BJMB
Brazilian Journal of Motor Behavior
Special issue:
Effects of aging on locomotor patterns
!
Santos et al.
2022
VOL.16
N.5
372 of 384
Walking speed does not affect age-differences in ankle muscle beta-band intermuscular
coherence during treadmill walking
PAULO C. R. SANTOS
1,2
| INGE ZIJDEWIND
3
| CLAUDINE LAMOTH
4
| LILIAN T. B. GOBBI
5,#
| TIBOR HORTOBÁGYI
4,6,7,8
1
Department of Computer Science & Applied Mathematics, Weizmann Institute of Science, Rehovot, Israel.
2
Center of Advanced Technologies in Rehabilitation, Sheba Medical Center, Ramat Gan, Israel.
3
Department of Biomedical Sciences of Cells and Systems, University Medical Center Groningen, University of Groningen, Groningen, Netherlands.
4
Department of Human Movement Sciences, University Medical Center Groningen, University of Groningen, Groningen, Netherlands.
5
São Paulo State University (UNESP), Institute of Biosciences, Posture and Gait Studies Laboratory (LEPLO), Rio Claro, SP, Brazil.
6
Institute of Sport Sciences and Physical Education, University of Pécs, Pécs, Hungary.
7
Department of Kinesiology, Hungarian University of Sport Science, Budapest, Hungary.
8
Somogy County Kaposi Mór Teaching Hospital, Kaposvár, Hungary.
#
In memorium
Correspondence to:!Paulo C. R. Santos, Department of Computer Science & Applied Mathematics, Weizmann Institute of Science, Rehovot, Israel
email: paulo-cezar.rocha-dos-santos@weizmann; paulocezarr@hotmail.com
https://doi.org/10.20338/bjmb.v16i5.323
HIGHLIGTHS
• Fast walking speed was ~9% slower in older
vs. younger individuals.
• Aging affected ankle but not thigh muscle beta-
band coherences.
• Older vs. younger individuals walked with
~53% lower ankle muscle beta-band
coherence.
• Walking speed did not affect age-differences
in ankle muscle beta-band coherence.
ABBREVIATIONS
BF Biceps femoris
CoP Center of pressure
CV Coefficient of variation
d Effect sizes
GL Gastrocnemius lateralis
PL Peroneus longus
RF Rectus femoris
SL Soleus
SPPB Short Physical Performance
Battery
ST Semitendinosus
TA Tibialis anterior
VL Vastus lateralis
η²
p
Partial eta squared
PUBLICATION DATA
Received 08 11 2022
Accepted 11 12 2022
Published 15 12 2022
BACKGROUND: By examining whether age and speed each differently affect beta coherence during walking, we
can extend the limited evidence on age-related impairment in the neural control of walking.
AIM: We determined the effects of age and walking speed on intermuscular beta coherence between lower
extremity muscle pairs and the association between stride characteristics and intermuscular beta coherence in
these muscle pairs.
METHOD: Older (n=12) and younger (n=14) individuals walked on a treadmill at fixed (1.2 m/s) and fast (~1.3x
preferred) speeds for 3min. For 100 dominant leg strides, we measured length, width, stance, swing time, cadence
and intermuscular beta coherence (15-35Hz) for the synergistic thigh (biceps femoris (BF)-semitendinosus, rectus
femoris (RF)-vastus lateralis (VL)) and ankle (Gastrocnemius lateralis (GL)-soleus (SL), Tibialis anterior (TA)-
peroneus longus (PL)) and the antagonistic (RF-BF and TA-GL) muscle pairs in swing and stance phases.
RESULTS: Comparing fast vs. fixed speed, participants walked with longer strides (21%), faster cadence (12%),
and greater coefficient of variation (CV) of stride length (14%), narrower stride width (-20%), and shorter stance (-
5%) and swing times (-14%) and with stronger TA-GL beta coherence in early stance (69%, all p<0.01). Older vs.
younger individuals walked with slower fast gait speed (~9%), higher CV of stride length (21%), weaker GL-SL (-
47%) and TA-PL (-60%) beta coherences during the late swing and early stance phase, respectively (all p<0.01).
No Group*Condition interactions occurred.
CONCLUSION: While old age seems to affect synergistic ankle but not thigh muscle beta coherence, based on a
lack of speed effect on coherence and a lack of association between spatiotemporal gait variables and ankle
muscle beta coherence, variables other than intermuscular beta coherence most likely underlie age-differences
in the neural control of walking speed.
KEYWORDS: Aging | Gait | EMG | Oscillatory coupling | Neuromuscular control
INTRODUCTION
Healthy human aging modifies the biomechanics of walking
1
, resulting in 16% per
decade decreases in self-selected walking speed after age 60
1–4
. A disproportionately
slower walking speed for a given age and sex at mid-life predicts multiple health conditions
BJMB! ! ! ! ! ! ! !
Brazilian(Journal(of(Motor(Behavior(
(
Santos et al.
2022
VOL.16
N.5
373 of 384
Special issue:
Effects of aging on locomotor patterns
later in life, including cognitive decline and mental health, falls and risk of falls, fractures,
adverse clinical events, hospitalization, mortality, and survival
4–7
. Although age-effects on
walking speed have been thoroughly examined
8–11
, the mechanisms underlying age-typical
reductions in the capacity to modulate neuromuscular control with changes in walking speed
have not yet been fully elucidated. Such information could be insightful because age can
affect locomotor muscle activation in the amplitude but also in the frequency domain
12–14
.
For instance, while aging was related to an increase in muscle coactivation during walking
(accompanied by ~20% higher energy cost of transport in older individuals)
15
, the rate of
change in coactivation due to walking speed was substantially lower in older vs. younger
individuals
12
. Specifically, coactivation between knee flexors and knee extensors increased
by ~16% as walking speed increased to 1.8 from 1.2 m·s
-1
in younger but not in older (~3.5%)
12
. Together, a lower capacity to modulate coactivation may indicate older individuals’
impaired neural control of walking when speed is changing.
One way to determine if age affects the neural control of walking with respect to
speed is to measure intermuscular coherence among the active muscles. Intermuscular
coherence is a measure of the strength of synaptic inputs to the muscles activated while
walking
16–19
. The frequency at which the intermuscular coherence arises may be related to
the origin where the coherence is generated (spinal, cortical, subcortical)
20
. Based on lesion
data in neurological patients, intermuscular beta coherence at 15–35 Hz emanates from
cortical structures
19
, which are known to contribute to gait control
16–19
. Intra/intermuscular
beta coherence of synergistic lower extremity muscles measured during walking seems to
decrease with age
13,14,18
. However, it is unknown if age and speed each differently affect
beta coherence during walking. Therefore, the purpose of the present study was to
determine the effects of age and walking speed on intermuscular beta coherence of lower
extremity muscles and the association between stride characteristics and intermuscular beta
coherence of these muscles. We hypothesized an interaction between age and walking
speed in beta coherence so that older vs. younger individuals would have lower coherence
without modulating it with walking speed. We based this expectation on a lack of age-related
modulation in beta coherence during walking after muscle fatigue
14
and split-belt
perturbations
13
. Our walking speed data would extend these previous data suggesting
potentially a general age-related impairment in the neural control of walking.
METHODS
Participants
We recruited healthy younger (n=14, age = 23 [range 20–26] years, 7 Females) and
older volunteers (n=12, age = 71 [range 66–77] years, 5 Females) to participate in the study.
Inclusion criteria were: age 18 to 29 and > 65 years and either sex. Exclusion criteria were:
inability to walk unassisted on a treadmill; musculoskeletal injury in the lower limb or surgery
that could affect walking ability; self-reported pain in the lower extremities; and neurological
or cardiac diseases. The procedures of this study were conducted following the Declaration
of Helsinki
21
, approved by the Ethical Committee of the Department of Human Movement
Sciences, University Medical Center Groningen (#ECB2017.06.12_1), and participants
consented to participate by signing the informed consent document before testing.
BJMB! ! ! ! ! ! ! !
Brazilian(Journal(of(Motor(Behavior(
(
Santos et al.
2022
VOL.16
N.5
374 of 384
Special issue:
Effects of aging on locomotor patterns
Experimental Procedures and data acquisition
Mobility was assessed by Short Physical Performance Battery
22
. Participants then
walked on a treadmill to determine their ‘comfortable’ speed by progressively increasing belt
speed by 0.1 m/s until the participant signaled ‘comfortable’. This was considered the
habitual walking speed. Subjects walked at a fixed (1.2 m/s) and at a “fast speed” (20-30%
above comfortable walking speed) for 3 minutes, separated by 1-2 min of rest. We included
a fixed speed at 1.2 to have a controlled speed walking condition (eliminating speed as a
potential confounder), and we selected the range of 20-30% above comfortable walking
speed to induce the individuals to walk fast but not run. During treadmill walking, volunteers
wore a harness and were asked not to touch or hold onto the handrail and look at the wall
in front of them.
The treadmill had two embedded force plates, and we used the vertical ground
reaction forces as an event to align the data for analysis (data recorded at a sample
frequency of 1 kHz, M-gait, Motekforce, Amsterdam, NL). During each treadmill walking
condition, we recorded EMG activity using 8 wireless sensors (dimensions: 37×26×15 mm,
electrode material: silver; Trigno Wireless System - Delsys, Natick, MA, USA). According to
SENIAN (Surface Electromyography for the Non-Invasive Assessment of Muscles)
23
, the
electrodes were placed unilaterally on the dominant limb (determined by asking the
participants and confirmed by asking the participants to kick a ball) on the following muscles:
soleus (SL), gastrocnemius lateralis (GL), tibialis anterior (TA), peroneus longus (PL), vastus
lateralis (VL), rectus femoris (RF), biceps femoris (BF), and semitendinosus (ST). EMG
signals were sampled at 2 kHz. The areas where the sensors were placed, body hair was
removed, and the skin was cleaned with alcohol. Treadmill and EMG data acquisition were
electronically synchronized with a custom-built timer and event generator.
Data analysis
Ground reaction and moment of forces, acquired from force plates, were 15 Hz low-
pass second order zero-phase Butterworth filtered. Based on the minimum ground reaction
forces, heel contact and toe-off were determined using a threshold of 50 N. Based on the
ground reaction and moment of forces, we computed the center of pressure (CoP)
24
and
detected heel strikes and toe-offs of 100 strides. From these strides, the mean and
coefficient of variation (CV) of the stride length, step width, stance, swing time, and cadence
were calculated (see for details
25
).
First, we visually inspected EMG data to minimize noise and artifacts. Then, the
data were high-pass filtered (5 Hz) using a second-order Butterworth filter and full-wave
rectified using the Hilbert transform. EMG data were then downsampled to 1 kHz to the same
sample frequency as data from the force plate. We calculated, for each heel strike in both
speed conditions, the coherence (frequency-domain coupling between two EMG signals) for
the late swing phase: −400 to −50 ms before the heel strike, and early stance phase: 50 to
400 ms after the heel strike
14
. For each phase, intermuscular coherence was computed for
six muscle pairs (four synergistic [BF-ST, RF-VL, GL-SL, TA-PL] and two antagonistic [RF-
BF and TA-GL] muscle pairs).
To compute coherence, the auto-spectra (ƒ
xx
and ƒ
yy
)) of each muscle and cross-
spectrum of the muscle pairs (ƒ
xy
) via Welch’s periodogram method was calculated
14,26
. For
each of the 100 swing and stance phases, estimates were obtained using a 350 ms window
with nonoverlapping data segments, resulting in a resolution frequency of 2.86 Hz. Spectral
estimates of individual strides were then averaged across the 100 strides. Intermuscular
BJMB! ! ! ! ! ! ! !
Brazilian(Journal(of(Motor(Behavior(
(
Santos et al.
2022
VOL.16
N.5
375 of 384
Special issue:
Effects of aging on locomotor patterns
coherence was calculated by the squared modulus of the cross-spectrum divided by the
product of the two auto-spectrum for each frequency (λ)
27
:
Coherence is usually reported in a frequency range of 0–55 Hz (but for the study
proposal, we focused only on the range of 15-35 Hz – beta-band frequency), with values
ranging from 0 (absent) to 1 (correlated). Significant coherence was considered if the value
exceeded the confidence limit (at α = 0.05) for the number of segments (L) used to estimate
the spectrum
27
.
where α = 0.05 and L is the number of strides (100) used in the analysis. During the
coherence calculation, for each subject and muscle pair, we verified the cumulant density
plots and ensured that high coherences were accompanied by near zero-lag synchronization
suggesting that cross-talk did not affect coherences
19,28
. To compare groups, intermuscular
coherences of individual subjects were combined into pooled estimates for age groups for
each walking phase for the fixed and fast walking conditions. Coherence estimates were
Fisher transformed before pooling to stabilize variance
28,29
. Since our interest is in the beta-
band, we computed the cumulative sum (area), in the range of 15-35 Hz, for each group
(older and younger), phase (swing and stance), and walking speed condition (fixed and fast).
For statistical analysis, using SPSS for Windows (Version 25, IBM, Armonk, NY,
USA), we, firstly, tested the data normality by Shapiro–Wilk. When data were non-normal
distributed, data were log-transformed for further comparisons using T-tests or ANOVAs. T-
tests were used to compare the effects of age on groups’ characteristics (age, height, body
mass, and SPPB) and fast walking speed. We compared stride outcomes (length, speed,
velocity, swing and stance time) and intermuscular beta coherence by ANOVA with as
between factor Group (younger vs. older) and within factor Condition (fixed vs. fast). ANOVA
effect size was estimated using partial eta squared (η²
p
)
30
. In case of significant interactions,
adjusted Bonferroni (post hoc) corrections for each factor were made. For post hoc level of
comparisons, Cohen’s d was calculated and considered as 0.21–0.50, 0.51 to 0.79, and >
0.79 to indicate small, medium, and large effect sizes (d), respectively
30
. To identify if speed
was correlated with coherence, we computed a Spearman’s correlation between change in
walking speed (∆ = Fast – Fixed) of coherence with ∆ walking speed as well ∆ stride
outcomes when Condition main effect indicates differences for such outcomes. For
interpreting the strength of the association, we assumed that r-values ranging from 0–0.19,
0.2–0.39, 0.40–0.59, 0.6–0.79, and 0.8–1 as a very weak, weak, moderate, strong, and very
strong correlation, respectively
31
.
RESULTS
Participants
Younger and older groups were similar in height (1.75 ± 0.11 and 1.72 ± 0.77 m, p
𝐶
(
𝜆
)
% = %
|𝑓
𝑥𝑦
(
𝜆
)
|
𝑓
𝑥𝑥
(
𝜆
)
. 𝑓
𝑦𝑦
(
𝜆
)
1 −
(
𝛼
)
1
𝐿−1
BJMB! ! ! ! ! ! ! !
Brazilian(Journal(of(Motor(Behavior(
(
Santos et al.
2022
VOL.16
N.5
376 of 384
Special issue:
Effects of aging on locomotor patterns
= 0.61) and body mass (70.14 ± 13.51 and 73.9 ± 10.60 kg, p = 0.44). Each individual in
both age groups reached the maximum score (12) on the Short Physical Performance
Battery.
Stride outcomes
Table 1 shows T-tests and ANOVAs outcomes. A significant group difference in fast
walking speed was observed, indicating that fast walking speed was ~9% slower in older
(1.61 ± 0.18) vs. younger individuals (1.76 ± 0.06, d = 1.11). Regarding stride outcomes, a
main effect of Group was found for CV of stride length and a main effect of Condition for
stride length, width, stance, and swing time and CV of stride length. No Group by Condition
interaction was observed (p<0.05). For Group, post hoc indicated that older vs. younger had
a 21% higher CV of stride length (d = 0.90). For Condition, post hoc indicated for fast vs.
fixed speed a significant increase in the mean of stride length (21%, d = 1.9), cadence (12%,
d = 1.7), and CV of stride length (14%, d = 1.7) and a decrease in mean step width (20%, d
= 0.51), swing (5%, d = 0.91), and stance time (14%, d = 1.94, Table 2).
Table 1. Statistical descriptors for the significant outcomes for Group and Condition comparisons.
Main Effect / Interaction
Outcome
T
24
/ F
1,24
p-value
(η²
p
)
Group
Fast walking speed
2.93
0.007
1.11*
CV Length
8.94
0.006
0.27
GL-SL coherence – Swing
9.20
0.006
0.28
TA-PL coherence – Stance
8.26
0.008
0.26
Condition
Length
78.01
< 0.001
0.76
Width
53.92
< 0.001
0.69
Swing
24.66
< 0.001
0.51
Stance
67.20
< 0.001
0.74
Cadence
62.86
< 0.001
0.72
CV Length
8.08
0.009
0.25
TA-GL coherence – Stance
8.61
0.007
0.26
*Cohen d effects size for T-test
Coherence
ANOVA revealed a main effect of Group for GL-SL beta coherence during late swing
and for TA-PL beta coherence during early stance, and of Condition for TA-GL beta
coherence during early stance (Figure 1). ANOVA did not indicate Group by Condition
interaction (p<0.05). For Group, post hoc indicated that, after the adjustment for walking
speed, older vs. younger individuals had weaker beta coherences during the late swing and
early stance phases for GL-SL (47%, d = 0.92, Figure 1a) and TA-PL (60%, d = 1.25, all
p<0.01, Figure 1b). For Condition, individuals (from both older and younger combined)
walked during fast vs. fixed walking speed with 69% stronger TA-GL beta coherence during
early stance (d = 0.59, p<0.01, Figure 1c).