resonant oscillations (3).

The fact that the two running conditions gave similar results illustrates that the 2% grade was not a significant factor in modifying our impact results. Although Valiant (16) reported a change in impact values at more severe grades (up to 9% incline), our results concur with his data from more shallow inclines. Peak head accelerations during running were also similar to other reports using the same forehead attachment method (4). Studies with larger accelerometers or with bite bar attachment have reported greater head accelerations, as high as 2g (7,14). Finally, the degree of attenuation of peak acceleration between the tibia and head in the running conditions agrees with previous studies(4,14,16,18).

The results of the frequency analysis for the running trials also agree with other studies. The running condition displayed two peaks in the PSD plot for the tibia, one at 6-8 Hz and the other at 10-14 Hz, as found in previous studies (4,14). Further analysis of the frequency data in running produced transfer functions between the tibia and head that were similar to those reported elsewhere, with a gain in power below 5 Hz and attenuation of all higher frequencies (4,14).

Back to Top | Article Outline

Running Versus ILS

One of the central themes of this study was the comparison of in-line skating with running. One salient difference was that subjects chose to skate at a higher velocity than they chose to run. Interestingly, this increased velocity did not result in higher peak tibial accelerations, as has been reported in previous studies when running speed was increased(1). The lower impact during skating likely results from the greatly reduced or nonexistent flight phase during skating. This reduces the change in vertical displacement of the center of mass during the skating stride, and also lowers the vertical velocity of the foot at footstrike. The fact that the foot must roll forward during the initial portion of stance probably induces the skater to adopt a technique that emphasizes horizontal rather than vertical velocity at foot contact. This results in a smoother transition from swing to stance and a much lower impact for the body to attenuate, despite the higher progression velocity.

The frequency analysis of tibial accelerations also showed that the skating condition demonstrated less impact acceleration, with lower values than running for both peak and median frequency. The significance of the ILS power spectrum is both in its low magnitude single peak at ≅1.5 Hz and in the absence of any spectral power between 10 and 20 Hz. This frequency range is associated with the collision of the foot with the ground, and it has been suggested that this frequency band may be implicated in musculoskeletal injury(10,13). Some have suggested that removal of higher frequencies may reduce the likelihood of degenerative joint diseases(18). In-line skating may therefore prove to be an aerobic exercise workout that is less harmful to the joints in terms of repetitive impacts of the foot and ground.

As with the tibia, the head experienced 50% lower peak acceleration values during in-line skating than in running. These reduced magnitudes at the head may result both from the absence of a flight phase and the differences in foot velocity at contact between running and ILS as outlined above. Another contributor to the lower head accelerations may be greater knee flexion during the skating trials. In running, higher knee flexion angles have been shown to increase shock attenuation (7). Preliminary kinematic data from our lab (unpublished observations) suggest that ILS exhibits greater knee flexion than does running. Interestingly, increased knee flexion has also been associated with higher oxygen cost (2,7), yet trained subjects have higher oxygen demands during running as compared with in-line skating (8).

The attenuation of impact shock between tibia and head may have clinical importance. Although some level of impact shock is necessary for bone remodeling, higher levels of impact shock have been implicated in joint degeneration of the knees, hips and spine (13). It is interesting that despite greatly different magnitudes of impact peaks at the tibia for running versus ILS, the amount of attenuation between the tibia and head is nearly identical. Viewing the attenuation through the transfer function plots tells a slightly different story. In contrast to running, the skating condition displayed attenuation over the total frequency spectrum, including the so-called “active” portion below 5 Hz. This is likely because of the relatively small low frequency peak at the tibia, which is easily damped by the body before reaching the head. Because ILS does not require a large change in vertical velocity during the stride, the body can easily absorb the tibial acceleration with no disruption in head motion. Therefore, although the amount of attenuation is the same in terms of reduction in peak accelerations, there are different “input” frequencies at the tibia for the body to deal with.

In both running and in-line skating, the body seems to be attempting to minimize head acceleration for the current loading condition. Minimal head acceleration may improve locomotor performance by providing a stable visual field, reduce vestibular excitation, and allow for a stable gravitational reference (11,12). This may be valuable during in-line skating at high velocity, especially if the athlete's balance on skates is less stable than while running. It would be interesting to examine elite, highly skilled skaters to see if they forego some of the attenuation to increase skating velocity as balance becomes a less important issue.

Overall, using measurements in both time and frequency domains, in-line skating demonstrates considerably less impact shock than does running. When one considers the high aerobic benefits possible with ILS, these reduced impact characteristics indicate that in-line skating may be used as a viable training method without subjecting the athlete to excess musculoskeletal loading from footstrike. Further work is needed to examine the kinematic, kinetic, and neuromuscular characteristics of in-line skating to fully appreciate its potential as a training modality.

Figure 1-

Figure 1-

Figure 2-Exemplar time series of tibial acceleration signals during stance phase of running (

Figure 2-Exemplar time series of tibial acceleration signals during stance phase of running (

Figure 3-Exemplar power spectral density (PSD) curves for the tibial accelerations during each condition.

Figure 3-Exemplar power spectral density (PSD) curves for the tibial accelerations during each condition.

Figure 4-Exemplar time series of head acceleration signals during stance phase of running (

Figure 4-Exemplar time series of head acceleration signals during stance phase of running (

Figure 5-Exemplar PSD curves for the head accelerations during each condition.

Figure 5-Exemplar PSD curves for the head accelerations during each condition.

Figure 6-

Figure 6-

Figure 7-Transfer functions between tibial and head accelerations for each movement condition.

Figure 7-Transfer functions between tibial and head accelerations for each movement condition.

Back to Top | Article Outline


1. Clarke, T. E., L. B. Cooper, D. E. Clark, and C. L. Hamill. The effect of increased running speed upon peak shank deceleration during ground contact. In: Biomechanics IX-B, D. Winter, R. Norman, R. Wells, K. Hayes, and A. Patla (Eds.). Champaign, IL: Human Kinetics, 1985, pp. 101-105.
2. Frederick, E. C., T. E. Clarke, J. L. Larsen, and L. B. Cooper. The effects of shoe cushioning on the oxygen demands of running. In:Biomechanical Aspects of Sport Shoes and Playing Surfaces, B. Nigg, and B. Kerr (Eds.). Calgary: University of Calgary. 1983, pp. 107-114.
3. Gross, T. S. and R. C. Nelson. The shock attenuation role of the ankle during landing from a vertical jump. Med. Sci. Sports Exerc. 20:506-514, 1988.
4. Hamill, J., T. R. Derrick, and K. G. Holt. Shock attenuation and stride frequency during running. Hum. Mov. Sci. 14:45-60, 1995.
5. Hennig, E. M. and M. A. Lafortune. Tibial bone and skin accelerations during running. In: C. E. Cotton, M. Lamontagne, D. G. E. Robertson, and J. P. Stothart (Eds.). Proceedings of the Fifth Biennial Conference and Human Locomotion Symposium of the Canadian Society for Biomechanics. Ottawa: University of Ottawa, 1988, pp. 74-75.
6. Light, L. H., G. E. McLellan, and L. Klenerman. Skeletal transients on heel strike in normal walking with different footwear. J. Biomech. 13:477-480, 1980.
7. McMahon, T. A., G. A. Valiant, and E. C. Frederick. Groucho running. J. Appl. Physiol. 62:2326-2337, 1987.
8. Melanson, E. L., P. S. Freedson, and S. Jungbluth. Changes in ˙VO2max and maximal treadmill time after nine weeks of running or in-line skate training. Med. Sci. Sports Exerc. 28:1422-1426, 1996.
9. Melanson, E. L., P. S. Freedson, R. Webb, S. Jungbluth, and N. Kozlowski. Exercise responses to running and in-line skating at self-selected paces. Med. Sci. Sports Exerc. 28:247-250, 1996.
10. Nigg, B. M., G. K. Cole, and G. P. Bruggerman. Impact forces during heel-toe running. J. Appl. Biomech. 11:407-432, 1995.
11. Pozzo, T., A. Berthoz, and L. Lefort. Head kinematics during various motor tasks in humans. Prog. Brain Res. 80:377-383, 1990.
12. Pozzo, T., A. Berthoz, L. Lefort, and E. Vitte. Head stabilization during various locomotor tasks in humans. I. Patients with bilateral vestibular deficits. Exp. Brain Res. 85:208-217, 1991.
13. Radin, E. L., H. G. Parker, J. W. Pugh, R. S. Steinberg, I. L. Paul, and R. M. Rose. Response of joints to impact loading-III. J. Biomech. 6:51-57, 1973.
14. Shorten, M. R. and D. S. Winslow. Spectral analysis of impact shock during running. Int. J. Sports Biomech. 8:288-304, 1992.
15. Simon, S., E. L. Radin, and I. L. Paul. The response of joints to impact loading-II: in-vivo behavior of sub-chondral bone.J. Biomech. 5:267-272, 1972.
16. Valiant, G. A. Transmission and attenuation of heel strike accelerations. In: The Biomechanics of Distance Running. P. R. Cavanagh (Ed.), Champaign, IL: Human Kinetics, 1989, pp. 225-247.
17. Valiant, G. A., T. A. McMahon, and E. C. Frederick. A new test to evaluate the cushioning properties of athletic shoes. In:Biomechanics X-B, B. Jonsson (Ed.). Champaign, IL: Human Kinetics, 1987, pp. 937-941.
18. Voloshin, A. S., C. P. Burger, J. Wosk, and M. Arcan. An in-vivo evaluation of the leg's shock absorbing capacity. In:Biomechanics IX-B, D. Winter, R. Norman, R. Wells, K. Hayes, and A. Patla (Eds.). Champaign, IL: Human Kinetics, 1985, pp. 112-116.


©1997The American College of Sports Medicine