Monday, September 25, 2017



Matthew B. Yandell, Karl E. Zelik
Vanderbilt University, Nashville, TN USA

Humans value economy of locomotion and seem to
adopt a step frequency while shod that minimizes
metabolic cost at a given walking speed [1]. It has
also been observed that humans increase their step
frequency when walking barefoot (as compared to
shod) [2,3]; however, the reason for this increase
has not been explained.

Taken together, these empirical observations indicate
that either: (1) removing the shoes changes gait
dynamics sufficiently such that it shifts the metabolic
minimum step frequency for a given walking
speed, or (2) humans adopt a new step frequency
when walking barefoot that does not minimize metabolic

The goal of this study was to determine which of
these two possibilities better explains the shift in
step frequency when walking barefoot versus shod.
We performed an experiment that compared shod
versus barefoot walking, and estimated metabolic
power as well as the relative effects of shoe-specific
properties (e.g., mass, height) on step frequency. A
secondary goal was to investigate changes in center
of mass (COM) mechanics for shod versus barefoot

We studied 5 subjects (mean ± SD, 23 ± 4.3 years
old, 72.3 ± 16.0 kg, 177.4 ± 11.0 cm height) during
level walking on an instrumented split-belt treadmill
(Bertec). All subjects gave informed written consent
prior to participation. Three footwear variants were
tested: shod (using each subject’s personal athletic
shoes), barefoot, and weight-matched (WM) barefoot
(ankle mass was added to match shoe mass).
Each subject performed acclimation trials at 1.25
m/s to determine their baseline self-selected step
frequency for each footwear variant.

Next, each subject was asked to walk at 1.25 m/s
while matching a metronome frequency for the five
testing conditions in Table 1. A Cosmed K4b2 system
measured metabolic data for six minutes, and
ground reaction force (GRF) data was collected during
the last minute of each trial. Average metabolic
power during the last 2.5 minutes of each trial was
calculated from the equation given by Brockway
[4]. GRF data was low-pass filtered at 25 Hz (Butterworth,
3rd order, zero lag) and used to compute
individual limb COM power [5] for each limb. Statistical
comparisons between conditions were performed
using analysis of variance with Holm-Sidak
step down correction (alpha = 0.05).

Table 1: Conditions for each footwear variant
Footwear Variant
Step Frequency Shod Barefoot Barefoot WM
1.Shod SS X X X
2.Barefoot SS X
3.Barefoot WM SS X

We also estimated the effect of leg length difference
(barefoot vs. shod, due to shoe height) on step frequency
using regression equations reported in literature

We found that the barefoot and barefoot WM selfselected
step frequencies were significantly higher
than the shod self-selected frequency (P < 0.01, Table
2). No significant difference in step frequency
was observed between barefoot and barefoot WM
(P = 0.62). Mean shoe mass in this study was 296 g.

We observed from metabolic data that 4 out of 5
subjects exhibited a lower gross metabolic power
when walking barefoot at the shod self-selected step
frequency (~112 steps/min, Table 2) than barefoot
at the barefoot self-selected frequency (~121
steps/min). This difference did not reach statistical
significance, likely due to the low number of subjects.
However, through ongoing testing of additional
subjects we will attempt to distinguish if statistically significant differences exist.

The height of the shoe sole added 20 ± 6 mm (mean
± SD) to each participant’s leg length, as compared
to walking barefoot. Based on the published speedstep
length relationship [2,6], we estimated this 20
mm decrease in leg length when walking barefoot
would result in a 1.9 step/min increase in step frequency.
The predicted increase in step frequency
was substantially smaller than the empirical increase
of 8.4 step/min (Table 2) observed in this study.

Differences were also observed in COM power for
barefoot vs. shod gait, specifically in terms of a reduced
transient immediately following heelstrike
during barefoot walking.

Table 2: Self-selected (SS) step frequency for each
footwear variant/
Subject SS Step Frequency (steps/min)
Variant 1 2 3 4 5 AVG
Shod 112 123 103 112 111 112.2
Barefoot 118 130 118 119 118 120.6
WM 117 130 118 120 117 120.4

There are multiple factors which could contribute to
the increase in step frequency when a person transitions
from shod to barefoot walking. In this study
we sought to determine if properties of the shoe
could account for the altered dynamics.

We found that the addition of shoe mass to barefoot
walking had little to no effect on self-selected step
frequency, and thus did not account for the observed
barefoot vs. shod changes. Similarly, we estimated
that the reduction in leg length when walking barefoot
(due to the loss of shoe height) was only expected
to lead to a small increase in step frequency
(~1-2 steps/min), but this accounted for less than
25% of the observed barefoot vs. shod step frequency
increase, which was >8 steps/min.

Since these intrinsic shoe characteristics failed to
account for the observed change in step frequency,
other factors must be considered. One of these is
shoe length (and its effect on effective foot length),
which will be investigated in ongoing/future trials.
Another factor may be related to the subjective
comfort/discomfort experienced as a result of not
having shoe cushioning when walking barefoot.

Although it is difficult to define comfort, the lack of
cushioning when walking barefoot could have a
marked effect on a user’s subjective preference and
lead to altered behavior.

In summary, we found indications that when walking
barefoot, people may choose a self-selected step
frequency that is metabolically sub-optimal. The
increased step frequency observed for barefoot (as
compared to shod) walking was not well explained
by the mass or height properties of the shoe. We
speculate that cushioning provided by the shoe
and/or subjective user comfort may be significant
factors in the choice of self-selected barefoot walking
frequency, factors which are generally not integrated
into biomechanical walking models or our
theoretical understanding of gait. Shoe characteristics
could potentially be manipulated to encourage
desired locomotor behaviors, such as more favorable
biomechanics or reduced joint loading associated
with long-term injury risk.

1. Zarrugh MY, et al. Eur. J. Appl. Physiol. 33, no.
4, pp. 293–306, 1974.
2. Grieve DW, et al. Ergonomics 9, no. 5, pp. 379–
399, 1966.
3. Lythgo N, et al. Gait and Posture 30, no. 4, pp.
502–506, 2009.
4. Brockway JM. Hum. Nutr. Clin. Nutr. 41, no. 6,
pp. 463–471, 1987.
5. Donelan JM, et al. J. Biomech. 35, no. 1, pp. 117–
124, 2002.
6. Kuo AD. J. Biomech. Eng. 123, no. 3, pp. 264-
269, 2001.

No comments: