O

MGBJ

ett1

sa

i

e1na

wdd

ajalsvaotst

pgTm

gQGoMe

sz

sGa

1596

A

RIGINAL ARTICLE

odality-Specific, Multitask Locomotor Deficits Persist Despiteood Recovery After a Traumatic Brain Injury

radford J. McFadyen, PhD, Jean-François Cantin, PhD, Bonnie Swaine, PhD, Guylaine Duchesneau, MPs,

ulien Doyon, PhD, Denyse Dumas, PT, Philippe Fait, MSc

a(

M

Btomipibt

sledt

birpaacm

avPlmospesc

saacc

ABSTRACT. McFadyen BJ, Cantin J-F, Swaine B, Duch-sneau G, Doyon J, Dumas D, Fait P. Modality-specific, multi-ask locomotor deficits persist despite good recovery after araumatic brain injury. Arch Phys Med Rehabil 2009;90:596-1606.

Objective: To study the effects of sensory modality ofimultaneous tasks during walking with and without obstaclesfter moderate to severe traumatic brain injury (TBI).

Design: Group comparison study.Setting: Gait analysis laboratory within a postacute rehabil-

tation facility.Participants: Volunteer sample (N�18). Persons with mod-

rate to severe TBI (n�11) (9 men, 3 women; age, 37.56�3.79y) and a comparison group (n�7) of subjects withouteurologic problems matched on average for body mass indexnd age (4 men, 3 women; age, 39.19�17.35y).

Interventions: Not applicable.Main Outcome Measures: Magnitudes and variability for

alking speeds, foot clearance margins (ratio of foot clearanceistance to obstacle height), and response reaction times (bothirect and as a relative cost because of obstacle avoidance).Results: The TBI group had well-recovered walking speeds

nd a general ability to avoid obstacles. However, these sub-ects did show lower trail limb toe clearances (P�.003) acrossll conditions. Response reaction times to the Stroop tasks wereonger in general for the TBI group (P�.017), and this grouphowed significant increases in response reaction times for theisual modality within the more challenging obstacle avoid-nce task that was not observed for control subjects. A measuref multitask costs related to differences in response reactionimes between obstructed and unobstructed trials also onlyhowed increased attention costs for the visual over the audi-ory stimuli for the TBI group (P�.002).

Conclusions: Mobility is a complex construct, and theresent results provide preliminary findings that, even afterood locomotor recovery, subjects with moderate to severeBI show residual locomotor deficits in multitasking. Further-ore, our results suggest that sensory modality is important,

From the Center for Interdisciplinary Research in Rehabilitation and Social Inte-ration and the Department of Rehabilitation, Faculty of Medicine, Laval University,uébec (McFadyen, Fait); Center for Interdisciplinary Research in Rehabilitation ofreater Montreal and Montréal Rehabilitation Institute, Montréal (Swaine); Schoolf Rehabilitation (Swaine) and the Department of Psychology (Doyon), University ofontréal, Montréal; and Québec Rehabilitation Institute (Cantin, Dumas, Duch-

sneau), Québec, Québec, Canada.Supported by the Canadian Institutes of Health Research (grant no. 64408).No commercial party having a direct financial interest in the results of the research

upporting this article has or will confer a benefit on the authors or on any organi-ation with which the authors are associated.

Correspondence to Bradford J. McFadyen, PhD, Center for Interdisciplinary Re-earch in Rehabilitation and Social Integration, 525 Hamel, Québec, Québec, Canada,1M 2S8, e-mail: brad.mcfadyen@rea.ulaval.ca. Reprints are not available from the

uthor.

0003-9993/09/9009-00190$36.00/0doi:10.1016/j.apmr.2009.03.010

rch Phys Med Rehabil Vol 90, September 2009

nd greater multitask costs occur during sensory competitionie, visual interference).

Key Words: Attention; Gait; Locomotion; Rehabilitation.© 2009 by the American Congress of Rehabilitationedicine

EING MOBILE IS CRUCIAL for independent living andquality of life and is a major focus of physical rehabilita-

ion programs. Mobility, however, is a complex construct notnly requiring physical integrity related to the coordination ofuscles and movements about many joints but also cognitive

ntegrity to attend to and process sensory information and tolan and navigate the surrounding environment. Brain injuries,n particular a TBI, result in many different sequelae involvingoth cognitive and motor abilities that can affect one’s abilityo navigate the complex environments found in daily life.

Locomotor capacity and mobility after a TBI have not beentudied nearly as often as with other populations with neuro-ogic impairments (eg, stroke, cerebral palsy). Yet, work ismerging across different injury severities showing changes inynamic equilibrium after mild TBI1 annd slowing and cau-ious walking after moderate and severe TBIs.2,3

The study of combined cognitive and walking ability haseen performed by using different dual-task paradigms.4,5 Us-ng such paradigms, it has been well established that walkingequires attention6 and that this attention level is dependent onhysical factors such as gait speed7 and personal factors suchs age.5 Dual-tasking deficits after certain impairments havelso been studied although the majority of existing work hasoncentrated on cognitive impairments caused by dementias orotor impairments (eg, Parkinson’s disease).4

Despite sequelae in both cognitive and physical functionsfter TBI, the study of dual tasking in this population is onlyery recent. Studying persons with trauma-related concussion,arker et al8 showed acute effects related to shorter stride

engths when attention was divided by using verbal fluency orathematic tasks. The same general protocol was repeated

ver 4 periods, from 48 hours to 28 days, after a concussion1

howing deficits in dynamic equilibrium and dual-task effectsersisting up to 1 month after the concussion. Finally, Catenat al9 confirmed that dual tasking, using a “question and an-wers” task during gait, distinguishes concussed subjects fromontrol subjects.

To our knowledge, only 1 other study has looked at moreevere cases of TBI by using a dual-task paradigm. Vallée etl10 have shown that persons with good recovery from moder-te to severe TBI had residual deficits in both walking andognitive performance that was most evident when in moreomplex environments related to avoiding wider obstacles and

List of Abbreviations

TBI traumatic brain injury

pttcseasst

uflcsmspmdoipotilcmt

wavadersitlatcsm

sfnvVecwdmdd

wipwa

csdpraTtctbss

P

epIasglncgjbnmobjnsotRFwocvwip

I

cnhu1rrtatrva

1597MULTITASK DEFICITS AFTER TBI, McFadyen

erforming the Stroop word task. In a related publication fromhe same study, Cantin et al11 also provided the first evidencehat neuropsychologic tests of visuospatial deficits after TBIorrelate to obstacle clearance margins during adapted gaithowing some predictive ability of neuropsychologic tests forxecutive dysfunction during locomotion. One other study,12

lthough not looking directly at dual or multitasking, alsohowed greater variability in spatiotemporal gait parameters inubjects with TBI with increased complexity of the locomotorask (ie, walking fast or with eyes closed).

Many of the studies on dual tasking during walking havesed a wide range of simultaneous tasks including verbaluency, arithmetic calculations, other motor tasks such asarrying other objects, and reaction times to auditory and visualtimuli. The latter 2 tasks correspond to most daily environ-ents in which the processing of many visual and auditory

timuli is required while mobile. Most environments, whetherublic (eg, on the street) or private (eg, in the home), offerany visual stimuli that can be either related (eg, signs for

irection) or unrelated (eg, other people, objects, and so on) tone’s mobility. Locomotion itself relies heavily on vision forts control even in direct straight-ahead walking.13,14 When thehysical environment is more complex, such as in the presencef obstacles, visual control is important for safe lower-limbrajectories.15 It is now known that visual sampling of obstacless performed 2 to 3 steps from the obstruction16 and that leadimb clearance (first limb to cross an obstacle) involves visualontrol, whereas trail limb clearance (second limb to cross) isore reliant on feedforward information and kinesthetic con-

rol.15

Auditory information, such as engaging in a conversationith another person or listening to a public announcement, is

lso common during locomotion and has been shown to in-olve dual-task effects during walking. For example, Lajoie etl6 showed decreased reaction times to an auditory signaluring walking as compared with standing or sitting. Theyven found that attention was phase dependent with greatereaction times when the auditory signal was presented in thewing phase. Gérin-Lajoie et al17 have shown that circumvent-ng an obstacle is affected by a simultaneous auditory task suchhat young adults provide more clearance of the obstacle whenistening to the auditory message. These dual-task effects werelso shown to be accentuated with age.18 Anecdotal informa-ion from clinicians often refers to the levels of distractionsaused by auditory stimuli while working with patients havinguffered a TBI, but no study to date has looked specifically atodality issues.Overall, little is known about the differential effects of the

ensory modality of simultaneous tasks during walking. Of theew studies that have looked at simultaneous motor and cog-itive tasks involving visual or auditory modalities, most in-olve standing postures. Recently, Woollacott and Vanderelde19 have shown in healthy, young adults that dual-task

ffects attributed to the visual system may be caused by visualode processing rather than simple visual interference. Yet, itould appear that with impaired postural control that anyivision of attention (whether from a visual or an auditoryodality) will result in compromised balance.20 No study to

ate has directly considered auditory versus visual dual taskinguring obstacle avoidance.Understanding locomotor mobility in complex environments

ith different sensory stimuli is important to improve rehabil-tation interventions aimed at recovering function and socialarticipation for persons with a TBI. The purpose of the presentork was to build on our previous research3,10 studying the

bility of persons with TBI to avoid obstacles and to specifi- p

ally address, for the first time, the different influences ofimultaneous tasks involving visual or auditory sensory mo-alities. As in our previous articles,3,10 we have focused onersons who had recovered their locomotor skills and wereelatively independent and mobile in order to highlight residualnd persisting impairments salient to moderate and severeBIs. It was expected that despite good recovery, when simul-

aneous visual or auditory tasks are presented during the exe-ution rather than the planning phase for obstacle avoidance,hat subjects with TBI would persist in showing cautious gaitehavior (slower walking and higher obstacle clearances) andhow greater multiple task effects for visual interference buttill general attention deficits regardless of sensory modality.

METHODS

articipantsA convenience sample of persons who had a TBI and were

ither receiving or had received treatment from the multidisci-linary team at the TBI unit of the Quebec Rehabilitationnstitute was recruited. These subjects had suffered only 1 TBInd had severity ratings after their accident of moderate orevere based on a combination of the hospital admission Glas-ow Coma Scale score, duration of posttraumatic amnesia,ength of the loss of consciousness, and interpretation of theeuroradiologic examination.21 In addition, subjects had to beonsidered to be independent for walking and to walk at speedsreater than 1.0m/s without assistance or technical aids. Sub-ects with skull fractures or open head injuries, cognitive orehavioral problems affecting their participation, or any othereurologic or musculoskeletal problems affecting their loco-otion were excluded from the study. A convenience sample

f control subjects was also recruited from the Quebec Reha-ilitation Institute and university communities. These sub-ects were required to have no self-reported physical oreurologic problems and were matched on average for age andex to the TBI group. Finally, all subjects had to show normalr corrected-to-normal visual acuity as measured on a Snellenest, normal hearing as verified by an audiologist at the Quebecehabilitation Institute, and ability to understand and readrench. The latter was confirmed by the tests in audiology asell as by their understanding of instructions and performancen cognitive tests before data collection. Subjects also indi-ated that they could differentiate the 3 colors used for theisual task (see below) before testing commenced. The studyas approved by the ethics committee of the Quebec Rehabil-

tation Institute, and all subjects signed a consent form beforearticipating.

nstrumentationSubjects wore their own comfortable walking shoes and

lothes (loose pants were allowed) for the study. Triads ofoncollinear infrared markers were placed on each subject’sead, trunk, and each foot. Kinematic data were collected bysing 3 Optotrak sensor bars (model 3020a) at a frequency of00Hz. Toe points were statically digitized before collection toeconstruct their trajectories from foot markers later. Verbalesponses to the cognitive task (described later) were recordedhrough a microphone worn by the subject. Voice signals weremplified by using a channel mixer (MDR 6b) and then cap-ured on both a computer (1000Hz) and as part of a videoecording for each trial. A separate computer provided theisual and auditory stimuli to the subject and to the data-cquisition computer simultaneously. The visual stimuli were

resented to the subject simultaneously on five 43.2-cm (17in)

Arch Phys Med Rehabil Vol 90, September 2009

flt1amvstwpwcls

P

eantAtwi1

lsmrhapetqocasTai

cwotmamotsi

owcwwdcpbp

D

wafvrlmoaop(s(hd

mcgWg

Fpat

1598 MULTITASK DEFICITS AFTER TBI, McFadyen

A

at screen computer monitors (Compaq 1720c), 2 pairs forminghe sides of the walking path (approximately 1.6-m wide and1-m long in total) and 1 monitor at the end (fig 1). Thisrrangement allowed subjects to get visual information withoutoving their head too much from their desired focal point. A

ideo splitter (ST128PRO 8d) allowed the maintenance ofignal quality across the monitors. A 6th screen placed out ofhe sightline of the subject had a photo sensor attached to it andas used to acquire the exact timing of the visual signalresentation. The auditory signal was sent to both a set ofireless headphones worn by the subject and to the acquisition

omputer. The volume of the headphones was fixed at the sameevel (80db) for all subjects by using a calibration tone and aound-level meter (model 33-2055e).

rotocolAll subjects were evaluated by using separate clinical tests of

xecutive function and attention as well as of balance and gaitbility in addition to the laboratory evaluation. The significanteuropsychologic results of the Delis-Kaplan Executive Func-ion System Trail Making Tests and of the Test of Everydayttention will be presented along with the laboratory results. In

he laboratory, subjects were required to walk over the 11-malkway at natural speeds with and without an obstacle placed

n the middle of the walkway. The obstacle, when present, was

ig 1. Experimental setup showing the walking path, obstaclelacement, force plate used to trigger both cognitive tasks (arrow),nd orientation of computer monitors presenting the visual Stroopask.

22-cm wide and set in height and depth (length) to 15% of the d

rch Phys Med Rehabil Vol 90, September 2009

eg length of the subject and was placed approximately 5 to 6teps from the starting position. The obstacle was made of aetal frame with a vinyl roller window shade that could be

olled out over the frame to form a box shape (see fig 1). Weave used normalized obstacle heights in our studies3,10 tollow us to compare similar tasks across subjects. Fifteenercent of leg length is similar to a street curb and known tolicit anticipatory adjustments.3 Simultaneous visual or audi-ory stimuli were presented as described previously at a fre-uency corresponding to their average unobstructed stride timen half the trials. These stimuli were initiated at right heelontact in the middle area of the walkway by using a force platend a force trigger threshold of 15N. This position corre-ponded to 2 steps before the obstacle when it was present.herefore, during obstacle avoidance, the dual task was initi-ted at lead (first limb to clear the obstacle) foot contact beforets clearance followed by trail limb clearance.

The visual stimulus was adapted from the Stroop Word testommonly used in neuropsychology,22 presenting a singleord on the monitors that indicated 1 of 3 colors in the samer different color as their lexical meaning. Subjects were askedo name the color of the ink of the word and to ignore theeaning of the word. This test was chosen because it providesvisual task not requiring memory. The auditory task was aodified version of the Stroop test in which the words “man”

r “woman” were pronounced by either a man or womanhrough the earphones requiring subjects to name the speaker’sex and not the word heard. Therefore, again, subjects had tognore the lexical meaning.

All subjects were allowed to practice the Stroop tasks aheadf time and to perform each walking condition (with andithout obstacle, with and without dual task) before data

ollection began. All subjects began with the unobstructedalking conditions to ensure the attainment of comfortablealking speed. For the subsequent walking trials, obstacle andual-task conditions were presented in blocks of 5 trials andounterbalanced across subjects. We informed subjects of theresence and modality (visual or auditory) of the stimulusefore each trial and instructed them to respond as fast asossible when it was present.

ata AnalysesKinematic data were filtered by using a second-order Butter-

orth, zero lag filter with a cutoff frequency of 6Hz. Anverage of each dependent variable was calculated across trialsor each subject to be used for statistical analyses. Dependentariables analyzed included performance on the cognitive taskelated to response reaction times to the Stroop stimuli (calcu-ated difference between first stimulus and response onsets) andultitask costs comparing response reaction times between

bstructed and unobstructed walking (therefore, between dualnd triple tasking, the latter involving an adjustment for thebstacle at the same time as performing the cognitive task) anderformance on the locomotor task including walking speedaverage speed of the body center of mass over lead and trailtrides with and without obstacle) and foot clearance margindistance of the foot above the obstacle normalized to obstacleeight). In addition, intrasubject variability using the standardeviation values across trials for a subject was analyzed.Nonparametric statistical tests were used because they areore conservative than their parametric counterparts and be-

ause of the limited homogeneity of the data among the TBIroup. Differences between groups were tested with the Mann-hitney U test. For comparison across conditions within

roups, a Friedman test was first applied to test for any general

ifferences across conditions (except for the multitask costs in

wcf0eeAtw

cwoccPfas

fchaEwawrTcfptcoso

ttl

aTdasa(aPioscictgt(ano

A*†

‡

§

N

1599MULTITASK DEFICITS AFTER TBI, McFadyen

hich only 2 comparisons were necessary). When significantondition effects were detected, a Wilcoxon test was used forurther pair-wise comparisons. We set significance levels to.05. Although a number of comparisons were made in thisxploratory study, any corrections used to minimize type Irrors would have also increased the risk of type II errors.ccepting both arguments and maintaining our previous prac-

ice,3,10 all alpha values of significant comparisons (ie, P�.05)ere disclosed for the reader.

RESULTSFinal recruitment resulted in 11 persons with TBI and 7

ontrol subjects for group comparisons. All subjects werehite and French speaking. Table 1 shows the characteristicsf the TBI subjects along with the group medians for theontrol group. There were no differences between the TBI andontrol groups for average age or body mass index (both with�.05). The control group with 3 women did have a different

emale:male ratio (3:4 vs 2:9), but this was the only difference,nd, to our knowledge, sex differences have not yet beenhown for obstacle avoidance to date.

The subjects with TBI were found to have executive dys-unction and visuospatial attention deficits as compared withontrol subjects. Specifically, the TBI group had significantlyigher scores on visuospatial Trail Making Tests 1, 3, 4, and 5nd on the telephone search component (item 6) of the Test ofveryday Attention (P�.05) (table 2). The laboratory protocolas designed to provide each stimulus at a frequency equal to

pproximately the average stride time during unobstructedalking. Therefore, the stimuli should be triggered around each

ight heel contact. It was found post hoc that this was the case.he first stimulus (stimuli 1) was presented just after heelontact (approximately at 35ms for the visual stimuli and 54msor the auditory stimuli). The second stimulus (stimuli 2)resentation was always around second right heel contact al-hough slightly more variable across conditions and trials be-ause of the inherent variability in gait speeds. The responsensets for stimulus 1 were always in the early part of the lefttance phase around right toe-off (ie, at the beginning of lead

Table 1: Characteristics of the TBI Subjects and

N Sex Age (y) BMI G

1 M 17.03 20.72 M 38.09 19.53 M 17.41 22.14 F 56.44 21.85 M 26.8 19.46 M 30.34 25.727 F 42.91 31.028 F 33.27 18.949 M 52.87 19.38

10 M 47.13 29.0711 M 50.83 24.5

Median 38.09 21.78(IQR) (20.41) (5.67)CNTL

Median 40.21 24.11(IQR) (21.55) (3.48)

bbreviations: BMI, Body mass index; CNTL, Control; F, female; IQRLowest Glasgow Coma Scale at accident site or hospital.Number of days of posttraumatic amnesia.Number of months since injury.Clinical measure of gait speed over 10m.

bstacle clearance). Response onsets for stimulus 2 were, likeAm

his stimulus’ presentation, a bit more variable being either athe end of left swing after trail obstacle clearance or in earlyeft stance with no obstacle present.

There were no differences in gait speed between groups forny condition (fig 2A, B), and, if anything, the subjects withBI tended to walk faster. Only the TBI group showed generalifferences across conditions (�2�29.338, P�.001). Furthernalyses showed that these subjects slowed their gait speedlightly from the single walking task to the visual (P�.012)nd auditory (P�.001) dual tasks for unobstructed walkingsee fig 2A) and slowed from unobstructed walking to obstaclevoidance for each respective condition (no simultaneous task,�.001; visual task, P�.012; auditory task, P�.007). For

ntrasubject variability in gait speed (fig 2C, D), there wasnly 1 group difference during unobstructed walking with theimultaneous visual task (P�0.044) because of a decrease byontrol subjects for this condition from the unobstructed walk-ng (see fig 2C). There was also a general difference acrossonditions for the TBI group (�2�20.974, P�.001) but not forhe control group. Specifically, variability in speed for the TBIroup was significantly decreased for obstacle avoidance forhe conditions without stimulus (P�.021) and with visualP�.003) and auditory stimuli (P�.027). Subjects with TBIlso showed a significant decrease between the no simulta-eous task and the simultaneous visual task but only duringbstructed walking (P�.042).

ians and Interquartile Ranges for Both Groups

/15) PTA† (d) Time‡ (mo) Gait Speed (m/s)§

37 6.02 1.5821 2.66 1.4925 2.83 1.17

7 4.27 1.5820 1.45 1.5513 3.19 1.6477 124.83 1.4016 2.1 1.47

7 2.1 1.5215 1.22 1.41

4 24.16 1.640 16 2.83 1.52) 13 (3.04) (0.14)

NA NA NA

erquartile range; M, male; NA, not applicable.

Table 2: Medians, Interquartile Ranges, and P Values BetweenNeuropsychologic Test Scores of the TBI and Control Groups

TestsTBI

Scores Control Scores P Value

TEA 6 3.9 (1.31) 2.6 (0.37) .01TMT 1 21 (6.5) 12 (3.0) .04TMT 3 34 (16.5) 23 (4.0) .04TMT 4 76 (24.5) 54 (20.5) .04TMT 5 26 (5.5) 20 (5.0) .04

OTE. Interquartile ranges are in parentheses.

Med

CS* (

533

1046436

1365.

(2.5

NA

, int

bbreviations: TEA, Test of everyday attention; TMT, D-Kefs trailaking test.

Arch Phys Med Rehabil Vol 90, September 2009

aa3tsts3p(tabd

agte

srpbadasf

Fiapt n witA g; O,

1600 MULTITASK DEFICITS AFTER TBI, McFadyen

A

Lead toe clearance margins over the obstacle were similarcross groups for all conditions but clearly more variablecross the TBI group (fig 3A). Trail limb clearance margins (figB), however, were lower for the TBI group as compared withhe control group across walking conditions although the onlyignificant individual condition was for the simultaneous audi-ory task (P�.044). Also, only the subjects with TBI showed aignificant difference across all clearance margins (�2�1.727, P�.001) associated with lower trail clearances as com-ared with lead clearances for the no simultaneous taskP�.01), the simultaneous visual task (P�.005), and the simul-aneous auditory task (P�.001). Intrasubject variability of clear-nce margins (fig 3C, D) tended to be greater for the TBI group,ut this was only significantly so (P�.035) for the lead limb

ig 2. Walking speeds (A, B) and intrasubject variability in walkinndicated for subjects with TBI (dark boxes) and control subjects (pauditory Stroop tasks. The box plots indicate medians (thick horizonercentile range, whereas I-bars indicate full ranges. Outliers (emptyhe interquartile range. Significant differences (P<.05) are showbbreviations: _A, auditory Stroop tasks; NO, unobstructed walkin

uring the simultaneous visual task. There were no differences (

rch Phys Med Rehabil Vol 90, September 2009

cross conditions for the TBI group. The control group did showeneral differences across conditions (�2 � 11.49, P�.033) al-hough further pair-wise comparisons showed no specific differ-nces between any 2 conditions.

With respect to the absolute response reaction times to theimultaneous Stroop tasks, subjects with TBI showed longeresponse reaction times (P�.017) over all conditions as com-ared with the control group for the first stimulus (fig 4A, B),ut there were no significant differences between groups forny specific condition. For stimulus 2 (fig 4C, D), significantifferences were found between groups for the simultaneousuditory stimulus for both unobstructed (P�.035) and ob-tructed walking (P�.044). There was a general differenceound between all conditions and stimuli for both the TBI

eds (C, D) during unobstructed and obstructed walking. Data arexes) with no division of attention and with simultaneous visual andrs), the 75th (top of box) percentile range, and 25th (bottom of box)

es) represent values that were at least 1.5 times greater or less thanhin groups (thin solid line) and between groups (dashed line).obstructed walking; _V, visual Stroop tasks.

g spele botal bacircl

�2�11.782, P�.007) and the control groups (�2�25.543,

Pisnnsgwsaowmw

o

aTaasiv

cccetvf

Fud(rg ons:w

1601MULTITASK DEFICITS AFTER TBI, McFadyen

�.001). Specifically, subjects with TBI showed a significantncrease in response reaction time from unobstructed to ob-tructed walking for both stimuli 1 and 2 during the simulta-eous visual task (P�.01 and P�.005, respectively). This wasot the case for the auditory task for the TBI group or for eitherimultaneous task for the control group. In addition, for the TBIroup, the response reaction time for the first visual stimulusas longer than for the second visual stimulus for both unob-

tructed (P�.003) and obstructed (P�.001) walking. This waslso observed for subjects with TBI for the auditory task butnly for the obstructed condition (P�.005). No differencesere found for the control group across stimuli of the sameodality. Comparisons across modalities are presented laterith respect to the multitask cost variable.For intrasubject variability in response reaction times (fig 5),

ig 3. Clearance margins for lead (A) and trail (B) limbs and intrasnobstructed and obstructed walking. Data are indicated for subjecivision of attention and with simultaneous visual and auditory Strotop of box) percentile range, and the 25th (bottom of box) percenepresent values that were at least 1.5 times greater or less than troups (thin solid line) and between groups (dashed line). Abbreviatialking; _V, visual Stroop tasks.

nly obstructed walking, in particular for the simultaneous b

uditory condition (P�.033), showed greater variability in theBI group. Overall, significant differences across conditionsnd stimuli were found for both the TBI (�2�11.891, P�.007)nd control (�2�25.543, P�.001) groups, but further analyseshowed only 1 significant difference for the TBI group whoncreased (P�.001) their response reaction time variability forisual stimulus 1 from unobstructed to obstructed walking.A multitask cost was created to look specifically at the

hange in response reaction times caused by environmentalontext (ie, obstructed vs unobstructed walking) and specifi-ally to compare between sensory modalities. Group differ-nces as discussed previously for absolute response reactionimes showed the same tendency for multitask costs for theisual task (fig 6A, B) although it was not significant. It wasound that only subjects with TBI had significant differences

t variability in clearance margins for the same limbs (C, D) duringith TBI (dark boxes) and control subjects (pale boxes) without anysks. The box plots indicate medians (thick horizontal bars), the 75thange, whereas I-bars indicate full ranges. Outliers (empty circles)

terquartile range. Significant differences (P<.05) are shown within_A, auditory Stroop tasks; NO, unobstructed walking; O, obstructed

ubjects w

op tatile r

he in

etween the visual and auditory modalities, showing a gener-

Arch Phys Med Rehabil Vol 90, September 2009

atudjaoasgsdhosw

qosjeYiddtdm

Fw(rd eenu ks.

1602 MULTITASK DEFICITS AFTER TBI, McFadyen

A

lly increased attention cost for the simultaneous visual overhe auditory task (P�.002) for both stimuli. When each stim-lus was analyzed, only stimulus 2 maintained a significantifference between sensory modalities (P�.014). Control sub-ects maintained the same multitask cost for both visual anduditory tasks. It must be recalled from earlier, however, thatverall response times were still shorter for this control groups compared with the TBI group. When looking at the intra-ubject variability for multitask costs (fig 6C, D), significantroup differences (P�.011) were maintained for the first visualtimulus. Also, again, only subjects with TBI had significantifferences between sensory modalities showing a generallyigher variability in multitask costs for the simultaneous visualver the auditory task (P�.036) for both stimuli. No significanteparate pair-wise differences were found when each stimulus

ig 4. Response reaction times for stimulus 1 (A, B) and 2 (C, D) duriith TBI (dark boxes) and control subjects (pale boxes) with simulta

thick horizontal bars), the 75th (top of box) percentile range, andanges. Outliers (empty circles) represent values that were at leifferences (P<.05) are shown within groups (thin solid line) and betwnobstructed walking; O, obstructed walking; _V, visual Stroop tas

as analyzed separately. f

rch Phys Med Rehabil Vol 90, September 2009

DISCUSSIONMobility within the complex environments found daily re-

uires combined physical and cognitive abilities. Based on thebservation of normal walking speeds and relatively normalpatial-temporal performance, the locomotor skill of the sub-ects with TBI in the present study appeared to be well recov-red, and these subjects did not show general cautiousness.et, despite their good walking ability, processing of sensory

nformation was longer than normal, and residual attentioneficits for multitasking behavior remained. With respect to theifferent sensory modalities of the 2 simultaneous tasks, al-hough the TBI group showed general information processingeficits in these complex locomotor environments regardless ofodality, the presentation of simultaneous visual stimuli was

obstructed and obstructed walking. Data are indicated for subjectss visual and auditory Stroop tasks. The box plots indicate medians5th (bottom of box) percentile range, whereas I-bars indicate full

.5 times greater or less than the interquartile range. Significantgroups (dashed line). Abbreviations: _A, auditory Stroop tasks; NO,

ng unneouthe 2

ast 1

ound to elicit the greater multitask costs and differences be-

tt

fYfblcfias

dt

gvaltssNp

Fabwr solidS , visu

1603MULTITASK DEFICITS AFTER TBI, McFadyen

ween physical environments (unobstructed vs obstructed) overhe auditory stimulus.

Previous work3,10 had suggested that persons with TBI per-orm gait adjustments more cautiously than control subjects.et, the TBI group studied here did not show such caution. In

act, the subjects of this study showed, if anything, a riskierehavior by exhibiting lower clearance margins for the trailingimb across conditions. The trailing limb is considered to beontrolled without direct visual input involving instead feed-orward control and more reliance on proprioception.15 Look-ng at standing postural control, Redfern et al23 have proposedreweighting of information processing resources for sensory

ig 5. Intrasubject variability in response reaction times for stimulure indicated for subjects with TBI (dark boxes) and control subjectox plots indicate medians (thick horizontal bars), the 75th (top ohereas I-bars indicate full ranges. Outliers (empty circles) represen

ange. Significant differences (P<.05) are shown within groups (thintroop tasks; NO, unobstructed walking; O, obstructed walking; _V

ignals when attention is divided. However, we did not observe d

eterioration in trail limb clearance behavior with either simul-aneous visual or auditory tasks.

What was observed here was variability across the TBIroup for both limbs, whereas the control subjects were onlyariable for trail clearance and fairly consistent for lead clear-nce. With this in mind, we looked post hoc at general corre-ations of changes in lead clearance margins with those of therail limb over all conditions and found that the TBI grouphowed a significant positive correlation (R�.56, P�.001)uch that the lead and trail clearance margins changed together.o correlation was found in the control group supportingrevious findings for healthy adults24 as well as the previously

, B) and 2 (C, D) during unobstructed and obstructed walking. Datale boxes) with simultaneous visual and auditory Stroop tasks. The) percentile range, and the 25th (bottom of box) percentile range,es that were at least 1.5 times greater or less than the interquartileline) and between groups (dashed line). Abbreviations: _A, auditoryal Stroop tasks.

s 1 (As (paf boxt valu

iscussed independence of control of these limbs.15 The cor-

Arch Phys Med Rehabil Vol 90, September 2009

rsacesttlpatnsl

chrtaT

urtcobv

Fc7r he ing tions

1604 MULTITASK DEFICITS AFTER TBI, McFadyen

A

elated clearance margins between lead and trail limbs forubjects with TBI indicate interlimb coupling during obstaclevoidance and, therefore, a reduced ability to independentlyontrol the 2 limbs perhaps because of an inability to attend toach limb separately. In the previous studies3,10 in whichubjects with TBI produced higher lead clearances, perhapshis coupling was the reason for the higher trail limb clearancesoo because they would have been raised in consequence withead limb control. In the present study, the TBI group may havelanned lead clearance normally on average but failed to makeppropriate and independent adjustments for the nonvisiblerailing limb. This puts the subjects potentially at risk althougho subject ever hit the obstruction in the present study. A fewubjects with TBI did, however, show very low absolute trail

ig 6. Multitask costs (A, B) and intrasubject variability (C, D) for stontrol subjects (pale boxes) with simultaneous visual and auditory5th (top of box) percentile range, and 25th (bottom of box) perceepresent values that were at least 1.5 times greater or less than troups (thin solid line) and between groups (dashed line). Abbrevia

imb clearances of around 3 cm only (ie, corresponding to a c

rch Phys Med Rehabil Vol 90, September 2009

learance margin of approximately 0.03 or 3% of obstacleeight). Although such behavior could be seen as potentiallyisky, given that no one actually failed the task, it could also behe result of abnormally greater attention given to the avoid-nce task. Either way, multitasking behavior is altered in theBI group, despite good locomotor recovery.As for the difference between visual and auditory modalities

sed for the simultaneous stimuli, it is clear that the visual taskesulted in more obvious differences between obstacle condi-ions. Both stimuli were specifically triggered at lead heelontact corresponding to the final preparation for and executionf clearance when the obstacle was present. This period haseen shown to involve less visual scanning and, therefore,isual interference during obstacle avoidance.16 The data of the

1 and 2. Data are indicated for subjects with TBI (dark boxes) andp tasks. The box plots indicate medians (thick horizontal bars), the

range, whereas I-bars indicate full ranges. Outliers (empty circles)terquartile range. Significant differences (P<.05) are shown within: _A, auditory Stroop tasks; _V, visual Stroop tasks.

imuliStroontile

ontrol subjects, particularly the similar multitask costs across

mefttregtti

iaapsdrSA(frctamas

TmssuushtmsspasFo(vaaa

spcitasctgci

sruenaijgsspar

babapnaut

S

knsTaAnmcbaeseti

pnmrogivciaed

htp

1605MULTITASK DEFICITS AFTER TBI, McFadyen

odalities, generally support this idea. For the TBI group, how-ver, the increase in response reaction times and its variabilityrom unobstructed to obstructed walking for the visual task andhe multitask cost results showing greater costs for the visualask when clearing the obstacle suggest that, despite goodecovery, these subjects were more affected by visual interfer-nce. This may, in part, be because these subjects maintainreater visual attention for the obstacle during the late approacho and subsequent clearance of the obstacle. This behavior putshem more at risk for interference at a crucial time whennteraction with the environment is more precarious.

It is possible that such a deficit is caused by poorer process-ng abilities for spatial information. Previous work11 has showncorrelation between visuospatial processing ability and clear-

nce margins. When such correlations were performed for theresent data, it was found that there were significant relation-hips between changes in trail limb foot clearance marginifferences, both with and without a visual stimulus, and neu-opsychologic scores on the Delis-Kaplan Executive Functionystem Trail Making Test (subtest 4) and the Test of Everydayttention (subtest 7, telephone search while counting tones)

P�.05) for subjects with TBI. These relationships were notound in the control group. Therefore, it appears that there is aelationship between visual scanning ability as measured in thelinical setting and changes in foot clearance margin for therail limb when simultaneously performing another visual taskfter TBI. Obviously, problems with visual scanning abilityust be considered and put into perspective of the general

bility of persons with TBI to perform different daily life habitsuch as grocery shopping or taking public transport.

The auditory tasks were not without effects, however. TheBI group took longer on average to process auditory infor-ation as compared with the control group (significantly so for

timulus 2). Also, when avoiding the obstacle, the TBI grouphowed greater variability in response reaction time for stim-lus 1 and greater effects for the first versus the second stim-lus. Previous studies17 that have shown effects of auditorytimuli have often involved some working memory as well. Itas been shown that the capacity for attention may be restrictedo within the modality and not between them.25 However, aore recent study by Woollacott and Vander Velde19 has

hown that the actual interference that is important may be thepatial coding and not simply the visual modality. In theresent study, neither task had a strong spatial componentlthough the visual task was presented on several screensimultaneously, which may affect spatial coding to an extent.urther work will need to also consider time since injury, levelf locomotor recovery, types of tasks, and level of processingie, spatial vs nonspatial). Yet, the present results imply thatisual interference is perhaps more of an issue of concerncross environments even after very good locomotor recoverynd warrants further consideration in the context of mobilityssessment and rehabilitation interventions.

The differences between the effects of the first and secondtimuli are also interesting and warrant further discussion. Therotocol used here presented each stimulus around right heelontact for all conditions. Yet, when the obstacle was present,t turned out that the processing of the first stimulus was duringhe placement of the trail foot and preparation for lead clear-nce with the related response provided at the beginning of leadwing. For the second stimulus, processing was during traillearance and preparation for subsequent foot contact, whereashe related response was during re-establishment of steady stateait after obstacle avoidance. Thus, the 2 stimuli were pro-essed during different avoidance processes. Group differences

n multitask costs variability were only observed for the first G

timulus. In addition, the TBI group showed greater responseeaction times for the first as compared with the second stim-lus, particularly for obstructed walking. Given the discussionarlier about lead/trail coupling, second stimulus effects mayot have been as perturbing because they were at the same times trail limb clearance, which may not have been attended tondependently. Another explanation is that even though sub-ects were told when simultaneous stimuli would be present,reater first stimulus effects for the TBI group may indicate thelower initial processing time at first presentation. Overall, ifuch techniques of multitasking are to be used for clinicalurposes, it will be important to verify this first stimulus effectnd trail limb attention, particularly after TBI with poorerecovery.

It must be noted that real-life environments would probablye even more demanding, and attention would not be as primeds we see in the laboratory. This could result in greater effectsy an auditory stimulus than observed here. Although we havettempted here to use tasks with some ecologic validity (step-ing over obstacles with divided attention), the context andecessary laboratory control are still limiting. In addition,lthough there is no clear indication in the literature to informs about sex effects on such locomotor behavior, this may haveo be addressed in the future.

tudy LimitationsSeveral limitations of our descriptive study should be ac-

nowledged. The sample size, although relatively homoge-eous in walking ability within each group, was relativelymall. Although this is often the case in studies of gait afterBI,1-3,8,10,11 the work should be repeated with larger samplesnd across a larger range of severities and walking abilities.lso, because of the exploratory nature of the study and theumber of physical and cognitive factors to consider affectingobility, a number of comparisons were made. There is still

ontroversy as to whether corrections for multiple tests shoulde performed or not.26 We opted to simply present all P valuesnd not correct the data. Yet, the possible inflation of type Irrors should be considered when interpreting the reportedignificant findings of the present study. We do believe, how-ver, that the present observations provide important informa-ion for consideration in future studies and perhaps even clin-cal practice.

CONCLUSIONSMobility is a complex construct involving the integration of

hysical and cognitive abilities. Our study provided prelimi-ary evidence that even with good locomotor recovery after aoderate to severe TBI, residual locomotor deficits appear to

emain in relation to multitasking. In particular, the presentbservations suggest that the greatest multitask effects duringait may occur when there is sensory competition (ie, visualnterference). Further work should be performed over a widerariety of TBI severity and locomotor capacity as well asonsidering other confounding factors related to TBI (spastic-ty, comorbidity, age, and so on). Finally, low trail limb clear-nce margins by the subjects with TBI require further consid-ration and may suggest a strategy to couple the limbs andecrease attention resources.

Acknowledgments: We thank Guy St-Vincent, MSc, PEng, foris technical assistance; Eric Huard, Aud, and Janick Bisson, SLP, forheir consultation and the development of the Stroop presentationrogram; Mireille Beaudoin, SLP, for auditory testing; and Lucie

agnon, PT, for aid in the analyses of physiotherapy assessments.

Arch Phys Med Rehabil Vol 90, September 2009

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

a

b

cd

1606 MULTITASK DEFICITS AFTER TBI, McFadyen

A

References1. Parker TM, Osternig LR, Van Donkelaar P, Chou LS. Gait

stability following concussion. Med Sci Sports Exerc 2006;38:1032-40.

2. Ochi F, Esquenazi A, Hirai B, Talaty M. Temporal-spatial featureof gait after traumatic brain injury. J Head Trauma Rehabil 1999;14:105-15.

3. McFadyen BJ, Swaine B, Dumas D, Durand A. Residual effects ofa traumatic brain injury on locomotor capacity: afirst study ofspatiotemporal patterns during unobstructed and obstructed walk-ing. J Head Trauma Rehabil 2003;18:512-25.

4. Yogev-Seligmann G, Hausdorff JM, Giladi N. The role of exec-utive function and attention in gait. Mov Disord 2008;23:329-42.

5. Woollacott M, Shumway-Cook A. Attention and the control ofposture and gait: a review of an emerging area of research. GaitPosture 2002;16:1-14.

6. Lajoie Y, Teasdale N, Bard C, Fleury M. Attentional demands forstatic and dynamic equilibrium. Exp Brain Res 1993;97:139-44.

7. Kurosawa K. Effects of various walking speeds on probereaction time during treadmill walking. Percept Motor Skills1994;78:768-70.

8. Parker TM, Osternig LR, Lee HJ, Donkelaar P, Chou LS. Theeffect of divided attention on gait stability following concussion.Clin Biomech (Bristol, Avon) 2005;20:389-95.

9. Catena RD, van Donkelaar P, Chou LS. Cognitive task effectson gait stability following concussion. Exp Brain Res 2007;176:23-31.

0. Vallée M, McFadyen BJ, Swaine B, Doyon J, Cantin JF, DumasD. Effects of environmental demands on locomotion after trau-matic brain injury. Arch Phys Med Rehabil 2006;87:806-13.

1. Cantin JF, McFadyen BJ, Doyon J, Swaine B, Dumas D, ValléeM. Can measures of cognitive function predict locomotor behav-iour in complex environments following a traumatic brain injury?Brain Inj 2007;21:327-34.

2. Niechwiej-Szwedo E, Inness EL, Howe JA, Jaglal S, McIlroy WE,Verrier MC. Changes in gait variability during different chal-lenges to mobility in patients with traumatic brain injury. GaitPosture 2007;25:70-7.

3. Bruggeman H, Zosh W, Warren WH. Optic flow drives humanvisuo-locomotor adaptation. Curr Biol 2007;17:2035-40.

4. Patla AE, Prentice SD, Robinson C, Neufeld J. Visual control oflocomotion: strategies for changing direction and for going overobstacles. J Exp Psychol Hum Percept Perform 1991;17:603-34.

5. Mohagheghi AA, Moraes R, Patla AE. The effects of distant and

on-line visual information on the control of approach phase and e

rch Phys Med Rehabil Vol 90, September 2009

step over an obstacle during locomotion. Exp Brain Res2004;155:459-68.

6. Patla AE, Vickers JN. Where and when do we look as weapproach and step over an obstacle in the travel path? Neuroreport1997;8:3661-5.

7. Gérin-Lajoie M, Richards CL, McFadyen BJ. The negotiation ofstationary and moving obstructions during walking: anticipatorylocomotor adaptations and preservation of personal space. MotorControl 2005;9:242-69.

8. Gérin-Lajoie M, Richards CL, McFadyen BJ. The circumventionof obstacles during walking in different environmental contexts: acomparison between older and yunger adults. Gait Posture 2006;24:364-9.

9. Woollacott M, Vander Velde T. Non-visual spatial tasks revealincreased interactions with stance postural control. Brain Res2008;208:95-102.

0. Shumway-Cook A, Woollacott M, Kerns KA, Baldwin M. Theeffects of two types of cognitive tasks on postural stability in olderadults with and without a history of falls. J Gerontol A Biol SciMed Sci 1997;52:M232-40.

1. Gadoury M. Cadre de référence clinique pour l’élaboration deprogrammes de réadaptation pour la clientèle ayant subi un trau-matisme cranio-cérébral (adulte) 2eme édition. Québec: Sociétéde l’assurance automobile du Québec; 2001.

2. Golden CJ. Identification of brain disorders by the Stroop Colorand Word Test. J Clin Psychol 1976;32:654-8.

3. Redfern MS, Muller ML, Jennings JR, Furman JM. Attentionaldynamics in postural control during perturbations in young andolder adults. J Gerontol A Biol Sci Med Sci 2002;57A:B298-303.

4. Patla AE, Rietdyk S, Martin C, Prentice S. Locomotor patterns ofthe leading and the trailing limbs as solid and fragile obstacles arestepped over: some insights into the role of vision during loco-motion. J Mot Behav 1996;28:35-47.

5. Rees G, Frith C, Lavie N. Processing of irrelevant visual motionduring performance of an auditory attention task. Neuropsycho-logia 2001;39:937-49.

6. Perneger TV. What is wrong with Bonferroni adjustments. BrMed J 1998;136:1236-8.

Suppliers. Northern Digital Inc, 103 Randall Dr, Waterloo, ON, CA, N2V

1C5.. Samson Technologies Corp, 575 Underhill Blvd, Syosset, NY

11791.. Hewlett-Packard USA, 20555 SH 249, Houston, TX 77070.. StarTech.com, 45 Artisans Cres., London, ON, CA, N5V 5E9.

. Radio Shack, 300 RadioShack Circle, Fort Worth, TX 76102-1964.