ORIGINAL COMMUNICATION

Effects of dual tasking on the postural performance of peoplewith and without multiple sclerosis: a pilot study

Jesse V. Jacobs • Susan L. Kasser

Received: 7 September 2011 / Revised: 4 November 2011 / Accepted: 7 November 2011 / Published online: 8 December 2011

� Springer-Verlag 2011

Abstract People with multiple sclerosis (MS) exhibit

both cognitive and postural impairments. This study

examined the effects of MS and of dual tasking on postural

performance, and explored associations among dual-task

postural performance, cognitive capacity, fear of falling,

and fatigue. Thirteen subjects with MS (Expanded Dis-

ability Status Scale: 0–4.5) and 13 matched subjects

without MS performed three tasks of standing postural

control, with and without dual tasking amid an auditory

Stroop task: (1) step initiation, (2) forward leaning to the

limits of stability, and (3) postural responses to rotations of

the support surface. Two-factor general linear models were

used to evaluate differences between the groups (with or

without MS) and two conditions (single or dual tasking) for

each postural task. During step initiation, dual tasking

significantly delayed the onset of the anticipatory postural

adjustment (APA) more for the subjects with MS than for

those without MS, and step lengths increased for the sub-

jects with MS but decreased for those without MS. No

other significant group-by-condition interactions were

evident on the recorded variables of stepping, leaning,

postural responses, or Stroop-response accuracies and

latencies. The scores for the subjects with MS on the

Modified Fatigue Impact Scale significantly associated

with the change between single-task to dual-task conditions

in APA onset and foot-lift onset during step initiation as

well as in lean position variability and lean onset times

during forward leaning. The results suggest dual-task

effects were more evident during step initiation and are

associated with levels of fatigue for subjects with MS.

Keywords Dual task � Multiple sclerosis � Posture �Balance � Stroop

Introduction

Multiple sclerosis (MS) represents a common and disabling

neurologic disease associated with impaired postural con-

trol and, consequently, impaired mobility with an increased

risk of falls [1]. People with MS exhibit altered postural

control across several contexts of behavior, including static

stance in different sensory conditions, leaning or reaching

to the limits of stability, postural responses to a loss of

balance, as well as gait and gait initiation [2]. Many of

these activities reflect unique but overlapping neurological

functions [3–6], and many of these activities also associate

with past falls, future falls, or a person’s perceived risk of

falling for people with MS [7–10]. Therefore, when placed

in the context of the International Classification of Func-

tion, Disability, and Health (ICF), balance impairment

represents a multi-contextual construct at the level of both

function and activity that can affect the performance of

varied tasks and, subsequently, multiple roles of partici-

pation at work, in the home, or in leisure activity [11]. As

such, understanding how MS affects postural control across

many contexts of behavior may be required to identify and

develop optimal interventions for people with MS.

In addition to impaired postural control, approximately

half of individuals with MS exhibit cognitive impairment,

particularly during timed response tasks that challenge

processing speed [12, 13]. An activity’s cognitive demand

on an individual may represent an important contextual

J. V. Jacobs (&) � S. L. Kasser

Department of Rehabilitation and Movement Science,

University of Vermont, 305 Rowell Building,

106 Carrigan Drive, Burlington, VT 05405, USA

e-mail: jjacobs@uvm.edu

123

J Neurol (2012) 259:1166–1176

DOI 10.1007/s00415-011-6321-5

factor within the ICF by affecting the extent to which

people with MS present with impaired balance. Impor-

tantly, cognition and postural control appear to interact.

Specifically, people with MS report that states of confusion

or performing tasks under divided attention associate with

falling [1, 8], and performance on the timed up-and-go

(TUG) with a cognitive dual-task represents a significant

predictor of future falls in people with MS [14]. This

combination of postural and cognitive impairments in

people with MS, along with fall risk associating with the

performance of postural activities under conditions of

cognitive distraction, have thus lead to suggestions that

dual tasking should be included in clinical diagnostics and

interventions [2, 15–17]. Further research, however,

remains necessary to substantiate such recommendations.

The relative effects of dual tasking on the postural

performance of people with and without MS, however,

remain little tested, having been examined only during

steady-state gait or static stance. When walking with and

without performing a digit-recall task, subjects with mild-

to-moderate MS have been reported to exhibit greater

change with dual tasking in digit recall, walking speed and

swing time variability than a control group of subjects

without MS, and the extent of dual-task effect associated

with the subjects’ reported levels of fatigue [15]. Similarly,

recently-diagnosed subjects with clinically isolated syn-

drome have been reported to exhibit reduced gait velocity

and prolonged double support when walking while per-

forming a word list generation task, and these dual-task

associated changes in gait were not evident in a healthy

control group [16]. In addition to these studies of gait

performance, silent backward counting has been reported

to alter the variability of sway velocity when standing on

foam with eyes closed [18], but a modified Stroop task has

been reported to similarly affect sway parameters of sub-

jects with and without clinically isolated syndrome when

standing with eyes open on a firm surface [17].

Based on these studies, it is clear that very little is

known about whether and how dual tasking affects postural

control for people with MS. Most notably, other contexts of

impairment beyond standing balance under different sen-

sory conditions or gait require examination in order to

more thoroughly understand how MS affects posture con-

trol across multiple activities and environmental contexts.

Therefore, this study sought to quantify the effects of dual

tasking on the postural control of subjects with and without

MS when performing tasks in three different behavioral

contexts that have not yet been evaluated with dual tasking

in the literature: (1) step initiation, (2) forward leaning to

the limits of stability, and (3) postural responses to an

induced loss of balance. Dual tasking was required by way

of an auditory Stroop task, in which the subjects were

evaluated on both the speed and accuracy of their verbal

responses. This task was chosen because performance on

the cognitive task is easily quantified and because Stroop

tasks represent explicit timed response tasks capable of

challenging known impairments in processing speed in

people with MS [19]. Lastly, we explored associations

among dual-task related changes in postural performance

and clinical measures of cognitive impairment, fear of

falling, and fatigue as other potential contextual factors

within the ICF model that may affect the impact of dual

tasking on postural performance. We hypothesized that

subjects with MS would exhibit altered postural control

and less automated postural control, such that they would

be more susceptible to changes in postural control when

dual tasking. We further hypothesized that increased cog-

nitive impairment, fear of falling, and fatigue would limit

information processing capacity, increase processing

demands, and/or modify attention selection strategies so as

to associate with increased dual-task costs on postural

performance.

Methods

Subjects

Thirteen subjects with MS and 13 subjects without MS

participated in this pilot study (Table 1). Subjects with MS

were included if they were diagnosed by a neurologist to

have MS and have an Expanded Disability Status Scale

(EDSS) score of less than six, as well as no uncorrected

hearing or visual impairments. Subjects without MS were

included if they had no self-reported neurological, mus-

culoskeletal, or psychiatric disorders, as well as no

uncorrected hearing or visual impairments, and they were

matched to the subjects with MS to be of the same sex and

to be within 2 years of age. All subjects gave written

informed consent to participate in the protocol, which was

approved by the local institutional review board.

Protocol

The subjects stood barefoot on a moveable force platform,

with their arms at their sides and their feet placed in par-

allel at a heel-to-heel distance that corresponded to 11% of

their body height in order to perform three tasks: (1) cued

step initiation, (2) cued forward leaning to the limits of

stability, and (3) feet-in-place postural responses to toes-up

rotations of the support surface. For each of these tasks,

trials were randomly ordered for the subjects to either

perform the postural tasks alone or while also verbally

responding to an auditory Stroop task.

Prior to performing the postural tasks, however, the

subjects completed the Montreal Cognitive Assessment

J Neurol (2012) 259:1166–1176 1167

123

(MoCA; [20]), the Activities-specific Balance Confidence

(ABC) Scale [21], and the Modified Fatigue Impact Scale

(MFIS; [22]). Each of these scales has been validated in

subjects with MS [23–25]. The subjects also completed the

timed up-and-go (TUG), with and without a dual-task of

serial backwards counting by three, in order to assess a

common clinical analog of dual-tasked balance impairment

that has been validated in people with MS [14, 26]. The

total time needed to stand up, walk 3 m, turn around, walk

back, and sit down was recorded by stopwatch.

The subjects were then instructed on the auditory Stroop

task. The Stroop task required the subjects to verbally

respond into a headset microphone with the word ‘‘high’’ or

‘‘low’’ in response to the words ‘‘high’’ or ‘‘low’’ presented

in either a congruent or incongruent tone, such that their

response matched the tone of the presented stimulus rather

than the stated word. The subjects were instructed to respond

both as quickly and as accurately as possible to the stimuli.

The subjects were then presented with eight randomly

ordered practice stimuli that consisted of two stimuli per

possible permutation of congruent and incongruent word-

tone pairs. After ensuring the subjects understood the task,

they then responded to another eight Stroop stimuli, the

responses to which would serve as a single-task reference for

when performing the Stroop task with the postural tasks.

For the experimental task of step initiation, the subjects

were instructed to stand with equal weight under both feet

and then to step with their preferred limb, followed by a

matching step with the other limb, as soon as possible in

response to a visual ‘‘GO’’ signal presented on a monitor,

which was positioned 2.45 m ahead and at eye level. The

subjects were then instructed that some trials would require

them to step as soon as possible in response to the ‘‘GO’’

cue without the presentation of any ‘‘high’’ or ‘‘low’’

stimuli, whereas other trials would require them to first

begin responding as quickly and accurately as possible to

the ‘‘high’’ and ‘‘low’’ stimuli, and then step as soon as

possible when presented with the ‘‘GO’’ signal while

continuing to respond to the ‘‘high’’ and ‘‘low’’ stimuli. As

with the single-task Stroop condition, eight randomly

ordered Stroop stimuli were consecutively presented such

that the ensuing Stroop stimulus was presented just after

the subject responded to the prior stimulus. Without the

subjects’ knowledge, the ‘‘GO’’ signal was randomly pre-

sented to coincide with the second, third, or fourth Stroop

stimulus. The ‘‘GO’’ signal was paired with all possible

permutations of congruent or incongruent tone-word pairs.

Data were collected from 3 s prior to presenting the ‘‘GO’’

signal to 7 s after the ‘‘GO’’ signal.

The forward leaning task was presented similarly to the

step initiation task, except the subjects were instructed to

respond as soon as possible to the ‘‘GO’’ signal by leaning

forward as far as possible without losing their balance and

to hold that forward position for several seconds until

instructed to return to their natural standing position. The

subjects were further instructed to lean by bending at their

ankles, without bending at the waist, and to keep their feet

in place without lifting their heels. Data were collected

Table 1 Group characteristics

RR relapse-remitting, SPsecondary-progressivea MoCA scores are out of a

maximum score of 30. Scores

less than 26 indicate cognitive

impairment; two subjects with

MS scored \26b MFIS scores are out of a

maximum of 84. Scores higher

than 38 indicate disabling or

activity limiting levels of

fatigue; seven subjects with MS

scored [38c ABC scores are out of a

maximum of 100%. Scores

lower than 41% indicate fall

history in people with MS; no

subjects scored \41%

Characteristic Group Statistic (P value)

MS Control

Sex distribution

(number of females, males)

8.5 8.5 Fisher’s Chi2 = 0.00

P = 1.00

Age (year)

Mean (95% CI)

50 (43–56) 50 (43–57) T24 = 0.09

P = 0.93

Height (cm)

Mean (95% CI)

168 (162–174) 167 (160–173) T24 = 0.36

P = 0.72

Weight (kg)

Mean (95% CI)

71 (64–78) 66 (57–75) T24 = 0.89

P = 0.38

MoCA scorea

Mean (95% CI)

27 (26–29) 29 (28–29) T24 = 1.71

P = 0.10

MFIS scoreb

Mean (95% CI)

37 (28–47) 9 (2–16) T24 = 5.25

P \ 0.00005

ABC scorec (%)

Mean (95% CI)

79 (72–87) 98 (96–100) T24 = 5.19

P \ 0.00005

Duration since diagnosis (year)

Mean (95% CI)

9.5 (6.75–12.25) Not applicable Not compared

Type of MS (number of subjects) 12 RR, 1 SP Not applicable Not compared

EDSS score

Median (range)

2.5 (0–4.5) Not applicable Not compared

1168 J Neurol (2012) 259:1166–1176

123

from 3 s prior to presenting the ‘‘GO’’ signal to 10 s after

the ‘‘GO’’ signal.

The postural response task was also presented similarly

to the other two postural tasks, although the subjects looked

straight ahead at a fixation point on the wall rather than

having to respond to a go cue on the monitor. The subjects

were instructed to maintain standing balance in response to

the platform rotations, which were initiated at a random

time and consisted of a 5�, toes-up displacement with peak

velocity of 23�/s and duration of 490 ms. Due to equipment

limitations, the onset of the platform movement was gen-

erated by a manual key press of the experimenter, rather

than being triggered by the Stroop presentation software.

Thus, although the subjects were actively engaged in

responding to the consecutively presented Stroop stimuli,

the initiation of the platform movement and the onset of a

Stroop stimulus were not precisely coincidental.

Data collection and processing

Both kinematic and kinetic measures were used to quantify

postural performance. When performing the experimental

tasks, the subjects stood on a biomechanical force plate

(AMTI OR6-6; Watertown, MA, USA), from which the

forces and moments were used to generate the center of

pressure (CoP) under the subjects’ feet. The force-plate

signals were sampled at 1,000 Hz and then the CoP signals

were low-pass filtered offline to 10 Hz. Three-dimensional

kinematics were collected from a 7-camera passive-marker

motion capture system (VICON, Centennial, CO, USA).

Markers were placed to define the head, trunk, upper arms,

forearms, thighs, shanks, and feet. The kinematics were

then combined with published body segment parameters in

order to estimate the center of body mass (CoM) [27]. In

addition, a marker on the first toe was used to identify step

parameters during the step initiation task. Kinematics were

collected at 100 Hz and then low-pass filtered offline to

10 Hz. All outcome measures were processed using

MATLAB software (Mathworks, Natick, MA, USA).

The following outcome measures were derived to

quantify performance during step initiation (Fig. 1a). The

CoP traces were used to derive the onset latency and peak

displacement of the lateral and posterior components of the

anticipatory postural adjustment (APA; a posterior–lateral

shift in the CoP toward the initial swing limb that acts to

move the CoM forward and toward the stance limb in

preparation for stepping). Foot-lift onset latency, maximum

step velocity, and the end of the step were determined from

differentiating the first-toe marker’s position traces and then

identifying the onset of an excursion in the velocity trace

above zero, maximum velocity, and the end of the step as a

return to zero velocity. Step length was determined from the

toe marker’s displacement between foot-lift onset and the

end of the step. The duration of the APA was determined as

the time of APA onset to the time of foot-lift onset.

The following outcome measures were derived to

quantify performance during forward leaning (Fig. 1b).

The CoP traces were used to determine the onset latency of

forward leaning, maximum APA amplitude (the peak

backward shift in the CoP that also corresponds to the onset

of forward leaning), the average forward leaning position,

and the variability of the forward leaning position. The

establishment of the forward leaning position was deter-

mined by identifying the time that the CoP first displaced

beyond the average endpoint CoP position, and this aver-

age endpoint position was defined from the final second of

each recorded trial. The average lean position and lean

position variability were then defined as the mean and

standard deviation of the CoP position, respectively, from

the establishment of the forward lean position to the end of

the trial. We also quantified the initial distance of the CoM

from the front of the feet as the average distance during the

500 ms that preceded the ‘‘GO’’ signal. In addition, we

quantified the average CoM displacement of the estab-

lished forward leaning position as well as the minimum

distance between the CoM and the front of the feet.

The following outcome measures were derived to

quantify performance during postural responses to platform

rotations (Fig. 1c). The CoP traces were used to determine

the peak and time-to-peak forward and backward response

displacements as well as the peak-to-peak displacements. In

addition, we quantified the initial distance of the CoM from

the heel as the average distance during the 500 ms that

preceded the perturbation, as well as the peak backward

CoM displacement and the minimum distance between the

CoM and the heels in response to the platform rotations.

Performance on the Stroop task was determined by

reaction time using Superlab software (Cedrus, San Pedro,

CA, USA). This software identified the onset of the voice

response into the headset’s microphone relative to stimulus

presentation. Two experimenters recorded response accu-

racy during the study, which was determined from the first

attempted response following the presentation of a stimu-

lus. When performing the Stroop task with either the step

or lean tasks, average Stroop response latencies and

accuracy were determined only from the stimuli that were

presented at the same time as the ‘‘GO’’ signal. Because

Stroop stimuli were not time locked to the platform rota-

tions during the postural response task, average Stroop

response latencies and accuracy during the postural

response task were determined from all presented stimuli.

For all postural tasks, the CoP and CoM positions or

displacements were normalized as a percentage of the

subjects’ foot length or heel-to-heel stance width. All out-

come measures were determined for each trial, and then

averaged by condition and subject for statistical analysis.

J Neurol (2012) 259:1166–1176 1169

123

Each outcome measure during the single-task condition was

subtracted from the value of that measure obtained during

the dual-task condition in order to generate measures of

dual-task effects that could be applied to regression analy-

ses evaluating whether average ABC scale scores, total

MoCA scores, total MFIS scores, and EDSS scores of the

subjects with MS provide significant independent predictors

of dual-task effects on postural performance.

Statistical analysis

Statistical analysis was carried out using SPSS version 19

(IBM, Armonk, NY, USA). Differences between the sub-

jects with and without MS in age, height, weight, ABC

scores, MoCA scores, and MFIS scores were determined

by two-tailed t tests. Group differences in sex distribution

were determined by a Fisher’s exact test. Two-factor

general linear models were used to evaluate differences in

the outcome measures of postural and Stroop performance

between the groups with and without MS (between-groups

factor) and between the single- and dual-task conditions

(repeated-measures factor). Stepwise linear regression

models were used to identify significant independent

associations for the group with MS among ABC scores,

MoCA scores, MFIS scores, and EDSS scores with dual-

task related differences in outcome measures that were

identified to exhibit significant group main effects or sig-

nificant group-by-condition interactions. Dual-task related

changes in TUG completion times were also associated by

Pearson correlations with these dual-task related changes in

outcome measures of step initiation, leaning, or postural

responses that exhibit significant differences between

subjects with and without MS. Significance was defined as

a p value less than 0.05.

Results

Step initiation

Dual tasking differentially affected the step initiation of the

subjects with and without MS. Specifically, the onset

Fig. 1 Representative CoP, CoM, and kinematic traces from indi-

vidual subjects with and without MS to illustrate measures of a step

initiation, b forward leaning to the limits of stability, and c postural

responses to rotations of the support surface. Solid, black tracesrepresent the single-task condition, and the dashed, gray traces

represent the dual-task condition. The left column within each of a, b,

and c represents traces from subjects without MS, and the rightcolumn represents traces from subjects with MS. Text, arrows,

brackets, and shade boxes highlight outcome measures derived from

these traces

1170 J Neurol (2012) 259:1166–1176

123

latencies of the APA’s posterior component were signifi-

cantly more delayed in the dual-task condition relative to

the single-task condition for the subjects with MS (Fig. 2a).

In addition, step lengths increased from the single-task to

the dual-task condition for the subjects with MS, but step

lengths decreased for the subjects without MS (Fig. 2b).

The group-by-condition interactions did not reach statisti-

cal significance for Stroop response times [F1,24 = 0.87,

P = 0.36], Stroop response accuracy [F1,24 = 2.68,

P = 0.12], peak step velocity [F1,24 = 0.72, P = 0.40],

foot-lift onset [F1,24 = 2.42, P = 0.13], APA duration

[F1,24 = 0.14, P = 0.71], lateral APA onset [F1,24 = 1.67,

P = 0.21] or peak amplitude [F1,24 = 0.23, P = 0.64], or

posterior APA peak amplitude [F1,24 = 0.06, P = 0.82].

Step initiation was further affected by dual tasking for

both groups, and the subjects with MS exhibited other

aspects of altered step initiation compared to the subjects

without MS across both conditions (i.e., main effects of

group and of condition were additionally evident). Spe-

cifically, the onset of the APA’s lateral component was

significantly later, APA durations were significantly

longer, and foot-lift onset latencies were significantly

later in the dual-task condition than in the single-task

condition (Table 2). In addition, Stroop responses were

significantly later and less accurate in the dual-task

condition than in the single-task condition (Table 2).

Regarding differences between groups, the subjects with

MS exhibited significantly longer APA durations and

significantly later foot-lift onset times than the subjects

without MS (Table 2).

Forward leaning

Dual tasking did not differentially affect the forward

leaning of the groups with and without MS. The group-by-

condition interactions did not reach statistical significance

for Stroop response times [F1,24 = 0.42, P = 0.52], Stroop

response accuracy [F1,24 = 0.68, P = 0.42], initial ante-

rior-posterior CoM position [F1,21 = 1.69, P = 0.21],

average CoM displacement [F1,21 = 0.59, P = 0.45],

minimum CoM distance from the front of the feet [F1,21 =

0.67, P = 0.42], average lean velocity [F1,24 = 1.61,

P = 0.22], peak APA amplitude [F1,24 = 0.27, P = 0.61],

lean onset time [F1,24 = 1.80, P = 0.19], average CoP

displacement [F1,24 = 2.71, P = 0.11], or lean position

variability [F1,24 = 3.65, P = 0.07]. Dual tasking did,

however, affect the forward leaning of both groups (sig-

nificant condition main effects), and the subjects with MS

exhibited altered lean parameters compared to the subjects

without MS (significant group main effects). Specifically,

the dual-task condition associated with significantly

delayed lean onset and Stroop reaction times as well as

significantly decreased CoM displacements relative to the

single-task condition (Table 2). In addition, the subjects

with MS exhibited significantly later lean onset times,

significantly smaller CoP displacements, and significantly

larger variability of the CoP leaning position than the

subjects without MS (Table 2).

Postural responses

Dual tasking did not differentially affect the postural

responses of the groups with and without MS. The group-

by-condition interactions did not reach statistical signifi-

cance for Stroop response times [F1,24 = 0.14, P = 0.71],

Stroop response accuracy [F1,24 = 0.01, P = 0.94], initial

CoM position from the heel [F1,23 = 0.08, P = 0.78], peak

backward CoM displacement [F1,23 = 1.07, P = 0.31],

minimum CoM distance from the back of the feet

[F1,23 = 0.46, P = 0.51], peak and time-to-peak backward

CoP displacement [F1,24 = 0.40, P = 0.54; F1,24 \ 0.001,

Fig. 2 Group mean (95% confidence interval) a APA onset latencies

and b step lengths for each group and condition. Gray lines with filledcircles represent measures for the group with MS, and black lines withunfilled boxes represent measures for the group without MS. These

measures are highlighted for their significant group-by-condition

interaction effects

J Neurol (2012) 259:1166–1176 1171

123

P [ 0.99], peak and time-to-peak forward CoP displace-

ment [F1,24 = 0.23, P = 0.64; F1,24 = 0.06, P = 0.82], or

peak-to-peak CoP displacement [F1,24 = 0.61, P = 0.44].

Dual tasking did, however, affect the postural responses of

both groups (significant condition main effects), and the

subjects with MS exhibited altered response parameters

compared to the subjects without MS (significant group

main effects). Specifically, the dual-task condition associ-

ated with significantly delayed time-to-peak backward CoP

displacements and significantly farther initial CoM posi-

tions from the heels relative to the single-task condition

(Table 2). In addition, the subjects with MS exhibited

significantly later time-to-peak backward CoP displace-

ments and significantly larger peak-to-peak CoP displace-

ments than the subjects without MS (Table 2).

Associations with fatigue, cognition, balance

confidence, and disease severity for the subjects

with MS

During step initiation, higher MFIS scores provided the

only significant independent predictor of dual-task related

increases in APA onset times and foot-lift onset latencies

[bMFIS = 11.57, r2 = 0.40, F1,11 = 7.36, P = 0.020 for

APA onset times; bMFIS = 12.02, r2 = 0.44, F1,11 = 8.73,

P = 0.013 for foot-lift onset times]. During forward lean-

ing, higher MFIS scores and lower MoCA scores provided

significant independent predictors of dual-task related

increases in lean onset times [overall model r2 = 0.67,

F2,10 = 10.19, P = 0.004; bMFIS = 8.76, T = 3.20, P =

0.009; bMoCA = -43.81, T = 2.72, P = 0.022]. In addition,

Table 2 Mean (95% CI) Outcomes demonstrating significant main effects between groups or conditions

Outcome measure Condition Group Main effect reaching significance

MS Control

Step initiation task

Foot-lift onset time (ms) Single task 1,013 (941–1,085) 919 (860–979) Group: F1,24 = 5.81, P = 0.024

Dual task 1,288 (1,088–1,488) 1,054 (935–1,174) Condition: F1,24 = 20.67, P \ 0.0,005

APA duration (ms) Single task 531 (477–585) 447 (417–477) Group: F1,24 = 6.71, P = 0.016

Dual task 551 (479–624) 475 (443–508) Condition: F1,24 = 5.21, P = 0.032

Lateral APA onset time (ms) Single task 482 (445–518) 468 (424–512) Condition: F1,24 = 11.86, P = 0.002

Dual task 685 (507–863) 560 (445–675)

Stroop reaction time (ms) Single task 1,107 (992–1,222) 1,063 (902–1,224) Condition: F1,24 = 8.32, P = 0.008

Dual task 1,247 (1,148–1,347) 1,337 (1,051–1,623)

Stroop response accuracy (% correct) Single task 98 (94–100) 99 (97–100) Condition: F1,24 = 9.18, P = 0.006

Dual task 95 (89–100) 90 (80–99)

Forward leaning task

Lean onset time (ms) Single task 755 (705–805) 715 (658–771) Group: F1,24 = 4.84, P = 0.038

Dual task 942 (778–1,105) 779 (703–856) Condition: F1,24 = 11.94, P = 0.002

CoP lean displacement (% foot length) Single task 25 (21–30) 32 (28–36) Group: F1,24 = 6.03, P = 0.022

Dual task 25 (20–30) 30 (27–34)

CoP lean variability (% foot length) Single task 3.4 (3.0–3.9) 2.1 (1.8–2.3) Group: F1,24 = 21.44, P \ 0.0005

Dual task 3.0 (2.6–3.4) 2.1 (1.7–2.5)

CoM lean displacement (% foot length) Single task 33 (28–38) 38 (33–43) Condition: F1,21 = 5.16, P = 0.034

Dual task 32 (26–38) 36 (31–41)

Stroop reaction time (ms) Single task 1,114 (989–1,239) 1,041 (862–1,219) Condition: F1,24 = 6.90, P = 0.015

Dual task 1,227 (1,086–1,368) 1,205 (1,065–1,345)

Postural response task

Time-to-peak backward

CoP displacement (ms)

Single task 264 (243–284) 224 (210–239) Group: F1,24 = 12.75, P = 0.002

Dual task 268 (246–289) 230 (218–242) Condition: F1,24 = 4.97, P = 0.035

Peak CoP displacement

(% foot length)

Single task 35 (30–39) 28 (25–31) Group: F1,24 = 4.76, P = 0.039

Dual task 36 (27–46) 28 (25–-31)

Initial CoM distance from heel

(% foot length)

Single task 49 (44–54) 47 (42–51) Condition: F1,23 = 4.56, P = 0.044

Dual task 51 (46–55) 49 (45–52)

1172 J Neurol (2012) 259:1166–1176

123

higher MFIS scores provided the only significant inde-

pendent predictor of lean position variability [bMFIS =

0.0003, r2 = 0.41, F1,11 = 7.50, P = 0.019]. MFIS,

MoCA, ABC, and EDSS scores did not provide significant

independent predictors of dual-task related changes in any

other tested variables, including those pertaining to pos-

tural responses.

Effects of dual-tasking on the TUG, and associations

with dual-task effects on the three laboratory postural

tasks

The subjects with MS exhibited significantly longer TUG

completion times [group main effect: F1,21 = 18.38,

P = 0.00033], and TUG times increased more with dual

tasking for the group with MS than for those without MS

[group-by-condition interaction: F1,21 = 4.80, P = 0.040].

The mean (95% CI) TUG completion times without dual

tasking were 8.5 (7.2–9.7) seconds for the subjects with

MS and 6.2 (5.7–6.7) seconds for the subjects without MS.

With dual tasking, mean TUG times were 10.4 (8.7–12.0)

seconds for the subjects with MS and 6.5 (5.8–7.2) seconds

for the subjects without MS. Higher MFIS scores provided

the only significant independent predictor of dual-task

related increases in TUG completion times [bMFIS = 0.083,

r2 = 0.33, F1,11 = 5.43, P = 0.040]. Dual-task related

differences in TUG completion times did not significantly

correlate with any evaluated measure of dual-task related

differences in step initiation, leaning, or postural

responses [range of Pearson r = 0.014–0.482, range of

P = 0.10–0.96].

Discussion

The results partially supported our hypothesis that subjects

with MS would exhibit altered postural control and would

be more susceptible to changes in postural control when

dual tasking. Significant differences between subjects with

and without MS were evident across all tasks, but the

effects of dual tasking were significantly different between

subjects with and without MS only during the TUG and

during step initiation. Specifically, dual tasking differen-

tially affected the two groups’ TUG completion times,

APA onset times and step lengths. Although no other sig-

nificant interactions were evident between groups and

conditions, many parameters across all three of the labo-

ratory tasks of posture control were sensitive to dual

tasking for both groups. During step initiation, the onset

and duration of the APA and foot lift were later and longer,

and Stroop responses were also more delayed and less

accurate. During forward leaning, dual tasking delayed and

decreased the lean, and Stroop responses were again

delayed. During the postural response task, dual tasking

delayed the time to peak backward CoP displacements, and

the subjects prepared for the task by starting with a more

forward initial CoM position. Thus, although all tasks were

sensitive to dual tasking, the TUG and step initiation were

most sensitive to identifying MS-associated changes in

postural control under dual-task conditions.

As a pilot study, the evidence remains insufficient to

direct clinical decisions, but they would suggest that clin-

ical interventions on balance would most need to account

for cognitive interactions during postural transitions and

gait activities. These results are consistent with reports that

fall risk and perceived fall risk are increased when per-

forming activities under cognitive distraction as well as

during gait or postural transitions [8, 14]. Based on these

findings, further research should focus on the effects of

dual-task training on functional tasks that require gait or

postural transitions, as well as the effectiveness of behav-

ioral strategies to prioritize these postural activities over

cognitive distractions.

Although group differences in dual-task effects were

most evident during the TUG and step initiation, all tasks

were sensitive to differences between subjects with and

without MS across the two conditions, as evidenced by the

significant statistical main effects of group. The subjects

with MS exhibited APAs of longer duration and, conse-

quently, more delayed foot-lift latencies when initiating a

step than the subjects without MS. The subjects with MS

also exhibited delayed leaning with smaller displacements

and a more variable leaning position than the subjects

without MS. Lastly, the subjects with MS responded to the

platform rotations with delayed and larger CoP displace-

ments relative to the subjects without MS. Thus, the greater

sensitivity of step initiation to identify larger dual-task

costs in subjects with MS than leaning or postural response

tasks does not appear to be because the leaning or postural

response tasks were less sensitive to identifying MS-related

alterations in postural control.

It was unexpected that differences in Stroop response

accuracies or latencies were not evident between the

groups with and without MS. The Stroop task was chosen

because previous studies had identified that people with

MS exhibit delayed Stroop responses compared to subjects

without MS, even in mild stages of the disease [28–32].

These other studies, however, generally employed a more

complex visual color-word paradigm with more than two

response choices, whereas our study employed a two-

choice auditory Stroop paradigm. These differences in

sensory modality and task complexity may have decreased

the ability of this study’s Stroop task to detect differences

associated with MS.

The Stroop task, however, still impacted postural per-

formance on all three postural tasks, and two of the

J Neurol (2012) 259:1166–1176 1173

123

postural tasks impacted Stroop performance. In addition,

the step initiation of subjects with and without MS was

differentially affected by the Stroop task. Thus, the Stroop

task was sufficient to compete with the processing demands

of the postural tasks. This study’s auditory Stroop task has

been used to evaluate dual-task effects on gait in healthy

young adults as well as older adults with and without

balance impairments [33–35]. The findings of these studies

vary, such that dual-task costs depend on the complexity of

the gait tasks, may affect either Stroop performance or gait

performance, and are dependent on how the subjects pri-

oritize performance toward either the Stroop task or the

gait task. Therefore, this study’s Stroop task is sufficient to

elicit dual-task effects on postural control, and whether

effects were evident likely relate to a complex interaction

among combined task demands, processing capacity, and

prioritization of attention to either the cognitive or postural

task. Further study may benefit from manipulating the

difficulty of each postural task as well as the instructional

set to prioritize either the cognitive task or the postural task

in order to disentangle how MS affects processing

demands, capacity, and selection strategies within each

postural activity.

The relative effects of dual tasking on step initiation,

leaning, and postural responses suggest a difference in the

nervous system’s utilization of competing executive neural

resources for performing the cognitive and postural tasks.

Dual tasking affected five of the measured parameters

during step initiation, three parameters of leaning, and two

parameters of postural responses. The relative effectiveness

of dual tasking to change performance on these postural

tasks could reflect a learning effect due to the ordering of

the tasks. The results, however, are consistent with the

increased level of complexity and dynamic postural coor-

dination required of step initiation relative to leaning as

well as the increased level of voluntary influence on step

initiation and leaning relative to induced postural

responses.

This study also demonstrated that the TUG was sensitive

to differences in dual-task effects between subjects with

and without MS. Interestingly, dual-task related changes in

TUG performance did not correlate with the dual-task

related changes in any measure of step initiation, leaning,

or postural responses, and this lack of association occurred

despite many shared associations with fatigue. These

results may reflect differences in the influence of the cog-

nitive tasks on the neural functions underlying gait per-

formance versus those underlying the other postural

activities. Although the TUG includes step initiation, the

relative influence of step initiation on the TUG completion

time compared to the continuous gait and turns required of

the task is likely very low, thereby diminishing associations

between TUG completion times and measures of step

initiation. The differences in cognitive processing for the

continuous backward counting task required of the TUG

versus the discrete responses required of the Stroop task

may also have contributed to the lack of associations

among the subjects’ performance on the TUG and their

performance on the other postural tasks.

The regression analysis demonstrated that dual-task

effects on TUG, step initiation and forward leaning were

associated with our subject sample’s characteristics. Most

notably, increased TUG times and delays in step initiation

or leaning were associated with higher self-reported levels

of fatigue. These results parallel a previous report that

dual-task related decreases in the walking speed of subjects

with MS are also significantly correlated with their self-

reported levels of fatigue [15]. Thus, further study on the

effects of dual tasking on the posture and gait of subjects

with MS may benefit from evaluating interventions that

modify fatigue.

As a pilot study, this study offers preliminary evidence

that the effects of MS on dual tasking may be relatively

stronger for some postural activities (e.g., gait and step

initiation) than others and, likewise, fatigue appears more

strongly associated with dual-task effects during gait, step

initiation and leaning than when responding to an induced

loss of balance. This study, however, represents a small

sample of people with generally mild, relapse-remitting

MS. This study’s findings, therefore, provide insights into

the relative effects of dual tasking on these postural

activities, but likely under-estimate differences between

people with and without MS and require further study to

assess their generalizability. In addition, this study focused

on the relative effects of one type of cognitive task on the

performance of multiple postural activities. Thus, there

remains a need for larger studies that could provide sub-

group analyses on people at different stages of MS severity

and with different types of MS, as well as providing

thorough examinations of clinically meaningful changes in

postural performance under dual-task conditions. Further

investigation could also evaluate the relative effects of

different cognitive tasks that challenge other cognitive

processes than those involved in Stroop tasks. Together,

such studies will provide a more thorough understanding

for how task demands and a person’s characteristics

interact across different postural activities to elicit balance

impairment in order to accurately inform clinical decision-

making.

In summary, postural control is sensitive to dual tasking

across several contexts of control, and having mild MS

likewise associates with altered postural control across

these tasks. Increased dual-task costs on postural control,

however, are relatively limited in subjects with mild,

relapse-remitting MS to gait and step initiation over lean-

ing to the limits of stability or responding to an induced

1174 J Neurol (2012) 259:1166–1176

123

loss of balance. Dual-task costs, though, appear to increase

with higher levels of fatigue for gait, step initiation and

leaning to the limits of stability. Further study is, therefore,

warranted on the responsiveness of dual-task costs on

posture control for clinical diagnostics and prognostics, as

well as on the effects of dual-task training or of interven-

tions to modify fatigue in order to improve or protect

posture, balance, and mobility.

Acknowledgments The authors wish to thank Parminder Padgett

and Kelley Groll for their assistance during data collection and their

contributions to data processing, as well as the University of Ver-

mont’s Department of Neurology and the Greater New England

Chapter of the National Multiple Sclerosis Society for their assistance

with subject recruitment. This study was funded by a Research

Incentive Grant from the College of Nursing and Health Sciences at

the University of Vermont.

Conflict of interest None.

References

1. Finlayson ML, Peterson EW, Cho CC (2006) Risk factors for

falling among people aged 45 to 90 years with multiple sclerosis.

Arch Phys Med Rehabil 87(9):1274–1279. doi:10.1016/j.apmr.

2006.06.002 (Quiz 1287)

2. Cameron MH, Lord S (2010) Postural control in multiple scle-

rosis: implications for fall prevention. Curr Neurol Neurosci Rep

10(5):407–412. doi:10.1007/s11910-010-0128-0

3. Massion J (1992) Movement, posture and equilibrium: interaction

and coordination. Prog Neurobiol 38(1):35–56. pii:0301-0082

(92)90034-C

4. Jacobs JV, Horak FB (2007) Cortical control of postural

responses. J Neural Transm 114(10):1339–1348. doi:10.1007/

s00702-007-0657-0

5. Takakusaki K (2008) Forebrain control of locomotor behaviors.

Brain Res Rev 57(1):192–198. doi:10.1016/j.brainresrev.2007.

06.024

6. Del Percio C, Brancucci A, Bergami F, Marzano N, Fiore A, Di

Ciolo E, Aschieri P, Lino A, Vecchio F, Iacoboni M, Gallamini

M, Babiloni C, Eusebi F (2007) Cortical alpha rhythms are cor-

related with body sway during quiet open-eyes standing in ath-

letes: a high-resolution EEG study. Neuroimage 36(3):822–829.

doi:10.1016/j.neuroimage.2007.02.054

7. Cattaneo D, De Nuzzo C, Fascia T, Macalli M, Pisoni I, Cardini

R (2002) Risks of falls in subjects with multiple sclerosis. Arch

Phys Med Rehabil 83(6):864–867. pii:S000399930267265X

8. Nilsagard Y, Denison E, Gunnarsson LG, Bostrom K (2009)

Factors perceived as being related to accidental falls by persons

with multiple sclerosis. Disabil Rehabil 31(16):1301–1310. doi:

10.1080/09638280802532639

9. Matsuda PN, Shumway-Cook A, Bamer AM, Johnson SL,

Amtmann D, Kraft GH (2011) Falls in multiple sclerosis. PM R

3(7):624–632. doi:10.1016/j.pmrj.2011.04.015 (Quiz 632)

10. Kasser SL, Jacobs JV, Foley JT, Cardinal BJ, Maddalozzo GF

(2011) A prospective evaluation of balance, gait, and strength to

predict falling in women with multiple sclerosis. Arch Phys Med

Rehabil. doi:10.1016/j.apmr.2011.06.004

11. Peterson EW, Cho CC, Finlayson ML (2007) Fear of falling and

associated activity curtailment among middle aged and older

adults with multiple sclerosis. Mult Scler 13(9):1168–1175. doi:

10.1177/1352458507079260

12. Langdon DW (2011) Cognition in multiple sclerosis. Curr Opin

Neurol 24(3):244–249. doi:10.1097/WCO.0b013e328346a43b

13. Hoffmann S, Tittgemeyer M, von Cramon DY (2007) Cognitive

impairment in multiple sclerosis. Curr Opin Neurol 20(3):275–

280. doi:10.1097/WCO.0b013e32810c8e87

14. Nilsagard Y, Lundholm C, Denison E, Gunnarsson LG (2009)

Predicting accidental falls in people with multiple sclerosis–a

longitudinal study. Clin Rehabil 23(3):259–269. doi:10.1177/

0269215508095087

15. Hamilton F, Rochester L, Paul L, Rafferty D, O’Leary CP, Evans

JJ (2009) Walking and talking: an investigation of cognitive-

motor dual tasking in multiple sclerosis. Mult Scler 15(10):

1215–1227. doi:10.1177/1352458509106712

16. Kalron A, Dvir Z, Achiron A (2010) Walking while talking–

difficulties incurred during the initial stages of multiple sclerosis

disease process. Gait Posture 32(3):332–335. doi:10.1016/

j.gaitpost.2010.06.002

17. Kalron A, Dvir Z, Achiron A (2011) Effect of a cognitive task on

postural control in patients with a clinically isolated syndrome

suggestive of multiple sclerosis. Eur J Phys Rehabil Med.

pii:R33112340

18. Negahban H, Mofateh R, Arastoo AA, Mazaheri M, Yazdi MJ,

Salavati M, Majdinasab N (2011) The effects of cognitive loading

on balance control in patients with multiple sclerosis. Gait Pos-

ture. doi:10.1016/j.gaitpost.2011.06.023

19. Denney DR, Gallagher KS, Lynch SG (2011) Deficits in pro-

cessing speed in patients with multiple sclerosis: evidence from

explicit and covert measures. Arch Clin Neuropsychol

26(2):110–119. doi:10.1093/arclin/acq104

20. Nasreddine ZS, Phillips NA, Bedirian V, Charbonneau S,

Whitehead V, Collin I, Cummings JL, Chertkow H (2005) The

Montreal Cognitive Assessment, MoCA: a brief screening tool

for mild cognitive impairment. J Am Geriatr Soc 53(4):695–699.

doi:10.1111/j.1532-5415.2005.53221.x

21. Powell LE, Myers AM (1995) The Activities-specific Balance

Confidence (ABC) Scale. J Gerontol A Biol Sci Med Sci 50A(1):

M28–M34

22. Ritvo PG, Fischer JS, Miller DM, Andrews H, Paty D, LaRocca

NG (1997) Multiple sclerosis quality of life inventory: a user’s

manual. National Multiple Sclerosis Society, New York, NY

23. Cattaneo D, Regola A, Meotti M (2006) Validity of six balance

disorders scales in persons with multiple sclerosis. Disabil

Rehabil 28(12):789–795. doi:10.1080/09638280500404289

24. Flachenecker P, Kumpfel T, Kallmann B, Gottschalk M, Grauer

O, Rieckmann P, Trenkwalder C, Toyka KV (2002) Fatigue in

multiple sclerosis: a comparison of different rating scales and

correlation to clinical parameters. Mult Scler 8(6):523–526

25. Krupp L, Rosicki C, Urbanovich A, Serafin D, Charvet L, Bhise

V, Greenblatt D, Giacinto J, Christodoulou C (2011) The Mon-

treal Cognitive Assessment (MOCA) as a Screening Tool for

Cognitive Functioning in Multiple Sclerosis (MS). In: American

Academy of Neurology Annual Meeting, Honolulu, Hawaii.

26. Nilsagard Y, Lundholm C, Gunnarsson LG, Dcnison E (2007)

Clinical relevance using timed walk tests and ‘timed up and go’

testing in persons with multiple sclerosis. Physiother Res Int

12(2):105–114

27. Dempster WT (1955) Space requirements of the seated operator:

geometrical, kinematic, and mechanical aspects of the body with

special reference to the limbs. vol v. 1. Wright Air Development

Center, Air Research and Development Command, U.S. Air

Force

28. Deloire MS, Bonnet MC, Salort E, Arimone Y, Boudineau M,

Petry KG, Brochet B (2006) How to detect cognitive dysfunction

at early stages of multiple sclerosis? Mult Scler 12(4):445–452

29. Denney DR, Lynch SG (2009) The impact of multiple sclerosis

on patients’ performance on the Stroop Test: processing speed

J Neurol (2012) 259:1166–1176 1175

123

versus interference. J Int Neuropsychol Soc 15(3):451–458. doi:

10.1017/S1355617709090730

30. Macniven JA, Davis C, Ho MY, Bradshaw CM, Szabadi E,

Constantinescu CS (2008) Stroop performance in multiple scle-

rosis: information processing, selective attention, or executive

functioning? J Int Neuropsychol Soc 14(5):805–814. doi:

10.1017/S1355617708080946

31. Vitkovitch M, Bishop S, Dancey C, Richards A (2002) Stroop

interference and negative priming in patients with multiple sclero-

sis. Neuropsychologia 40(9):1570–1576. pii:S0028393202000222

32. Balsimelli S, Mendes MF, Bertolucci PH, Tilbery CP (2007)

Attention impairment associated with relapsing-remitting multi-

ple sclerosis patients with mild incapacity. Arq Neuropsiquiatr

65(2A):262–267. pii:S0004-282X2007000200014

33. Siu KC, Catena RD, Chou LS, van Donkelaar P, Woollacott MH

(2008) Effects of a secondary task on obstacle avoidance in

healthy young adults. Exp Brain Res 184(1):115–120. doi:

10.1007/s00221-007-1087-9

34. Siu KC, Chou LS, Mayr U, Donkelaar P, Woollacott MH (2008)

Does inability to allocate attention contribute to balance con-

straints during gait in older adults? J Gerontol A Biol Sci Med Sci

63(12):1364–1369. pii:63/12/1364

35. Siu KC, Lugade V, Chou LS, van Donkelaar P, Woollacott MH

(2008) Dual-task interference during obstacle clearance in heal-

thy and balance-impaired older adults. Aging Clin Exp Res

20(4):349–354. pii:4886

1176 J Neurol (2012) 259:1166–1176

123