J Neurophysiol 107: 2560 –2569, 2012.
First published February 1, 2012; doi:10.1152/jn.00950.2011.
Stability of gait and interlimb coordination in older adults
T. Krasovsky,1,2 M. C. Baniña,1,2 R. Hacmon,1,2 A. G. Feldman,2,3,4 A. Lamontagne,1,2
and M. F. Levin1,2
1
School of Physical and Occupational Therapy, McGill University, Montreal; 2Feil & Oberfeld Research Centre, Jewish
Rehabilitation Hospital (research site of the Center for Interdisciplinary Research in Rehabilitation of greater Montreal),
Laval, Quebec; 3Department of Physiology, University of Montreal, Montreal; and 4Institut de réadaptation Gingras-Lindsay
de Montreal, Montreal, Canada
Submitted 24 October 2011; accepted in final form 30 January 2012
Krasovsky T, Baniña MC, Hacmon R, Feldman AG, Lamon-
tagne A, Levin MF. Stability of gait and interlimb coordination in
older adults. J Neurophysiol 107: 2560 –2569, 2012. First published
February 1, 2012; doi:10.1152/jn.00950.2011.—Most falls in older
adults occur when walking, specifically following a trip. This study
investigated the short- and longer term responses of young (n ⫽ 24,
27.6 ⫾ 4.5 yr) and older adults (n ⫽ 18, 69.1 ⫾ 4.2 yr) to a trip during
gait at comfortable speed and the role of interlimb coordination in
recovery from tripping. Subjects walked on a self-paced treadmill
when forward movement of their dominant leg was unexpectedly
arrested for 250 ms. Recovery of center of mass (COM) movements
and of double-support duration following perturbation was deter-
mined. In addition, the disruption and recovery of interlimb coordi-
nation of the arms and legs was evaluated. Although young and older
subjects used similar lower limb strategies in response to the trip,
older adults had less stable COM movement patterns before pertur-
bation, had longer transient destabilization (⬎25%) after perturbation,
required more gait cycles to recover double-support duration (older,
3.48 ⫾ 0.7 cycles; young, 2.88 ⫾ 0.4 cycles), and had larger phase
shifts that persisted after perturbation (older, ⫺83° to ⫺90°; young,
⫺39° to ⫺42°). Older adults also had larger disruptions to interlimb
coordination of the arms and legs. The timing of the initial disruption
in coordination was correlated with the disturbance in gait stability
only in young adults. In older adults, greater initial COM instability
was related to greater longer term arm incoordination. These results
suggest a relationship between interlimb coordination and gait stabil-
ity, which may be associated with fall risk in older adults. Reduced
coordination and gait stability suggest a need for stability-related
functional training even in high-functioning older adults.
perturbation; aging; falls; arms; balance
FALL-RELATED INJURY in older adults places a heavy burden on
the health care system with over 19 billion dollars spent in
direct medical costs in the U.S. in 2000 alone (Stevens et al.
2006). Identification of mechanisms leading to falls or to
recovery from falls in older adults can assist in fall prevention
because it can lead to more effective interventions (Chang et al.
2004).
Most falls in older adults occur when walking (Berg et al.
1997; Nevitt et al. 1991) due to tripping (34%) or slipping
(25%). The fact that young adults rarely fall during walking
suggests that gait stability is affected by aging. Gait stability
can be defined as the ability to maintain functional locomotion
despite perturbations (England and Granata 2007). As a min-
Address for reprint requests and other correspondence: T. Krasovsky,
School of Physical and Occupational Therapy, 3654 Promenade Sir William
Osler, Montreal, Quebec, Canada H3G 1Y5 (e-mail: tal.krasovsky@mail.
mcgill.ca).
2560 0022-3077/12 Copyright © 2012 the American Physiological Society
Downloaded from journals.physiology.org/journal/jn (045.230.084.251) on September 29, 2022.
imum, this ability is provided by position- and velocity-depen-
dent resistance to perturbation-elicited deviations from the
ongoing locomotion. Gait stability depends on the interplay
between central and peripheral factors, which include the
central pattern generator, muscle-reflex properties, and the
mechanical interaction of the body with the environment (Taga
et al. 1991). Although the above definition of gait stability is
generally accepted, quantification of stability is controversial,
especially in older adults. Gait stability has been quantified
using responses of the motor system to time-dependent
changes in the environment and neuromuscular properties (for
a recent review, see Hamacher et al. 2011). Other measures
quantify the ability to maintain functional locomotion despite
larger perturbations, such as path obstacles or trunk displace-
ments. Transient and long-lasting responses to perturbations
have been quantified in terms of limb strategies used (Eng et al.
1994), parameters quantifying immediate changes in and grad-
ual return to a stable gait pattern (Bruijn et al. 2010), and by the
phase resetting of gait reflecting the ability to restore body
equilibrium in the environment (Feldman et al. 2011). Gait
stability is a global property of the motor system, implying that
all body segments are involved in the short- and longer term
responses to perturbation (Marigold and Misiaszek 2009).
In young healthy adults, three typical strategies have been
identified in response to a trip perturbation (Eng et al. 1994).
The first is a lowering strategy, in which the perturbed leg is
lowered to the ground with reduced step length and step time.
The second is an elevating strategy, which consists of pro-
longed maintenance of the perturbed leg in the air, and then the
leg arriving at the same step length or longer with a longer
swing duration. The third is a delayed lowering (or combined)
strategy, when an elevation strategy is attempted but eventually
the leg is lowered at a shorter step length. Lowering and
combined strategies were further shown to require more energy
and more recovery steps (de Boer et al. 2010; Forner-Cordero
et al. 2005). In older adults, some evidence suggests that
lowering responses may be more prevalent than in younger
subjects (Pijnappels et al. 2005; Roos et al. 2008). This may
suggest that older adults use less efficient responses to pertur-
bation.
Decreased gait stability in older adults may result from
deterioration in musculoskeletal properties, reflex reactions,
and central neural control. With aging, movement becomes less
automatic and more cognitively challenging (Heuninckx et al.
2005). Deficits in motor performance are usually not limited to
walking but also occur in other tasks, such as maintaining
interlimb coordination when swinging limbs or cycling. In
www.jn.org
 STABILITY AND COORDINATION IN OLDER ADULTS
addition, older adults have lower temporal coupling between
arms (Greene and Williams 1996) and between arms and legs
(Serrien et al. 2000) that further decreases with increasing
movement speed. The timing of gait events in the lower limbs
(heel strikes) is less symmetrical (decreased temporal cou-
pling) and less stable in older adults (Plotnik et al. 2007).
Response of the body to perturbation involves several fac-
tors, including context-dependent coordinated responses of the
legs (Eng et al. 1994; Schillings et al. 1996) and reactions of
the arms (Pijnappels et al. 2010; Bruijn et al. 2010), in the
short-term (1–2 steps after perturbation). Coordinated initial
arm and leg responses may deteriorate in older adults, which
may contribute to a decrease in gait stability in this population.
Roos et al. (2008) identified differences in initial arm move-
ments following a trip between young and older adults. They
suggested that arm movement served to elevate the center of
mass (COM), thus assisting recovery in young but not older
adults. Pijnappels et al. (2005) showed that improper place-
ment of the leg during recovery was related to falling in older
adults. These studies only examined immediate and short-term
responses to a trip perturbation. There may also be longer term
changes in gait stability after a perturbation (Forner-Cordero et
al. 2003), but the role of interlimb coordination of arms and
legs in such recovery remains unclear. We evaluated both
immediate and longer term effects of perturbation on gait
stability and interlimb coordination in young and older adults.
We hypothesized that older adults would be less stable and
would have less coordinated whole body and limb responses to
a perturbation during gait compared with younger adults. We
further hypothesized that interlimb coordination in the arms
and legs may play an important role in the recovery of stable
walking following perturbation and that this relationship de-
creases with age.
METHODS
Participants
Twenty-four young adults (12 male; mean age 27.6 ⫾ 4 yr) and 18
older adults (9 male; mean age 69.1 ⫾ 4 yr) volunteered for the study.
All subjects signed informed consent forms approved by the institu-
tional ethics committee. All subjects except two young and three older
adults were right-handed (Edinburgh Handedness Scale ⬎25) (Old-
field 1971). Leg dominance was evaluated by asking the subject to
identify the leg preferred for kicking a ball and was identical to hand
dominance in all but one subject. Subjects were excluded if they had
any orthopedic, vestibular, or neurological disorders affecting gait.
Experimental Procedure
Testing protocol. Subjects walked on a self-paced treadmill (Ste-
phenson et al. 2009) and wore a safety harness attached to the ceiling
to prevent falling (Fig. 1A). For safety purposes, two switches could
be pressed to stop the treadmill movement in case of an emergency.
Subjects could rest for up to 2 min between trials and wore a heart rate
monitor to ensure that heart rate did not exceed 70% of maximum
(220 beats/min minus subject age).
Subjects were allowed to habituate to treadmill walking at their
comfortable pace. Once comfortable gait speed stabilized, i.e., did not
vary by more than 10%, the pace was set for each subject with a
metronome. Subsequent data collection consisted of 2 blocks of six
40-s trials (n ⫽ 12 trials total). Subjects were not required to follow
J Neurophysiol • doi:10.1152/jn.00950.2011 • www.jn.org
Downloaded from journals.physiology.org/journal/jn (045.230.084.251) on September 29, 2022.
2561
the metronome beat following perturbation. Trip perturbations were
introduced using a custom-built device arresting forward movement
of the dominant leg during swing. Two 110-cm light wooden rods
were attached to ankle cuffs (Fig. 1A) and inserted into cylindrical
electromagnets fastened to posts 50 cm behind the subject. Rod
movement occurred with minimal friction. The device had a potenti-
ometer that monitored forward and backward movement of the
wooden rods, and baseline range of movement (swing length) was
calculated. In pseudorandomly selected steps (1–2 perturbations per
40-s trial), the device transiently blocked forward leg motion of the
dominant leg for 250 ms at 20 –30% swing length. Early swing
perturbations were chosen because they allow for a wider range of
possible responses (Schillings et al. 1996) compared with perturba-
tions in late swing, when only a leg lowering strategy is typically
used. A force transducer embedded in the rod measured perturbation
force. Perturbation duration was determined in a pilot study (n ⫽ 3)
where a 250-ms arrest generated a consistent perturbation with no
forces in the opposite direction (“backlash”). One to two trials in
which no perturbation occurred were randomly included in each block
to prevent anticipation. Because the response to the first perturbation
may differ from those to subsequent perturbations (c.f. Marigold and
Patla 2002), the first three perturbations of each subject were omitted
from the analysis. Reflective markers (36 in total) were placed on
body landmarks according to the Vicon Plug-in-Gait model (Vicon,
Oxford, UK). Specifically, heel markers were placed on the calcanei,
and finger markers were placed on the dorsums of the second meta-
carpals. Movement kinematics were recorded at 120 samples/s using
a 12-camera Vicon motion analysis system and filtered using a
sixth-order low-pass Butterworth filter (dual-pass, cutoff 10 Hz). A
sixth-order Butterworth filter was chosen because, compared with the
more often used fourth-order filter, it better suppressed noise above
the cutoff frequency.
Clinical tests. Performance of gait-related activities was assessed in
older adults by a physical therapist using the valid and reliable
Functional Gait Assessment (FGA) (Walker et al. 2007; Wrisley et al.
2004; Wrisley and Kumar 2010). The FGA evaluates functional
ambulation in community-dwelling older adults on a 30-point scale in
which ⱕ22 indicates fall risk (Wrisley and Kumar 2010). The sub-
ject’s self-confidence (self-efficacy) when performing daily activities
including walking was assessed with the valid and reliable Activities-
Specific Balance Confidence questionnaire (ABC) (Myers et al. 1998;
Powell and Myers 1995), a self-report questionnaire measured on a
100-point scale.
Data Analysis
Immediate and longer term responses: stability. Following the
definition above (Introduction), we quantified gait stability by param-
eters characterizing immediate and longer term effects on the steady-
state gait pattern. Immediate effects of the leg perturbation were
classified into three typical response strategies, as described in the
Introduction (Fig. 1B). Characteristics of the immediate and longer
term effects of perturbation were analyzed using the recovery of COM
movements and the double-support times and long-lasting phase shifts
in the gait pattern. COM position was calculated using Vicon’s
Plug-in-Gait model. For each trial, each step was time-normalized to
100 points, and during a preperturbation period of 8 steps, the position
and velocity of the COM were calculated in 3 axes (mediolateral,
anteroposterior, and vertical). COM velocity was calculated using a
2-point central difference algorithm filtered with a 25-Hz cutoff
low-pass second-order Butterworth filter (Post et al. 2000). For each
axis, COM trajectory during unperturbed walking was described as a
stable-state attractor by mean (⫹SD) velocity vs. position phase plots.
The Euclidean distance of COM from the steady state after a pertur-
bation was calculated following the method of Bruijn et al. (2010) as
2562 STABILITY AND COORDINATION IN OLDER ADULTS

Fig. 1. Experimental setup: 3 strategies of

responses to perturbation (A) and basic de-
scriptors of gait stability (B and C). A: exper-
imental setup. Subjects walked on a self-
paced treadmill where treadmill speed was
determined via a custom-built microcontroller
and changed according to the anteroposterior
position of a string attached to the subject’s
back. Subjects wore a harness connected to a
metal bar attached to the ceiling, with no
constraint to motion on the treadmill. Velcro
leg cuffs were attached to both legs to equate
sensation, and an electromagnet automatically
clamped the rod attached to the cuff on the
dominant leg for a period of 250 ms at ⬃20%
of swing length. B: phase plots of 3 types of
postperturbation leg strategies (lowering,
combined, elevation). Plots describe the an-
teroposterior position (abscissa) and velocity
(ordinate) of the heel marker on the perturbed
leg. Thin solid black lines indicate 3 preper-
turbed cycles; thick gray line indicates per-
turbed cycle, and dashed lines indicate 2 post-
perturbation cycles. C: an example of descrip-
tors of stability (Bruijn et al. 2010) following
a perturbation (vertical dotted line). The x-
axis is normalized time, where 100% is the
time from heel strike to the consecutive heel
strike of the contralateral leg. The y-axis is the
Euclidean distance from a mean phase plot
(attractor), normalized by standard deviations
of the mean and summed over 3 movement
axes. Gray line represents the fitted exponen-
tial function (see text). Immediate measures
of stability include time to peak distance (␶)
and value of peak distance (B), whereas lon-
ger term recovery is described by the slope of
the recovery function (␤) and mean distance
from the attractor at the 5th step after pertur-
bation (A).

DIST100⫻s⫹i ⫽ 冑兺 共兵关
3

j⫽1
N(j,i) ⫺ P(j,100⫻s⫹i)兴 ⁄ S(j,i)其2 ⫹ 兵关DN(j,i) ⫺ DP(j,100⫻s⫹i)兴 ⁄ DS(j,i)其2兲

where j is the movement axis, N and DN are the mean nonperturbed
position and velocity respectively, P and DP are the instantaneous
position and velocity after perturbation, S and DS are the instanta-
neous SDs of the mean position and velocity, respectively, s is the step
index following the perturbation, and i is the index within the step
[1–100] (Fig. 1C).
Because the distance at each time point depends on the SD of the
stable state, and to verify that differences did not depend on baseline
variability, the mean SD of the stable state was defined as
Fi ⫽ ACOM ⫹ (BCOM ⫺ ACOM) ⫻ e⫺␤COM(i⫺␶COM)
This analysis yielded variables that defined the immediate (BCOM,
␶COM) and longer term (␤COM, ACOM) responses of the COM to
perturbation (Fig. 1C). BCOM is the peak of the distance curve, which
indicates the degree of COM disruption, and ␶COM is the time to
BCOM, which indicates the duration of the initial destabilizing re-
sponse. ␤COM, the slope of the exponential function, describes the rate
of return of the function to baseline following the initial response, and

冑
100 3
ACOM is the mean value of the distance of the function from the
1
SD ⫽ 兺 兺
100 i⫽1 j⫽1
关S(j,i)
2
⫹ DS(j,i)
2
兴 attractor at the fifth postperturbed step, indicating the recovery of a
stable gait pattern at a time when full recovery was expected. Trials in
which ACOM was higher than twice the baseline SD were discarded to
To identify variables related to postperturbation recovery of stable- ensure reliable decay of the distance function following perturbation
state walking, we fit an exponential function to the distance curve (Post et al. 2000). On the basis of this criterion, 6% of trials were
from the peak onwards (Post et al. 2000; Bruijn et al. 2010): removed in young adults and 9% of trials in older adults.

J Neurophysiol • doi:10.1152/jn.00950.2011 • www.jn.org

Downloaded from journals.physiology.org/journal/jn (045.230.084.251) on September 29, 2022.
 STABILITY AND COORDINATION IN OLDER ADULTS
Given that in unperturbed walking, an increase in the duration of
double leg support correlates with impaired balance (Winter et al.
1990), the time required to recover steady-state walking was also
described by the number of gait cycles required for the return of the
double-support duration to preperturbation values (double-support
recovery index). Because the SD of double-support time in nonper-
turbed walking did not exceed 10% of the mean, we defined recovery
of double-support duration as the first gait cycle when double-support
duration was within 10% of baseline values.
The longer term (residual) effect of the perturbation on whole body
stability was quantified by computing long-lasting phase shifts of all
four limbs (Feldman et al. 2011; Yamasaki et al. 2003). This measure
is complementary to the COM analysis, because the COM stability
analysis quantifies the return of the COM to the preperturbed stable
state, whereas phase resetting describes the residual change in the
movement pattern, i.e., changes that persist after the perturbation. A
long-lasting phase shift was defined as a shift in the locomotor rhythm
that persisted for at least three cycles after perturbation. Phase shifts
were computed by comparing the timing of gait events: the foot
contact for the legs and the front-to-back movement direction reversal
for the arms. These events were identified on the basis of movements
of heel and finger markers, respectively, in the sagittal plane before
and after perturbation. The anteroposterior position of each limb in the
three cycles before perturbation was projected forward, and the
minimal time between the actual and projected foot contact/direction
reversal was divided by the preperturbed cycle period. In the case of
more than one event occurring within the duration of one preperturbed
cycle, the first event was used (Ford et al. 2007). The obtained value
was multiplied by 360°, resulting in a range of possible phase shifts
from ⫺180° to ⫹180°. A negative value denoted a phase advance, and
a positive value denoted a phase delay.
Immediate and longer term responses: interlimb coordination. The
immediate effect of the perturbation on interlimb coordination of arms
and legs was described using the same approach outlined above for
the COM. The difference in the angular position of limb pairs, arms
(shoulder flexion angle) or legs (hip flexion angle), was used. The
angular difference rather than the continuous relative phase (c.f.
Scholz et al. 1987) was used, because the latter is indicated only for
sinusoidal signals (Peters 2003). Shoulder and hip angles were com-
puted using the Plug-in-Gait model. Shoulder angles in three move-
ment planes (flexion/extension, abduction/adduction, and rotation)
were calculated relative to thorax movement, and hip angles (flexion/
extension, abduction/adduction, and rotation) were calculated relative
to pelvic movement. The baseline interlimb coordination was quan-
tified by calculating the difference between the angular position
signals of the two limbs in each plane for each time point. The angular
position of the limb on the perturbed side was subtracted from that of
the nonperturbed side, and the phase plot was produced as above. The
parameters of initial and longer term recovery derived were ␶, B, and
A for arms (␶arms, Barms, Aarms) and legs (␶legs, Blegs, Alegs). These
parameters were used as explanatory variables for COM stability. In
addition, the baseline SD of the difference signal was calculated for
arm and leg trajectories to verify that baseline conditions were similar
between groups.
Statistical Analyses
A one-way ANOVA was used to test the effect of group on
response strategy. Because response strategy selection was assumed to
affect recovery (de Boer et al. 2010), it was entered as an independent
factor in further analyses. Generalized estimation equations (GEE;
Wald ␹2 statistic) (Zeger and Liang 1986, 1988) were used to test the
effect of group and response strategy on gait stability and interlimb
coordination. GEE was selected because of its ability to accommodate
correlated and missing data. To prevent variations due to different
walking speed in individual trials, walking speed was set as a
covariate in the model. For phase shifts of all four limbs (circular
J Neurophysiol • doi:10.1152/jn.00950.2011 • www.jn.org
Downloaded from journals.physiology.org/journal/jn (045.230.084.251) on September 29, 2022.
2563
variables), Harrison-Kanji tests were performed (circular 2-way
ANOVAs with response pattern and group as independent variables;
Harrison and Kanji 1988). A repeated-measures ANOVA was used to
test for group differences in the duration of coordination disruption for
arm, leg, and COM movements. Spearman correlations were used to
correlate gait stability with interlimb coordination and the clinical
scores (for older adults). All analyses were carried out in SPSS
(version 17.0) except for circular data analyses, which were computed
in Matlab 6.5 (The MathWorks, Natick, MA) using the circular
statistics toolbox (Berens 2009). Significance was set at P ⬍ 0.05 for
all tests. For the correlations, Bonferroni corrections were used to
adjust for the number of planned comparisons (P ⬍ 0.006).
RESULTS
Anthropometric data for all subjects are detailed in Table 1.
Walking speed was similar for both groups. The perturbation
generated a consistent mean force of 38.1 ⫾ 8.3 N that did not
vary by response strategy or group. All subjects easily adapted
to treadmill walking and produced rhythmical locomotor pat-
terns in the arms and legs (Fig. 2A). Perturbation disrupted the
rhythmical patterns in all four limbs. Gait speed in the first
three postperturbed steps decreased by a small but significant
percentage in both groups (by 2.6% in young subjects and by
5.1% in older adults, P ⬍ 0.05 for group). This change was not
correlated with any of the analyzed variables related to recov-
ery of steady-state walking.
Role of Response Strategy in Recovery
of Steady-State Walking
We identified three strategies in response to perturbation. In
48% of cases in young adults (percentage of cases was deter-
mined for each subject and averaged across group) and 64% of
cases in older adults, a lowering strategy was used. A com-
bined strategy was used in 44% and 31% of cases in young and
older subjects, respectively, and an elevation strategy was used
in 8% and 5% of cases, respectively. There were no group
differences in strategy selection (Fig. 3A1).
Response strategy was related both to the immediate and
longer term characteristics of COM trajectories following the
perturbation. In both groups, the immediate COM responses
were related to the strategy used (Fig. 3A2, ␶COM: ␹22 ⫽ 56.69,
P ⬍ 0.001; Fig. 3A3, BCOM: ␹22 ⫽ 8.98, P ⬍ 0.01). A lowering
strategy generated the largest disruption in COM trajectory
(BCOM, P ⬍ 0.01) and the longest duration of the initial
response (␶COM, P ⬍ 0.001). In contrast, an elevation response
generated the smallest disruption to COM trajectory. In the
Table 1. Anthropometric and kinematic data for young
and older adults
Variable Young Adults Older Adults
Height, cm 172.1 ⫾ 7.7 167.5 ⫾ 10.5
Weight, kg 71.6 ⫾ 10.8 69.2 ⫾ 13.3
Age, yr 27.6 ⫾ 4.5 69.1 ⫾ 4.2
FGA 27.8 ⫾ 2.3
ABC 93.5 ⫾ 6.6
Comfortable gait speed, m/s 1.07 ⫾ 0.09 1.06 ⫾ 0.17
Swing length, mm 661.1 ⫾ 39 645.1 ⫾ 68
Swing time, s 0.44 ⫾ 0.02 0.43 ⫾ 0.03
Data are means ⫾ SD for young (n ⫽ 24) and older adults (n ⫽ 18). FGA,
Functional Gait Assessment (30-point scale); ABC, Activities-Specific Bal-
ance Confidence questionnaire (100-point scale).
2564 STABILITY AND COORDINATION IN OLDER ADULTS

Fig. 2. Typical examples of phase shifts that persisted long after perturbation (A) and characteristics of center-of-mass (COM) stability (B). A: whole body phase
shifts in a lowering response for a young (left) and older adult (right) following a trip perturbation (vertical black line). Movement cycles over a period of 4 s
before and after perturbation are shown for the heel of the nonperturbed and perturbed leg (top two panels) and the contralateral and ipsilateral arms (bottom
two panels). The preperturbed cycles of the anteroposterior (AP) position signal are projected forward (dashed gray traces), and the timing of the peaks in the
position signal are calculated in reference to the unperturbed cycle duration. B: COM stability as manifested by the distance of the COM from the mean
nonperturbed COM phase plot for a young (left) and older adult (right). For the older adult, the duration of the initial response (time to peak of the distance
function) is longer and the slope of the return to baseline is less steep. Onset of perturbation is marked by a vertical line.

longer term, response strategy was related to COM stability at
the fifth step after perturbation (Fig. 3B1, ACOM: ␹22 ⫽ 24.51,
P ⬍ 0.001) and consequently, also to the rate of return to
steady-state walking (Fig. 3B2, ␤COM: ␹22 ⫽7.01, P ⬍ 0.04). In
addition, a lowering strategy was associated with longer dou-
ble-support recovery time (Fig. 3B3, ␹22 ⫽ 38.68, P ⬍ 0.001)
and larger long-term phase shifts for all limbs (␹22 ⬎ 300.3,
P ⬍ 0.001).
Effect of Age on Recovery of Steady-State Walking
Older adults had larger baseline SDs of COM trajectory
during nonperturbed walking (␹22 ⫽ 16.80, P ⬍ 0.001). How-
ever, by the fifth step after perturbation, COM trajectories
(ACOM) returned to baseline in both groups.
Characteristics of both the short- and longer term responses
were different in older compared with young adults. In older
adults, the duration of the initial response was longer (95% CI
␶COM young: 68.8 to 78.9% of step duration, older: 80.6 to
102.8%; ␹22 ⫽ 8.23, P ⬍ 0.006, with group ⫻ strategy
interaction at ␹22 ⫽ 9.87, P ⬍ 0.008, indicating that this
difference was specific to lowering and combined strategies;
Fig. 2B). However, the degree of disruption (BCOM) was
smaller in older adults (␹22 ⫽ 8.70, P ⬍ 0.004), which may
have been due to the larger baseline SD of nonperturbed
walking.
J Neurophysiol • doi:10.1152/jn.00950.2011 • www.jn.org
Downloaded from journals.physiology.org/journal/jn (045.230.084.251) on September 29, 2022.
Older adults had lower rates of recovery from perturbation
(␤COM: ␹22 ⫽ 4.24, P ⬍ 0.04), took more steps to recover
double-support duration (␹22 ⫽ 8.68, P ⬍ 0.004), and had larger
phase shifts (⫺83° to ⫺90°) compared with young adults
(⫺39° to ⫺42°; ␹22 ⬎ 22.0, P ⬍ 0.001).
Interlimb Coordination and Recovery
of Steady-State Walking
Baseline SDs of arm and leg interlimb coordination were
similar between groups. In both groups, the initial disruption in
arm coordination (␶arms) occurred later than that of the COM
and the legs (F1,39 ⫽ 45.68, P ⬍ 0.001), whereas the durations
of the coordination disruptions in COM and legs did not differ
in either group. In both groups, response strategies were related
to the amount of disruption in arm and leg interlimb coordi-
nation at the fifth step after perturbation (Aarms: ␹22 ⫽ 28.87,
P ⬍ 0.001; Alegs: ␹22 ⫽ 12.14, P ⬍ 0.003) such that the
coordination between the two arms and the two legs was more
disrupted at the fifth step if a lowering strategy was used.
Interlimb coordination in both arms and legs at the fifth step
after perturbation (Aarms, Alegs) was more disrupted in older
compared with young adults. Although by the fifth step after
perturbation, the COM trajectory (ACOM) returned to baseline
in both groups, arm and leg coordination remained relatively
 STABILITY AND COORDINATION IN OLDER ADULTS 2565

Fig. 3. Whole body stability following perturbation: effect of response strategy and group. A: variables related to the immediate response. 1, Distribution of
response strategies in both groups; 2, time to peak distance of COM from phase plot attractor (␶COM); 3, value of peak distance of COM from phase plot attractor
(BCOM). B: variables related to the longer term response. 1, Distance from attractor at the 5th step after perturbation (ACOM); 2, slope of the recovery function
(␤COM); 3, number of gait cycles required for double-support time to return to baseline ⫾ 10%. Values are means; error bars indicate SD. *P ⬍ 0.05; **P ⬍
0.01. NS, not significant.

more disrupted in the older group (Aarms: ␹22 ⫽ 8.90, P ⬍
0.004; Alegs: ␹22 ⫽ 10.74, P ⬍ 0.002).
In both groups, the larger initial destabilization of COM
(BCOM) led to a larger disruption in arm coordination (Barms:
young: r ⫽ 0.63, P ⬍ 0.006, older: r ⫽ 0.72, P ⬍ 0.007).
However, only in young adults was the duration of the initial
disruption of COM trajectory (␶COM) related to the duration of
the initial disruption in arm and leg coordination (␶arms: r ⫽
0.56, P ⬍ 0.005; ␶legs: r ⫽ 0.65, P ⬍ 0.005). In young adults,
in addition, the duration of the initial disruption of leg coordi-
nation (␶legs) was also related to double-support recovery time
(r ⫽ 0.59, P ⬍ 0.02).
These relationships were absent in older adults. However, in
the older group, a larger initial destabilization (BCOM) affected
the disruption in arm coordination after five steps (Aarms: r ⫽
0.62, P ⬍ 0.007) such that the arms remained less coordinated
after five steps when the initial destabilization was larger.
Recovery of gait stability or interlimb coordination in the older
adults was not related to clinical tests of walking performance
(FGA) or balance self-confidence (ABC).
DISCUSSION
We investigated gait stability and interlimb coordination in
young and older adults following a perturbation to the domi-
nant leg during walking at comfortable speed. Perturbation
J Neurophysiol • doi:10.1152/jn.00950.2011 • www.jn.org
Downloaded from journals.physiology.org/journal/jn (045.230.084.251) on September 29, 2022.
resulted in a disruption to the steady-state walking pattern in all
subjects followed by a return to baseline walking pattern and a
long-lasting phase shift in all four limbs. Gait stability was
related to the pattern of leg movement used in response to the
perturbation such that when a lowering strategy was used, a
longer lasting and stronger disruption of stability occurred.
Older adults had more variable COM patterns at baseline, an
initially longer COM destabilization following perturbation,
slower COM recovery, and a larger residual phase shift in
movements of the arm and legs. In addition, arm and leg
coordination of older adults remained more disrupted for up to
five steps following perturbation. In young adults only, the
temporal evolution of arm and leg coordination was correlated
with COM stability. In older adults, however, a stronger initial
destabilization led to a larger and longer term disruption in arm
coordination.
The older participants were all able to recover from the trip
perturbation and regain their baseline walking pattern. They
used the same strategies as young adults when reacting to the
perturbation (Pavol et al. 2001). In young adults, use of a
lowering strategy is typical for mid- and late-swing perturba-
tions (Schillings et al. 1996) and is considered less energy
efficient than an elevation strategy (de Boer et al. 2010). We
showed that use of this strategy was related to a longer lasting,
more unstable postperturbation state compared with the use of
2566 STABILITY AND COORDINATION IN OLDER ADULTS
combined or elevation strategies. Previous studies have anec-
dotally suggested that strategy selection might differ between
young and older adults (Pijnappels et al. 2005; Roos et al.
2008). Although we did not find evidence to support a different
distribution of response strategies in young and older adults,
older adults differed from young adults in immediate effects
and longer term responses to perturbation. Specifically, older
adults had a 25% longer initial destabilization, took longer to
recover COM stability and double-support duration, and had
larger whole body long-lasting phase shifts. These differences
were observed in a group of high-functioning, physically active
older adults with FGA scores of 27.8 ⫾ 2.3 out of 30,
compared with a reported average of 27.1 ⫾ 2.3 for people
aged 60 – 69 yr and 24.9 ⫾ 3.9 for people aged 70 –79 yr
(Walker et al. 2007). None of the older adults in this study was
at risk of falling (FGA ⱕ 22), and yet we observed that the
period of decreased stability following perturbation was longer
in this group. Thus the clinical tests may be less sensitive to
minor changes in the ability to recover from perturbation, as
also implied by the lack of relationship between clinical mea-
sures of balance (FGA) and balance self-confidence (ABC) and
the recovery of steady-state walking. This is consistent with
preliminary results of Marone et al. (2011), who suggested that
balance self-confidence (ABC scores) had no influence on the
kinematic response to an induced trip in older adults. However,
it is possible that more differences would be observed in a
group of older adults with more functional heterogeny.
Age-related differences in gait stability such as those de-
scribed here could result from several factors, including de-
clines in muscle strength and/or central and reflex reactions. In
healthy, physically active older adults, peak isometric strength
decreases by one-third by the age of 65 yr. Although neuro-
logical factors such as reaction times also change with age
(Thelen et al. 1996), the magnitude of these differences is small
(10 –25 ms) and unlikely to explain age differences in tripping
responses (Schultz 1995; Schultz et al. 1997). In this study, age
differences in gait stability were not specific to the first step,
indicating a prolonged period of instability following a forward
loss of balance (Forner-Cordero et al. 2003). Furthermore, the
finding that the response observed in both young and older
adults involved the whole body may indicate the involvement
of a central mechanism. It is well known that in cats, the motor
cortex and specifically the corticospinal tract are involved in
obstacle avoidance during gait by modifying forelimb eleva-
tion according to obstacle characteristics (Drew 1988, 1993;
Lajoie et al. 2010). Microstimulation of pyramidal tract neu-
rons generated a phase shift in the locomotor patterns in the cat
hindlimb (Orlovsky 1972) similar to phase shifts observed in
this study. In humans, both the corticospinal tract and other
supraspinal locomotor centers (mesencephalic locomotor re-
gion, subthalamic locomotor region) have been implicated in
the control of swing length modulation during gait (Shik and
Orlovsky 1976) with possible differences between step short-
ening and lengthening reactions (Varraine et al. 2000). The
long-term whole body responses observed in this study may be
mediated by the superposition of a supraspinal signal on a
spinal locomotor command. A change in this control sequence
or a slowing of transmission in locomotor pathways in older
adults could be responsible, in part, for the difference in the
resulting locomotor patterns following perturbation.
J Neurophysiol • doi:10.1152/jn.00950.2011 • www.jn.org
Downloaded from journals.physiology.org/journal/jn (045.230.084.251) on September 29, 2022.
Results from this study also suggest that arm and leg
coordination following perturbation are linked to recovery of
COM stability. In the short term following a perturbation to the
torso, early-onset muscle activity occurred in shoulder and arm
muscles (Misiaszek 2003). Bruijn et al. (2010) compared short-
and long-term gait stability of young adults following a per-
turbation to the torso with and without arm swing. They
showed that walking without arm swinging increased the time
to peak distance from the attractor (␶) and decreased the slope
of the return to baseline (␤) for the trunk. Despite differences
in the methodology used in our study (analysis of COM instead
of a trunk rigid body), we found similar results when compar-
ing responses between different strategies at similar speeds: a
lowering response was related to longer initial destabilization
and a longer recovery phase (Fig. 2). We further found, using
the same technique for evaluating changes in interlimb coor-
dination, that timing of the disruption in arm and leg interlimb
coordination was coupled with COM stability only in young
adults. In older adults, however, the timing of the initial
interlimb coordination response was not related to COM sta-
bility. This suggests that older adults were less able to adapt the
arm and leg responses to the changes in COM stability result-
ing from perturbation. This finding is consistent with those of
Roos et al. (2008), who investigated the role of arm movements
in the sagittal plane immediately following a trip in young and
older adults. These authors found a positive correlation be-
tween vertical arm movement and COM displacement in young
but not older adults. It was suggested that older adults did not
use the arms to assist COM elevation following a trip. We
investigated movement in three dimensions and showed that
the larger disruptions in interlimb coordination persisted in
older adults in the longer term. In older but not young adults,
the disruption in the arm interlimb coordination after five steps
was still related to the perturbation, suggesting a longer term
global effect.
A mechanism by which the arms may initially assist recov-
ery in the case of a trip has been suggested by Pijnappels et al.
(2010). They showed that during the leg elevation response
elicited by a trip, the arm swing amplitude increased and was
associated with a large body rotation in the transverse plane,
supporting the increased step length required to prevent a fall.
These authors focused on elevation responses during over-
ground walking. In our study, however, trip perturbations led
to a majority of lowering or combined responses that involve
premature lowering of the perturbed leg. In this type of re-
sponse, a transition from swing to stance occurs in one leg and
from stance to swing in the other (phase shift; Feldman et al.
2011). To completely regain the preperturbed whole body
coordination pattern and regardless of the initial response to
perturbation, the arms have to switch movement direction
(from forward to backward or vice versa) such that the anti-
phase pattern between the arms and legs is restored and a
similar phase shift is observed in upper and lower limbs (Fig.
2). We have shown that in both young and older adults, the
initial response of the arms occurred after that of the legs and
COM. However, in older adults, the arms and legs remained
less coordinated for up to five steps after the perturbation,
whereas COM stability was restored in both groups.
Gait can be considered as resulting from translation of body
equilibrium in the environment (Feldman et al. 2011). To
prevent falling, subjects can transiently slow down or speed up
 STABILITY AND COORDINATION IN OLDER ADULTS
the translation in response to perturbation, which can be
detected as a long-lasting phase shift in gait. Thus this shift
may be involved in fall prevention by allowing subjects to
regain gait stability following a perturbation, as also follows
from the analysis of reactions to perturbations in a simple
biomechanical model of gait (Yamasaki et al. 2003). There-
fore, the ability to perform the coordinated phase shift in all
four limbs required to successfully recover from perturbation
may be decreased in older adults, and interlimb coordination,
especially for the arms, may be less stable.
It was previously found that even during unperturbed gait,
older adults have a less stable bilateral leg coordination (Plot-
nik et al. 2007), as manifested by a more variable relative phase
between strides. The temporal stability of interlimb coordina-
tion between arms or between arms and legs was also lower in
older adults in tasks unrelated to gait, such as bimanual cycling
(Greene and Williams 1996) or arm-leg swinging during sitting
(Fujiyama et al. 2009; Serrien et al. 2000). These deficits were
more pronounced during anti-phase movements as in walking.
Our leg perturbation caused a disruption in arm and leg
coordination in both young and older adults. However, older
adults had more difficulty recovering coordination following
the perturbation. The performance of anti-phase limb swinging
(even without walking) is considered to be an “effortful”
process, compared with in-phase movement, which is consid-
ered to be more “automatic” (Greene and Williams 1996). An
effortful process typically requires increased attention. The
reduced stability of anti-phase movements in older adults’
performance may be related to a reduction in cognitive re-
sources (Sparrow et al. 2005; Wishart et al. 2000). A pertur-
bation places additional load on cognitive resources of older
adults (for review, see Maki and McIlroy 2007) and increases
the disruption to interlimb coordination (Heuninckx et al.
2004). Although attention was not directly measured in this
study, it is possible that the perturbation placed an increased
load on the system and generated a larger disruption to inter-
limb coordination.
In our study, early-to-mid swing perturbations generated
relatively more lowering responses compared with previous
studies involving treadmill walking with perturbations (c.f.
Schillings et al. 2000). The difference may partly be explained
by the type of perturbations used in different studies. In our
study, the perturbation consisted of a posterior force applied to
the ankle (pulling from behind), whereas in previous studies,
perturbations consisted of the placement of an obstacle in the
walking path. Two studies (Forner-Cordero et al. 2003; Ko-
bayashi et al. 2000) investigated a similar type of posterior pull
perturbation in young healthy subjects. In the single-subject
study of Kobayashi et al. (2000), early-to-mid swing perturba-
tions resulted in a lowering strategy and a phase advance
similar to results reported here. In the report of Forner-Cordero
et al. (2003), different strategies were observed for early-swing
perturbations, and the authors hypothesized that strategy selec-
tion may be optimized in terms of safety or energy consump-
tion. In contrast to our study, both of these previous studies
used a motor-driven treadmill, which may have led to different
responses because subjects had to “keep up” with the treadmill
speed.
To quantify gait stability and interlimb coordination, we
used the method described by Bruijn et al. (2010), which
enables a fine-grained comparison of differences in recovery of
J Neurophysiol • doi:10.1152/jn.00950.2011 • www.jn.org
Downloaded from journals.physiology.org/journal/jn (045.230.084.251) on September 29, 2022.
2567
steady-state walking between different conditions. Bruijn et al.
(2010) suggested that the initial response to a destabilizing
perturbation (quantified by ␶ and B) is indicative of the steady-
state walking pattern, whereas ␤ and A quantify the recovery
of steady-state walking. The authors suggest that lower values
of ␶ indicate less stability in the initial response to perturbation.
Results from our study suggest an additional interpretation. We
found that ␶COM was related to response strategy. For example,
a lowering strategy, known to be less energy efficient and
require longer recovery, was associated with higher values of
␶COM.. Combined and elevation strategies, which are less
destabilizing (de Boer et al. 2010), were associated with lower
values of ␶COM. A higher value of ␶ implies that the least stable
point occurs later in the recovery process. This means that the
initial destabilization of the COM caused by the perturbation
was longer in duration. These results coincide with higher
values of A and lower values of ␤, indicating that both stages
of the recovery were less efficient when a lowering strategy
was used compared with combined or elevation strategies.
The method of Bruijn et al. (2010) relies on the similarity of
baseline standard deviations to enable comparisons. In our
case, older adults had larger baseline variability of COM
trajectories. However, by the fifth step after perturbation, COM
stability returned to similar values with respect to baseline SD
in both groups. The higher baseline SD in older adults did not
affect the temporal evolution of the response (␶COM) but may
have resulted in lower values of initial COM disruption (BCOM)
in this group. On the other hand, the higher variability in COM
movement in older adults is important given the link between
gait variability and falls (c.f. Hausdorff et al. 2001). Our data
suggest that group differences were not due to baseline differ-
ences, since other measures of gait stability, namely, phase
shifts and double-support recovery, were also disrupted in
older subjects.
This study has several limitations. Our sample of older
adults was relatively high functioning, and more group differ-
ences between young and older adults could possibly be found
in a sample of older adults with a wider range of FGA scores.
The use of a self-paced treadmill limited the ability to control
gait speed (cadence control only), but it did enable a more
functional examination of responses to perturbation than use of
a motorized treadmill. It also allowed us to produce multiple
perturbations in each subject. Finally, the use of a metronome
to pace walking may have unnaturally stabilized gait patterns
or introduced an increased cognitive load by adding a distrac-
tor. However, we did not insist that subjects follow the met-
ronome pacing after perturbations, and perturbations were
initiated only after walking patterns stabilized during each trial.
In conclusion, results from this study indicate that in older
adults, gait stability and stability of interlimb coordination
were more disrupted following perturbation in both the imme-
diate and longer term. The longer period of destabilization in
older adults following a perturbation suggests that older adults
would be more vulnerable to additional perturbations during
this period, for example, when walking on unstable terrain
(rocks, ice, etc.). This finding in a group of high-functioning
older adults with no fall risk could indicate a possible need for
functional training of stability using a perturbation approach
(Marigold and Misiaszek 2009). In young adults but not in
older adults, COM stability was related to the timing of the
disruption in arm and leg interlimb coordination, but only in
2568 STABILITY AND COORDINATION IN OLDER ADULTS
older adults did the disruption to gait affect longer term arm
interlimb coordination. This suggests a relationship between
interlimb coordination and recovery of gait stability in older
adults.
ACKNOWLEDGMENTS
Thanks are extended to Dr. J. Fung for useful discussions, to R. Guberek for
assistance with subject recruitment and evaluation, to G. Chilingaryan for
statistical advice, and to participants of the study.
GRANTS
This work was supported by Collaborative Health Research Program,
Natural Sciences and Engineering Research Council of Canada, and Fonds de
recherche du Quebec—Nature et technologies. T. Krasovsky’s doctoral studies
are supported by a Canadian Institutes of Health Research (CIHR) Locomotor
Team Grant and the Heart and Stroke Foundation of Canada (Focus on Stroke).
M. F. Levin holds a Tier 1 Canada Research Chair in Motor Recovery and
Rehabilitation. A. Lamontagne holds a New Investigator Salary Award from
CIHR.
DISCLOSURES
No conflicts of interest, financial or otherwise, are declared by the author(s).
AUTHOR CONTRIBUTIONS
Author contributions: T.K., A.G.F., A.L., and M.F.L. conception and design
of research; T.K., M.C.B., and R.H. performed experiments; T.K., M.C.B., and
R.H. analyzed data; T.K., A.G.F., A.L., and M.F.L. interpreted results of
experiments; T.K., A.G.F., and M.F.L. prepared figures; T.K. drafted manu-
script; T.K., M.C.B., R.H., A.G.F., A.L., and M.F.L. edited and revised
manuscript; T.K., M.C.B., R.H., A.G.F., A.L., and M.F.L. approved final
version of manuscript.
REFERENCES
Berens P. CircStat: A MATLAB toolbox for circular statistics. J Stat Softw 31:
1–21, 2009.
Berg WP, Alessio HM, Mills EM, Tong C. Circumstances and consequences
of falls in independent community-dwelling older adults. Age Ageing 26:
261–268, 1997.
Bruijn SM, Meijer OG, Beek PJ, van Dieën J. The effects of arm swing on
human gait stability. J Exp Biol 213: 3945–3952, 2010.
Chang JT, Morton SC, Rubenstein LZ, Mojica WA, Maglione M, Suttorp
MJ, Roth EA, Shekelle PG. Interventions for the prevention of falls in
older adults: systematic review and meta-analysis of randomised clinical
trials. Br Med J 328: 680, 2004.
de Boer T, Wisse M, van der Helm FCT. Mechanical analysis of the
preferred strategy selection in human stumble recovery. J Biomech Eng 132:
071012– 071018, 2010.
Drew T. Motor cortical cell discharge during voluntary gait modification.
Brain Res 457: 181–187, 1988.
Drew T. Motor cortical activity during voluntary gait modifications in the cat.
I. Cells related to the forelimbs. J Neurophysiol 70: 179 –199, 1993.
Eng JJ, Winter DA, Patla AE. Strategies for recovery from a trip in early and
late swing during human walking. Exp Brain Res 102: 339 –349, 1994.
England SA, Granata KP. The influence of gait speed on local dynamic
stability of walking. Gait Posture 25: 172–178, 2007.
Feldman AG, Krasovsky T, Baniña MC, Lamontagne A, Levin MF.
Changes in the referent body location and configuration may underlie human
gait, as confirmed by findings of multi-muscle activity minimizations and
phase resetting. Exp Brain Res 210: 91–115, 2011.
Ford MP, Wagenaar RC, Newell KM. The effects of auditory rhythms and
instruction on walking patterns in individuals post stroke. Gait Posture 26:
150 –155, 2007.
Forner-Cordero A, Koopman HFJM, van der Helm FCT. Multiple-step
strategies to recover from stumbling perturbations. Gait Posture 18: 47–59,
2003.
J Neurophysiol • doi:10.1152/jn.00950.2011 • www.jn.org
Downloaded from journals.physiology.org/journal/jn (045.230.084.251) on September 29, 2022.
Forner-Cordero A, Koopman HJFM, van der Helm FCT. Energy analysis
of human stumbling: the limitations of recovery. Gait Posture 21: 243–254,
2005.
Fujiyama H, Garry MI, Levin O, Swinnen SP, Summers JJ. Age-related
differences in inhibitory processes during interlimb coordination. Brain Res
1262: 38 – 47, 2009.
Greene LS, Williams HG. Aging and coordination from the dynamic pattern
perspective. In: Changes in Sensory Motor Behavior in Aging, edited by
Ferrandez A and Teasdale N. Amsterdam: Elsevier, 1996, p. 89 –131.
Hamacher D, Singh N, van Dieën J, Heller M, Taylor W. Kinematic
measures for assessing gait stability in elderly individuals: a systematic
review. J R Soc Interface 8: 1682–1698, 2011.
Harrison D, Kanji GK. The development of analysis of variance for circular
data. J Appl Stat 15: 197, 1988.
Hausdorff JM, Rios DA, Edelberg HK. Gait variability and fall risk in
community-living older adults: a 1-year prospective study. Arch Phys Med
Rehabil 82: 1050 –1056, 2001.
Heuninckx S, Debaere F, Wenderoth N, Verschueren S, Swinnen SP.
Ipsilateral coordination deficits and central processing requirements associ-
ated with coordination as a function of aging. J Gerontol B Psychol Sci Soc
Sci 59: P225–P232, 2004.
Heuninckx S, Wenderoth N, Debaere F, Peeters R, Swinnen SP. Neural
basis of aging: the penetration of cognition into action control. J Neurosci
25: 6787– 6796, 2005.
Kobayashi M, Nomura T, Sato S. Phase-dependent response during human
locomotion to impulsive perturbation and its interpretation based on a neural
mechanism. Jpn J Med Elec Biol Eng 38: 20 –32, 2000.
Lajoie K, Andujar JÉ, Pearson K, Drew T. Neurons in area 5 of the
posterior parietal cortex in the cat contribute to interlimb coordination
during visually guided locomotion: a role in working memory. J Neuro-
physiol 103: 2234 –2254, 2010.
Maki BE, McIlroy WE. Cognitive demands and cortical control of human
balance-recovery reactions. J Neural Transm 114: 1279 –1296, 2007.
Marigold DS, Misiaszek JE. Whole-body responses: neural control and
implications for rehabilitation and fall prevention. Neuroscientist 15: 36 –
46, 2009.
Marigold DS, Patla AE. Strategies for dynamic stability during locomotion
on a slippery surface: effects of prior experience and knowledge. J Neuro-
physiol 88: 339 –353, 2007.
Marone JR, Rosenblatt NJ, Troy KL, Grabiner MD. Fear of falling does
not alter the kinematics of recovery from an induced trip: a preliminary
study. Arch Phys Med Rehabil 92: 2093–2095, 2011.
Myers AM, Fletcher PC, Myers AH, Sherk W. Discriminative and evalua-
tive properties of the activities-specific balance confidence (ABC) scale. J
Gerontol A Biol Sci Med Sci 53: M287–M294, 1998.
Nevitt MC, Cummings SR, Hudes ES. Risk factors for injurious falls: a
prospective study. J Gerontol 46: M164 –M170, 1991.
Oldfield RC. The assessment and analysis of handedness: the Edinburgh
inventory. Neuropsychologia 9: 97–113, 1971.
Orlovsky GN. The effect of different descending systems on flexor and
extensor activity during locomotion. Brain Res 40: 359 –372, 1972.
Pavol MJ, Owings TM, Foley KT, Grabiner MD. Mechanisms leading to a
fall from an induced trip in healthy older adults. J Gerontol A Biol Sci Med
Sci 56: M428 –M437, 2001.
Peters BT, Haddad JM, Heiderscheit BC, Van Emmerik REA, Hamill J.
Limitations in the use and interpretation of continuous relative phase. J
Biomech 36: 271–274, 2003.
Pijnappels M, Bobbert MF, van Dieën J. How early reactions in the support
limb contribute to balance recovery after tripping. J Biomech 38: 627– 634,
2005.
Pijnappels M, Kingma I, Wezenberg D, Reurink G, van Dieën J. Armed
against falls: the contribution of arm movements to balance recovery after
tripping. Exp Brain Res 201: 689 – 699, 2010.
Plotnik M, Giladi N, Hausdorff J. A new measure for quantifying the
bilateral coordination of human gait: effects of aging and Parkinson’s
disease. Exp Brain Res 181: 561–570, 2007.
Post AA, Peper CE, Daffertshofer A, Beek PJ. Relative phase dynamics in
perturbed interlimb coordination: stability and stochasticity. Biol Cybern 83:
443– 459, 2000.
Powell LE, Myers AM. The Activities-specific Balance Confidence (ABC)
scale. J Gerontol A Biol Sci Med Sci 50A: M28 –M34, 1995.
Roos PE, McGuigan MP, Kerwin DG, Trewartha G. The role of arm
movement in early trip recovery in younger and older adults. Gait Posture
27: 352–356, 2008.
 STABILITY AND COORDINATION IN OLDER ADULTS 2569

Schillings AM, Van Wezel BMH, Duysens J. Mechanically induced stum-
bling during human treadmill walking. J Neurosci Methods 67: 11–17, 1996.
Scholz JP, Kelso JAS, Schöner G. Nonequilibrium phase transitions in
coordinated biological motion: critical slowing down and switching time.
Phys Lett A 123: 390 –394, 1987.
Schultz AB. Muscle function and mobility biomechanics in the elderly: an
overview of some recent research. J Gerontol A Biol Sci Med Sci 50: 60–63, 1995.
Schultz A, Ashton-Miller JA, Alexander NB. What leads to age and gender
differences in balance maintenance and recovery? Muscle Nerve 20: 60–64, 1997.
Serrien DJ, Swinnen SP, Stelmach GE. Age-related deterioration of coordi-
nated interlimb behavior. J Gerontol B Psychol Sci Soc Sci 55: P295–P303, 2000.
Shik ML, Orlovsky GN. Neurophysiology of locomotor automatism. Physiol
Rev 56: 465–501, 1976.
Sparrow WA, Parker S, Lay B, Wengier M. Aging effects on the metabolic
and cognitive energy cost of interlimb coordination. J Gerontol A Biol Sci
Med Sci 60: 312–319, 2005.
Stephenson JL, Lamontagne A, De Serres SJ. The coordination of upper and
lower limb movements during gait in healthy and stroke individuals. Gait
Posture 29: 11–16, 2009.
Stevens JA, Corso PS, Finkelstein EA, Miller TR. The costs of fatal and
non-fatal falls among older adults. Inj Prev 12: 290 –295, 2006.
Taga G, Yamaguchi Y, Shimizu H. Self-organized control of bipedal loco-
motion by neural oscillators in unpredictable environment. Biol Cybern 65:
147–159, 1991.
Thelen DG, Ashton-Miller JA, Schultz AB, Alexander NB. Do neural
factors underlie age differences in rapid ankle torque development? J Am
Geriatr Soc 44: 804 – 808, 1996.
Varraine E, Bonnard M, Pailhous J. Intentional on-line adaptation of stride
length in human walking. Exp Brain Res 130: 248 –257, 2000.
Walker ML, Austin AG, Banke GM, Foxx SR, Gaetano L, Gardner LA,
McElhiney J, Morris K, Penn L. Reference group data for the Functional
Gait Assessment. Phys Ther 87: 1468 –1477, 2007.
Wishart LR, Lee TD, Murdoch JE, Hodges NJ. Effects of aging on
automatic and effortful processes in bimanual coordination. J Gerontol B
Psychol Sci Soc Sci 55: P85–P94, 2000.
Wrisley DM, Kumar NA. Functional Gait Assessment: concurrent, discrim-
inative, and predictive validity in community-dwelling older adults. Phys
Ther 90: 761–773, 2010.
Wrisley DM, Marchetti GF, Kuharsky DK, Whitney SL. Reliability,
internal consistency, and validity of data obtained with the functional gait
assessment. Phys Ther 84: 906 –918, 2004.
Yamasaki T, Nomura T, Sato S. Phase reset and dynamic stability during
human gait. Biosystems 71: 221–232, 2003.
Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous
outcomes. Biometrics 42: 121–130, 1986.
Zeger SL, Liang KY, Albert PS. Models for longitudinal data: a generalized
estimating equation approach. Biometrics 44: 1049 –1060, 1988.

J Neurophysiol • doi:10.1152/jn.00950.2011 • www.jn.org

Downloaded from journals.physiology.org/journal/jn (045.230.084.251) on September 29, 2022.