Brain Injury, August 2008; 22(9): 657–668

Mild traumatic brain injury induces prolonged visual processing

deficits in children

ODILE BROSSEAU-LACHAINE1,2, ISABELLE GAGNON3,4, ROBERT FORGET2,5,& JOCELYN FAUBERT1,2

1Ecole d’optometrie, Universite de Montreal, Montreal, Quebec, Canada, 2Centre de recherche interdisciplinaire en

readaptation du Montreal Metropolitain (CRIR), Montreal, Quebec, Canada, 3Trauma Programs, Montreal

Children’s Hospital, McGill University Health Center, Montreal, Quebec, Canada, 4School of Physical and

Occupational Therapy, McGill University, Montreal, Quebec, Canada, and 5Ecole de readaptation, Faculte de

Medecine, Universite de Montreal, Quebec, Canada

(Received 28 February 2008; accepted 14 May 2008)

AbstractPrimary objective: To compare the sensitivity to simple and complex visual stimuli of children who have sustained a mildtraumatic brain injury (mTBI) to that of matched non-injured children and to determine the evolution of visuo-perceptualperformance over time.Research design: A prospective design was used to assess 18 children with mTBI and 18 matched healthy controls(8–16 years of age).Methods and procedures: Sensitivity to static and dynamic forms of simple (first-order) and complex (second-order) stimuliwere assessed at 1, 4 and 12 weeks post-injury and at equivalent times for controls. Orientation and direction identificationthresholds were measured for all participants for static and dynamic conditions, respectively. In addition, sensitivity to radialoptic flow (inward vs outward), a complex motion stimulus, was assessed.Main outcomes and results: Thresholds measured from all complex stimuli were significantly affected for the mTBI childrenover time whereas no difference in threshold between groups across all testing conditions was found for simple, first-orderinformation. Sensitivity to all complex stimuli was still affected 12 weeks after the injury.Conclusion: These findings suggest that injured children present selective processing deficits for higher-order informationand that this deficit persists over relatively long periods. Such measures could be useful to assess children who havesustained mTBI and possibly contribute to identifying potential risks of returning these children to demanding physicalactivities.

Keywords: Visual perception, sensory integration, mild traumatic brain injury, paediatrics, second-order, optic flow

Introduction

A large portion of the human brain is involved invision and given the vulnerability of cerebellar andcortical tracts to shearing injury, a variety of visualimpairments have been reported following a trau-matic brain injury (TBI) [1–6]. The identifieddeficits are usually focused on oculomotor disorders,loss of acuity; loss of visual field; loss of binocularity;

cranial nerve palsies with extra-ocular motility, andfew authors have investigated the perception ofvisual stimuli or the integration of visual informationfor use in functional activities after TBI.

The perception of visual stimuli is generallyachieved through the attributes of the stimulithemselves. When objects are defined by attributessuch as luminance and colour, they are said to

Correspondence: Jocelyn Faubert, PhD, Visual Psychophysics and Perception Laboratory, Ecole d’optometrie, Universite de Montreal, 3744, Jean-Brillant,room 260-7, Montreal, Quebec, H3T 1P1, Canada. Tel: (514) 343-7289. Fax: (514) 343-2382. E-mail: jocelyn.faubert@umontreal.ca

ISSN 0269–9052 print/ISSN 1362–301X online � 2008 Informa Healthcare USA, Inc.DOI: 10.1080/02699050802203353

Downloaded By: [Canadian Research Knowledge Network] At: 18:35 28 October 2008

possess ‘first order’ (i.e. simple, linear) properties.On the other hand, when objects can only beperceived by their contrast, texture or depth, theyare said to have ‘second order’ (i.e. complex,non-linear) properties and additional informationprocessing is required to perceive the image [7, 8].Psychophysical and physiological evidence, alongwith computational modelling, suggest that twoseparate cortical mechanisms underlie the initialprocessing of first- and second-order stimuli, wherean additional processing step is needed for second-order stimuli [9–12].

There is little documentation of the specificimpact of mild TBI (mTBI) on the visual systemand particularly on the individual’s ability toperceive and integrate complex visual information.Indeed, studies investigating the consequences ofmTBI have mostly concentrated on investigatingcognitive functioning and generally report impairedinformation processing speed, working memory,attention, memory and executive functions[13–15]. More recently, balance deficits have beendocumented in mTBI children [16] and adults[17–21]. Examining the consequences of even amild brain injury on visual perception and processingis important because in order to behave and navigatethrough the environment in an efficient and adaptivemanner, one must first construct a visual representa-tion of one’s surroundings. To achieve this, one’svisual brain must integrate local information acrosstime and space into a coherent visual percept

Work conducted by this group [22, 23], investi-gated visuo-motor performances of children withMTBI and demonstrated that the injured childrenmanifested response-time deficits under certainexperimental conditions involving complex visuo-motor tasks (i.e. catching a moving stimulus), but nodeficits in reaction and movement time for classicalsimple or choice response time paradigms. Althoughthese studies were primarily aimed at identifyingresponse-time deficits, the question of whetherperceptual deficits of the visual system could beresponsible for those slower response times wasraised and it was questioned if the nature of thevisual stimulus itself (i.e. complex) could havecontributed to the difference in visuo-motor perfor-mance of children with and without mTBI. Othersused visual evoked potential (VEP) to investigatesimilar concepts in adults and have demonstratedthat higher-level VEPs (e.g. texture segregation,cognitive EPs), which permitted the detection ofsubtle deficits which remained indistinguishable withstandard electrophysiology or neuroimagery [24, 25]were sensitive to mild-to-moderate TBI andsuggested possible resulting visual information pro-cessing deficits after the injury. To the authors’knowledge, no study has investigated the perception

of complex visual stimuli in children who suffereda mTBI.

One way to investigate the capacity of individualsto visually integrate information is to use simpleorientation- and direction-identification paradigmsfor static and moving stimuli, respectively. One suchparadigm uses first- and second-order visual stimulithat differ in the amount of neural integrationneeded to perceive its direction or orientation;probing either early- or later-level visual corticalfunction [7, 9, 26, 27]. Measurements of thesensitivity to first- and second-order motion havebeen demonstrated to be very sensitive to subtleneural deficits in different populations such as non-pathological ageing [8], high-functioning autism[28, 29] and fragile X syndrome [30].

Another type of dynamic and complex visualinformation is referred to as optic flow motion.Optic flow is a type of complex motion informationthat has great ecological validity, since it exemplifiesthe visual pattern perceived as one navigates throughone’s environment [31]. Gibson [32] proposed thatthe direction of self-motion (as in locomotion) canbe directly perceived from the ‘focus of radialoutflow’ in the optical flow pattern. One paradigmto test optic flow is the presentation of radial motionstimuli (e.g. randomly generated white dots on ablack background) that travel towards (expandingfield) or away from (contracting field) the observer.At least one group has used the assessment of suchconcepts in adults with mTBI found destabilizingeffects of visual field motion (virtually moving theroom in different directions) for the duration of afollow-up study which lasted up to 30 days post-injury [33]. The assessment of the perception ofoptic flow information in children with mTBI cantherefore give important insight on their ability tointegrate and efficiently perceive their environmentas they navigate through it.

The present study assessed the perception ofsimple and complex visual stimuli after a mTBI inchildren, in the acute phase. The objectives were to(1) determine whether children who have sustained amTBI presented a decreased sensitivity for complexvisual stimuli when compared to matched non-injured children and (2) explore the evolution ofthe visual perceptual performance of children withand without mTBI during the first 3 monthsfollowing the injury in children with mTBI.

Methods

Participants

Eighteen children having sustained a mTBI and anequivalent number of children without any history ofbrain injury accepted to participate in this study.

658 O. Brosseau-Lachaine et al.

Downloaded By: [Canadian Research Knowledge Network] At: 18:35 28 October 2008

All participants were aged between 8–16 years.Children from the mTBI group were recruitedfrom consecutive hospital admissions to theTrauma and Injury Prevention Program from theMontreal Children’s Hospital (McGill UniversityHealth Center, Montreal, Canada) and to theEmergency Department of Sainte-Justine Hospital(Montreal University Hospital Center, Montreal,Canada). All injured participants were considerednormal on a standard neurological examination doneprior to discharge from the hospital. The inclusioncriterion for the injured group was a diagnosis ofmild TBI made by the professionals of the recruitinghospitals, as defined by the American Congress ofRehabilitation Medicine [34] and the WHOCollaborating Centre for Neurotrauma Task Forceon mTBI [35]. This includes at least one of thefollowing criteria after an acute brain injury resultingfrom mechanical energy to the head: confusion ordisorientation, loss of consciousness for 30 minutesor less, any loss of memory for events immediatelybefore or after the accident lasting less than 24 hoursand/or other transient neurological abnormalities notrequiring surgery. Also, a Glasgow Coma Scale(GCS) score of 13–15 within 30 minutes of theinjury [34, 35]. Participants received a complete eyeexam by an optometrist and had to have normal orcorrected-to-normal vision (acuity of 6/6 or better), anormal ocular health (pupillary reflexes, ocularmotilities, biomicroscopia and fundus) and binocu-lar vision (excluding heterophoria and tropia, normalstereoscopic vision). All participants had typicalacademic backgrounds (i.e. they attended age-appropriate classes, according to standard schoolschedules in Quebec) and development. They werescreened to exclude pre-morbid diagnosis of learningdisabilities, attention deficits and hyperactivity dis-order as well as regular use of psychostimulant drugsand/or behaviour problems. Children for the com-parison group were recruited among friends ofchildren with mTBI when possible or from thecommunity. The groups were matched as closely aspossible in terms of gender and chronological age.They had no history of any form of head injury. Themean chronological age of the children included inthe mTBI and control groups were 12.56 (SD 2.38)and 12.44 (SD 2.37), respectively (for children agedbetween 8–16 years).

Table I presents the characteristics of the braininjury sustained by this sample (mTBI group).Reported injuries were from accidental impacts tothe head (hits and/or falls) and during recreationalactivities, such as bicycling, playing hockey, basket-ball, soccer, football or skiing. The majority of thesample had a GCS score of 15 at the admission tothe hospital (average of 14.72). Eighty-three per centof children had sustained mTBI that could be

categorized as ‘simple concussion’, while 17% hadsustained ‘complex’ concussions, according to thisrecently introduced typology [36]. An informedwritten consent was obtained from all participantsand their legal guardians, prior to data collection.The ethics and scientific committees from theInstitutional Review boards of the MontrealChildren’s Hospital of the McGill UniversityHealth Centre and of Sainte-Justine Hospitalapproved the study.

Apparatus

The stimuli were presented and the data werecollected by a Power Macintosh G3 computer.Presentation was done on a 16-inch Apple monitorfor first- and second-order stimuli with a framerefresh rate of 75 Hz and a screen resolution of1024� 768 pixels. For optic flow stimuli, stimuliwere rear-projected (InFocus LCD projector,LP725) on a large light diffusing tangent screen(Da-lite; 1.06� 0.8 m) with a frame refresh rate of60 Hz and a screen resolution of 800�600 pixels.Stimulus generation and animation was controlledby the VPixx� graphics program, version 1.88

Table I. Participants characteristics for mTBI group.

Characteristics Frequency Percentage

GenderBoys 10 56Girls 8 44

Cause of injuryFalls 9 50Bicycle falls 4 22MVA, bicycle 1 6Hits (hockey, ball, etc) 4 22

Admission GCS13 0 014 5 2815 13 72

Duration of LOCNo LOC 11 610–1 min 6 33>1 min 1 6

Duration of PTA0–60 min 15 83>60 min 3 17

Concussion typesimple 15 83complex 3 17

Initial symptomsHeadache 14 78Nausea–Vomiting 8 44Dizziness 8 44Drowsiness 8 44Visual Problems 2 11Altered gait 2 11

GCS: Glasgow Coma Scale; LOC: loss of consciousness; MVA:motor vehicle accident; PTA: post-traumatic amnesia.

Mild traumatic brain injury induces prolonged visual processing deficits in children 659

Downloaded By: [Canadian Research Knowledge Network] At: 18:35 28 October 2008

(www.vpixx.com). Calibration and luminancereadings were regularly measured using a MinoltaCS-100 Chromameter. To minimize non-linearity inthe display, the luminance of the monitor wasgamma-corrected, implemented with a colourcalibration within the VPixx� program.

Psychophysical tasks

Participants underwent psychophysical testing todetermine their visuo-perceptual thresholds forstatic and dynamic first- and second-order stimuliand radial optic flow stimuli using two-alternativeforced choice (2-AFC) orientation or directiondiscrimination.

Static conditions. Figure 1 represents a schematic ofthe first- and second-order stimuli that were used toassess perceptual performances in children aftera mTBI.

The static conditions required participants toidentify the orientation of horizontal or verticalgratings, presented as first- or second-order patterns.Each stimulus was within a hard-edged circularregion at the centre of the display, subtending 10� indiameter at a viewing distance of 57 cm. The meanluminance of the display was 30.5 cd m�2

(u0 ¼0.1918, v0 ¼ 0.4344 in CIE [CommissionInternationale de l’Eclairage] u0 v0 colour space)where Lmin was 0.5 and Lmax was 60.5 cd m�2. First-order stimuli were a simple type of stimuli that wereluminance-defined (luminance-modulation depth)and were constructed by adding static grayscalenoise (1�1 pixel; measuring �1.86 minutes arc) toa static modulating sine wave grating. While second-order stimuli, a complex type of stimulus, were

texture-defined (contrast-modulation depth) andproduced by multiplying rather than adding thestatic grayscale noise to the modulating sine wavegrating, the noise dots’ individual luminances wererandomly assigned as a function of sin (x), where (x)ranged from 0–2 �. The average contrast of the noisewas set at half its maximum value. All stimuli had aspatial frequency of 0.5 cycle per degree (cpd).

The orientation-identification thresholds for thefirst-order stimuli were found by varying thecontrast, i.e. the luminance modulation depth,defined as the amplitude of the modulating sinewave, which ranged between 0.0–0.5:

Luminance modulation depth ¼Lmax � Lmin

Lmax þ Lmin

where Lmax and Lmin refer to the average highest andlowest local luminances in the stimulus. These first-order patterns were presented at five-to-six levels ofluminance modulation along log steps(0.04, 0.02, 0.01, 0.005, 0.0025 and 0.00125).

Thresholds for second-order stimuli were found byvarying the contrast-modulation depth of the staticpattern, defined as the amplitude of the modulatingsine wave, which ranged between 0.0–1.0:

Contrast modulation depth ¼Cmax � Cmin

Cmax þ Cmin

where Cmax and Cmin are the maximum andminimum local contrasts in the pattern. This typeof stimulus was also presented at five-to-sixpre-determined levels of contrast modulation(0.5, 0.25, 0.125, 0.0625, 0.0314 and 0.0156).

Dynamic conditions. The stimuli used for the direc-tion-identification task in the dynamic condition areillustrated in Figure 1(b). Except for their dynamiccharacteristic, the moving stimuli used for thedirection-identification task were identical with thestatic stimuli previously described for the orienta-tion-identification task in terms of physical proper-ties and parameters. The vertically-oriented gratingswere moving either to the right or to the left, with adrift frequency of 2 Hz. First- and second-orderdynamic patterns were also constructed by adding ormultiplying static grayscale noise to the modulatingsine wave grating. The dynamic first-order patternswere presented at five-to-six levels of luminancemodulation (0.02, 0.01, 0.005, 0.0025, 0.00125 and0.000625). For the dynamic second-order patterns,the same levels of contrast modulation werepresented as for static second-order stimuli.

Optic flow stimuli. Radial optic flow stimuli used inthis study simulate an observer moving in

Figure 1. (a) Static conditions: first- and second-order stimuli.First- and second-order static stimuli were presented eitherhorizontally or vertically to the participants at different depthsof modulation. The child was required to identify the orientationof the lines (i.e. horizontal or vertical) after each presentation.(b) Dynamic condition: first- and second-order stimuli. First- andsecond-order moving stimuli were presented moving either to theright or to the left of the participants at different modulationdepths. The children were required to discriminate the directionof the stimulus.

660 O. Brosseau-Lachaine et al.

Downloaded By: [Canadian Research Knowledge Network] At: 18:35 28 October 2008

translation through a circular tunnel [37].A fixation point appeared first for 1 second in thecentre of the screen (red dot subtending 0.5�).Then, a radial outward or inward moving patternappeared for 1 second. The pattern was constructedwith 150 moving white dots (10 cd m�2;0.52� 0.52� of visual angle), randomly distributed(with uniform density) on a dark background(1.8 cd m�2; for a Michelson contrast of 70%;u0 ¼ 0.1815 and v0 ¼0.5053) of large extent, sub-tending 104� � 79� of visual angle. In order tosimulate the optic flow motion seen in theenvironment, the elements follow a radial trajectoryfrom the centre of the pattern. In addition, dotspeed was projection-specified, meaning dot accel-eration increased as dots moved into the periphery.This acceleration was calculated using the square ofthe distance from the origin of expansion, i.e. by afactor of four as the distance doubled. Specifically,the dot speed was 7.5�s�1 at 10� and would travelat a speed of 30�s�1 at 20�of eccentricity. The dotswere following their trajectory in a continuousmanner, with no limited lifetime. Different levelsor proportion of dot coherence were presented. Forexample, for a coherence level of 50%, half of thedots were moving in the same direction and theother half were moving with the same speed but inrandom direction in the same path (jittering). Thisstimulus was presented at five-to-six levels along logsteps starting at 100% of coherence, to seek aminimum proportion of dots moving in the samedirection that yields a coherent percept. Thisstimulus is also qualified as ‘complex’ because ofthe larger neural integration required for perceivingthe overall direction of the optic flow motion.

Other measures. To ensure that participants hadnormal visual fields, they underwent the FrequencyDoubling Technology (FDT; HumphreyViewfinderTM Systems) perimetry assessment ateach testing session. The C-20 full thresholdmode test was used for both eyes independently.In the central 20� of the visual field, a centrallocation was tested and four locations weretested in each quadrant. Thresholds in dB wereobtained for every location, using a series of blackand white bands varying in contrast that flickerat 25 Hz.

Moreover, a questionnaire, based on theRivermead Post-Concussion Symptoms (PCS) wasadministered to both groups at every testing session[38]. Information about the trauma was extractedfrom the child’s medical record in order to describethe characteristics of the brain injury sustained(see Table I).

Testing procedures

Participants were tested during three separate ses-sions at three pre-determined times followingrecruitment to the study, which was after �1 week(mean�SD of 4.5 days �1.72 after recruitment and7.72 days �2.72 post-injury); 4 weeks (mean of 4.19weeks �0.54 after recruitment) and 12 weeks (meanof 12.47 weeks �0.99 after recruitment) and atcorresponding time intervals for controls. Thesetime sessions were chosen because of their impor-tance in the child’s recovery. The initial assessmentwas to seek a measure in the acute phase, the 4thweek corresponds to the end of an activity restrictionperiod imposed on children having sustained themost severe mTBI [39] and the 12th week waschosen because most post-concussion reportedsymptoms are generally resolved by that time [40].One child from the mTBI group could not completethe first evaluation because of the severity of hissymptoms (headaches), but did complete the twosubsequent assessments. In addition, two othermTBI children did not complete the study afterthe first and the second evaluation, for personalreasons.

The testing sessions took place at the VisualPsychophysics and Perception Laboratory located inSchool of Optometry of Universite de Montreal. If arecent eye exam on the child was not available, acomplete optometric assessment was done andchild’s vision was corrected to normal when neces-sary. Each session lasted �75 minutes (without theoptometric examination). The children were testedin a dimly lit room, sitting in a comfortable chair, at adistance of 57 cm from the screen upon which eachvisual stimulus was presented. The viewing of thedisplay was binocular and the children fixated a reddot centred on the screen. Each psychophysicalsession was of five testing conditions (i.e. static anddynamic conditions for first and second-orderstimuli and optic flow) where the order was counter-balanced across appointments and participants.Instructions were given verbally before each testand practice trials with feedback were completed tofamiliarize participants with every task.

For both first- and second-order static conditions,the children were required to identify the orientationof horizontally or vertically presented lines after eachpresentation. In the dynamic conditions, participantshad to identify the motion direction of first- andsecond-order patterns that were moving eitherrightward or leftward. Each stimulus was presentedfor 1 second, after which the child gave the experi-menter an appropriate verbal or non-verbal (i.e. ahand gesture) response, depending on the experi-mental condition (2-AFC). The experimenterinitiated each trial only when the participant

Mild traumatic brain injury induces prolonged visual processing deficits in children 661

Downloaded By: [Canadian Research Knowledge Network] At: 18:35 28 October 2008

attended to the screen and entered each responseobtained from the observer. The same generalprocedure was used for the optic flow stimuli.The children were required to discriminate whetherthe flow field was expanding (moving away from thecentre on the screen giving sensation of motiontowards the subject) or contracting (towards thecentre on the screen or giving the impression ofmotion away from the subject), after eachpresentation.

A method of constant stimuli was used to measurethresholds for all conditions. The pre-determinedmodulation depth (for first- and second-orderpatterns) or coherence levels (for optic flow) wererandomly varied in each experimental condition.Stimuli were presented 10 times for either orienta-tion or direction, for a total of 20 trials for eachmodulation depth or coherence level for everytesting condition. Weibull [41] functions werefitted to the obtained responses in order to estimateorientation- or direction-identification thresholds ata 75% correct level of performance (bootstrapprogram; Matlab v. 6.5).

Data analysis

Thresholds were measured for each child in each ofthe following conditions: orientation-identificationthresholds for first- and second-order static informa-tion, direction-identification thresholds for first- andsecond-order dynamic information and directiondiscrimination for optic flow stimuli. Log-transformdata of these thresholds were used in the statisticalanalysis. Also, threshold of central location from theFDT and scores on post-concussion symptomsquestionnaires were computed.

Analyses of variance were performed to investigatethe differences within each condition and testperformances of the children in the mTBI groupcompared with the control group. A two-way (groupand time) mixed model ANOVAs with repeatedmeasures on the time factor was used. Post-hocindependent t-tests with Bonferroni correction wereused to determine differences between groups ateach assessment time. One-way ANOVAs withrepeated measures were also used to assess thespecific time effect for each group, when needed.The ANOVAs were done only on complete data sets.However, bar graphs and t-tests were done using allavailable data. Relations between mTBI group’spost-concussion symptoms and performance onrelevant measures, such as contrast modulationdepth thresholds, were analysed with Pearsoncorrelation. Statistical analyses were carried outwith the SPSS statistical package (version 13;SPSS Inc.)

Results

Static conditions

Figure 2 shows the mean orientation-identificationthresholds for mTBI and control groups for first-order (Figure 2a) and second-order stimuli(Figure 2b) as a function of experimental sessionsin time. Separate analyses were performed for thetwo types of stimuli, because of their qualitativelydifferent defining attributes making a directcomparison non-informative. The two-wayANOVA performed on the log transform orienta-tion-identification thresholds for first-order staticstimuli (luminance modulation depth thresholds)revealed no statistically significant group-by-timeinteraction (F(2,62)¼ 2.515, p¼ 0.089), neither assignificant differences between groups across alltesting sessions (F(1,31)¼ 0.133, p¼ 0.717) or onthe time factor (repeated measures) (F(2,62)¼2.162, p¼ 0.124).

Figure 2. Static conditions. Mean orientation-identificationthresholds (þ1 SEM) for mild traumatic brain injury (mTBI)(open bars) and control children (filled bars) for first-order(a) and second-order (b) stimuli at 1, 4 and 12 weeks. (a) First-order static stimuli. No difference in thresholds was foundbetween groups across all testing sessions either on the timefactor. (b) Second-order static stimuli. A significant between-group effect was found but no main effect on the time factor,neither as for group� time interaction. Insets show a schematic ofthe conditions tested.

662 O. Brosseau-Lachaine et al.

Downloaded By: [Canadian Research Knowledge Network] At: 18:35 28 October 2008

The mixed within-between ANOVA conductedon orientation-identification thresholds for second-order stimuli (contrast modulation depth thresholds)did neither reveal an interaction effect of groupmembership and time (F(2,60)¼ 0.411, p¼ 0.665),nor a main effect of time ((2,60)¼1.80, p¼ 0.174).The analysis did, however, reveal a statisticallysignificant difference between groups (F(1,30)¼4.186, p¼ 0.05), indicating higher thresholds for themTBI group as compared to the control group.Specifically, post-hoc analysis revealed that injuredchildren’s thresholds were significantly higher thanthose of the non-injured at 4 weeks only (t¼ 2.56,p¼ 0.015, with Bonferroni correction). The absenceof significant effect on the time factor for both typeof stimuli (first- and second-order) demonstratesthat no learning effect was present betweenexperimental sessions for both groups.

Dynamic conditions

For dynamic conditions, a similar pattern of resultswas found for the first- and second-order motionstimuli, as for static conditions. Luminance modula-tion depth thresholds from dynamic first-orderstimuli did not differ significantly betweenmTBI and controls participants. The ANOVAdemonstrated no between-group effect for thisdirection-identification (F(1,30)¼0.007, p¼ 0.936)and no main effect for the time factor(F(2,60)¼ 0.650, p¼ 0.526). Figure 3(a) demon-strates the mean thresholds obtained for both groupswith these simple moving stimuli.

For the contrast modulation depth thresholds ofsecond-order moving patterns there was a strongbetween-group effect (mTBI vs controls)(F(1,30)¼ 13.334, p¼ 0.001), as shown inFigure 3(b). There was no main effect for the timefactor (F(2,60)¼ 0.832, p¼0.440) and no group bytime interaction (F(2,60)¼ 0.137, p¼ 0.872). Morespecifically, higher thresholds were obtained fromthe mTBI group compared to the control group atweek 1 (t¼ 2.786, p¼ 0.009), week 4 (t¼ 3.012,p¼ 0.005) and week 12 (t¼ 2.478, p¼ 0.019) (butnot significant at week 12 with Bonferroni correc-tion: p<0.017).

For first-order static and dynamic conditions, theanalysis showed no significant group or testing timeeffects. This implies that perceptual performancesfor simple visuo-spatial information are equalbetween groups and across time sessions.

In contrast, the second-order static and dynamicconditions analyses indicate that control childrenhad significantly better performances than childrenwho suffered from a mTBI. Moreover, the lack ofsignificant interaction between groups and testingtimes shows that difference between groups are

maintained across the testing period, up to 3 monthspost-injury.

Optic flow condition

Coherence thresholds were successfully obtained forall participants with the exception of one in eachgroup (mTBI group: week 12, controls: week 1)when tested with the optic flow condition. Thecoherence thresholds were significantly higher forthe mTBI group as compared to the control group,as revealed by the between-group effect of theperformed ANOVA (F(1,28)¼ 6.486, p¼0.017)(Figure 4).

There was a main effect for the time factor(repeated measures) (F(2,56)¼ 16.413, p< 0.001),but there was no group-by-time interaction(F(2,56)¼ 0.009, p¼ 0.991). Post-hoc analysis

Figure 3. Dynamic conditions. Mean direction-identificationthresholds (þ1 SEM) for mild traumatic brain injury (mTBI)(open bars) and control children (filled bars) for first-order(a) and second-order (b) stimuli at 1, 4 and 12 weeks. (a) First-order moving stimuli. No statistical difference in thresholds wasfound between groups across all testing sessions and no maineffect for the time factor. (b) Second-order dynamic stimuli.A significant between-group effect was found but no main effectfor the time factor and no group� time interaction. Insets show aschematic of the conditions tested.

Mild traumatic brain injury induces prolonged visual processing deficits in children 663

Downloaded By: [Canadian Research Knowledge Network] At: 18:35 28 October 2008

indicated that children with mTBI performed worse,exhibiting higher thresholds as compared to childrenin the control group, but only at 1 week post-injury(t¼ 2.605, p¼ 0.014, with Bonferroni correction).The time effect was present in each group, when thegroups were taken separately (for mTBI:F(2,26)¼ 8.205, p¼ 0.002; for controls: F(2,30)¼8.438, p¼ 0.001). Children from the mTBI grouppresented higher coherence thresholds to optic flowin the 1st week when compared to those of the 4th(p<0.001) and of the 12th (p¼ 0.028) week.Similar to the injured children, non-injuredparticipants improved their performance betweenweeks 1 and 4 (p¼ 0.007) and between weeks 1and 12 (p¼0.012). This implies a learningeffect between the first and the following assess-ments for both groups. Furthermore, the lack of asignificant interaction revealed that the differencesbetween groups were maintained throughout theassessment period after the injury and implies thatthere was no significant recovery across testingsessions.

Other measures

For the post-concussion symptoms, Figure 5demonstrates the mean scores obtained for mTBIand control children from the questionnaires, as afunction of experimental session in time.

Statistical analysis demonstrated a group-by-timeinteraction (F(2,52)¼ 5.161, p¼ 0.009). There wasno significant group effect for all testing sessionsalthough there was a tendency (F(1,26)¼ 3.77,p¼ 0.063) for mTBI participants to report

more post-concussion symptoms than controls.This implies that there was initially a significantdifference in the reported symptoms between thegroups, with mTBI children reporting more symp-toms compared to the control group (week 1(t¼ 4.464, p< 0.001) and then a recovery in thisdifference (week 4; t¼ 1.671, p¼ 0.106; week 12:t¼ 0.677, p¼ 0.504)). Moreover, there was a timesession effect with repeated measures(F(2,52)¼ 8.425, p¼ 0.001). When the groupswere analysed separately, the time effect was presentonly for mTBI children (F(2,20)¼ 7.469, p¼ 0.004)and injured children improved significantly betweenthe first and the 12th week (p¼0.028) only. Incontrast, no time effect was present for non-injuredchildren (F(2,32)¼ 0.578, p¼ 0.567).

For mTBI children, there was no significantcorrelation between the scores obtained on thepost-concussion symptoms questionnaire andthresholds from second order static (r¼ 0.353,p¼ 0.237), dynamic stimuli (r¼ 0.268, p¼0.376)and optic flow patterns (r¼0.039, p¼ 0.9) atweek 1. This indicates that there was no relationbetween the number and severity of symptomsreported after a mTBI and perceptual performancesto complex stimuli indicating that this study isassessing different elements with these measures.

An additional assessment was performed tocompare injured with non-injured children in termsof contrast sensitivity (FDT). Statistical analysisdemonstrated that there were no significant differ-ences in the central FDT measures, for right and lefteyes separately, between the two groups or acrosstesting sessions.

Figure 4. Optic flow condition. Mean coherence thresholds(þ1 SEM) for mild traumatic brain injury (mTBI) (open bars)and control children (filled bars) for radial optic flow stimuli at 1,4 and 12 weeks. A significant between-group effect and a maineffect for the time factor, but no group� time interaction wererevealed. Insets show a schematic of the conditions tested.

Figure 5. Rivermead post-concussion symptoms questionnaire.Mean scores (þ1 SEM) of mild traumatic brain injury (mTBI)(open bars) and control children (filled bars) at 1, 4 and 12 weeks.A group� time interaction and a main effect for the time factorwere revealed. Mild TBI children reported initially moresymptoms compared to controls and then a recovery in thisdifference was observed.

664 O. Brosseau-Lachaine et al.

Downloaded By: [Canadian Research Knowledge Network] At: 18:35 28 October 2008

Discussion

Deficits of higher level processing after mTBI

This investigation used a prospective design toassess visuo-spatial information perception in theacute phase after a mTBI in children. The overallfindings showed perceptual deficits for complexvisual information despite a normal neurologicalexamination at the time of hospital discharge.Indeed, injured children were selectively impairedon second-order, static or dynamic contrast-definedand optic flow pattern perception, compared withtheir non-injured controls. Furthermore, thisdiscrepancy in perceptual performance for complexstimuli was maintained throughout the assessedperiod, up to 3 months post-injury. In contrast,results obtained in this study showed thatsensitivities to simple visual information, morespecifically static or dynamic luminance-definedstimuli, were unaffected for children whosuffered a mTBI. Also, normal central contrastsensitivity (FDT) for a 25 Hz flickering stimuluswas found.

These findings revealed specific impairment formTBI children to integrate local elements in morecomplex visual patterns, such as second-order andoptic flow stimuli. This possibly reflects a deficiencyof the integrity of the occipito-parietal and occipito-temporal regions in higher level visual corticalfunctions, assessed with dynamic and static stimuli,respectively. The normal lower level functiondemonstrated with average first-order sensitivity, asopposed to the higher level function deficits,highlights a clear effect of complexity and a possiblegeneralized deficit in cortical integrative mechan-isms. This also suggests that the observed impair-ments did not result from a general visualimpairment.

The two types of dynamic and static stimuli arephysically similar, permitting a comparison of theneural integrity of the two cortical visual streamsat both early and later levels. In fact, first- andsecond-order stimuli tap into different hierarchicallevels in the visual cortex [12, 42]. First-orderpatterns are considered simpler because they areefficiently processed by the visual system.Specifically, local luminance variations are ana-lysed by neural detectors in area V1 for thedetection of direction or orientation. Whereas,second-order information is recognized as com-plex, as it requires additional non-linear neuralprocessing to be perceived. Higher visual areaswould be implicated in this processing, withrecruitment of more extended neural circuitry,such as V3 and VP, and V5 for both types ofdynamic stimuli [42].

Low and high levels processing in ageing and

other pathologies

As reported above, other studies from the laboratoryhave shown a complexity effect with comparablefirst- and second-order visual stimuli, such as in theinvestigation of visual perception in ageing [8].A larger sensitivity decrease was found forsecond-order stimuli in elderly, suggesting thatnon-pathological ageing may affect the additionalprocessing steps required for the analysis of higherlevel stimuli. Moreover, the authors suggested thatdiffuse and non-specific cell dysfunction in ageingcould account for their results. Similar to ageing, thisstudy proposes that the elevated thresholds forhigher-order stimuli in children who suffered amTBI reveal a diffuse, generalized later-level inte-gration malfunction rather than a region-specificimpairment.

Similar to other populations where neurobiologi-cal alterations are suspected such as individuals withautism and fragile X, this group has also demon-strated impairments of the neuro-integrativemechanisms used to detect complex motion andstatic stimuli [28–30]. It appears, therefore, thatusing the first- vs second-order (simple vs complex)methodology is quite sensitive to neurobiologicalalterations. Other studies have shown that thisapproach may not only be sensitive but is alsoquite selective in distinguishing among populations.For instance, this study has shown that in glaucomathe second-order motion processing was not moreaffected than the first-order motion processing [43].Glaucoma is an ocular disorder that affects theretinal ganglion cells of the retina that feed motionsensitive cells [44, 45] and these patients have beenshown to have motion processing deficits [44–48].As it is primarily a low-level neuropathology, it is notexpected that higher-level motion processing wouldbe more affected than the lower-level processing assupported by the results. It is clear, therefore, thatwhen one obtains a selective reduction of second-order processing, that this represents some form ofneurobiological alteration (anatomically and/or phy-siologically) which affects higher-level processing.An argument can be made that these techniques canbe useful in the assessment of these populations.

Another interesting finding is that perception ofoptic flow stimuli was also affected in this group ofchildren with mTBI. Impaired perception of opticalflow has also been shown in other populationspresenting cortical damage, such as patients withAlzheimer’s disease [49, 50]. This is interestingbecause optic flow is known to require complexvisual processing because signals from a multitude ofdirections must be integrated to generate a coherentpercept. This further supports the proposition that

Mild traumatic brain injury induces prolonged visual processing deficits in children 665

Downloaded By: [Canadian Research Knowledge Network] At: 18:35 28 October 2008

the mTBI group is still not at normal levels 12 weekspost-injury. However, if one had to select a specifictesting procedure between the first- and second-order grating tests or the optic flow one must favourthe former approach because, while the grating taskwas not affected by learning, there was a clearlearning effect for optic flow in both the controls andmTBI groups.

Visual deficits and symptoms after TBI

The investigation of visual perception per se after abrain injury in children has been neglected in theliterature, but other work had raised the possibilitythat tasks requiring intact integration of variousvisual information could be affected following amTBI. Indeed, a previous study had shown selec-tively decreased scores in preschool children whohad sustained a mTBI, when tested with interpretingvisual puzzles in a cognitive battery 6 months afterthe accident. This impairment persisted in the long-term, up to the age of 6.5 years [51]. In adults withmild-to-severe TBI, Lachapelle et al. [24] haveshown abnormal visual evoked potentials (VEP)with textures segregated by gradients of bothorientation and motion. In concordance with theseconclusions, they have suggested altered higher-order visual processing mechanisms after TBI.Another study showed postural dysfunctions inresponse to visual field motion in a virtual realityenvironment of young adult athletes after an mTBIthat was present up to 30 days post-injury. Theauthors suggested that trauma induced sensoryintegration dysfunction [17].

Children within the mTBI group reported morepost-concussion symptoms to the questionnaire thandid control children at the first assessment (week 1).Self-reported subjective symptoms have normalizedwith time, with children becoming asymptomatic, onaverage, at 1 and 3 months post-injury. However,results showed that even when the children becameasymptomatic, their ability to process complex visualinformation was affected. This demonstrates that thehigher-order measures were sensitive enough todetect subtle brain dysfunction even after very mildbrain injuries. An interesting finding here was thelack of significant correlations between thresholdsfor processing complex visual information and self-reported symptoms. This would lead one to believethat these functional tests and the self-reportedsymptoms scales measure different consequences ofbrain trauma. It also raises the question as to thevalidity of returning mTBI children to regularactivities based only on the resolution of reportedsymptoms.

Surprisingly, these findings demonstrated norecovery of visual processing abilities over the

assessment period for children with mTBI. Indeed,the impaired sensitivity to complex visual informa-tion lasted throughout the acute phase post-injury,i.e. encompassing the first 3 months after TBI [14].In a comparable paediatric sample, balance deficitswere also reported at 12 weeks after a mild TBI [16].While there is a debate in adult mTBI literaturewhether cognitive performance deficits persistbeyond the initial post-injury period, some studieshave reported sustained post-concussive impair-ments [52–55]. Furthermore, persistent neurobeha-vioural deficits in symptomatic sub-groups havebeen reported at least 12 months post-injury [56]and neurophysiological anomalies were revealed insymptomatic as well as asymptomatic concussedathletes even several weeks post-injury [57].

It is possible that this higher-order perceptualdeficit could have some consequences on activities ofdaily living. In fact, Gagnon et al. [58] have shown,with similar mTBI children, a lack of confidence intheir performance in physical activities 3 monthspost-injury, assessed with self-efficacy question-naires. Possibly, the perceptual impairmentsrevealed in this study might be expressed indemanding activities, such as team sports, thusaffecting the children’s self-confidence. This couldalso put them at greater risk of re-injury whenreturning to their physical and recreational activities.

Conclusion

In conclusion, the present study sheds some light onvisual perception skills of children who suffered amTBI. Injured children showed normal sensitivity tovisual perception of simple stimuli. In contrast, adifferent picture was revealed with stimuli necessi-tating more complex visual information processing.First, these findings demonstrate that the complexperceptual deficits persisted up to 3 months aftermTBI. In further studies a longer follow-up will beuseful to assess the resolution or the persistence ofthis perceptual deficit in time. Secondly, this studyargues that the psychophysical techniques usedpresently represent a sensitive tool in the assessmentof deficits after a mTBI in children. More specifi-cally, it is argued that assessing first- and second-order processing represents an optimal strategy forthe evaluation of the post-injury neurocognitivestatus. Thirdly, it is concluded that the perceptualtesting measures different functions than the self-reported symptom questionnaires. Finally, suchperceptual measures could significantly contributeto the clinical post-injury evaluation for preventingpremature returns to activities and sports andpotentially reduce risks of further injury.

666 O. Brosseau-Lachaine et al.

Downloaded By: [Canadian Research Knowledge Network] At: 18:35 28 October 2008

Acknowledgements

This work was supported by the Canadian Institutesof Health Research (operating grant MOP-74504and studentship to O.B.L and I.G.). We thank allparents and children for their participation to thestudy. Also, we would like to thank DebbieFriedman, Dre Marie Laberge and all the cliniciansinvolved in the recruitment of participants at theMontreal Children’s Hospital and at Sainte-Justine’sHospital.

Declaration of interest: The authors report noconflicts of interest. The authors alone are respon-sible for the content and writing of the paper.

References

1. Sabates N, Gouce M, Farris B. Neuro-ophthalmologicfindings in closed head trauma. Journal of Clinical Neuro-Ophthalmology 1991;11:273–277.

2. Ciuffreda KJ, et al. Occurrence of oculomotor dysfunctionsin acquired brain injury: a retrospective analysis. Optometry2007;78:155–161.

3. Heitger MH, et al. Eye movement and visuomotor armmovement deficits following mild closed head injury. Brain2004;127:575–590.

4. Suchoff IB, et al. The occurrence of ocular and visualdysfunctions in an acquired brain-injured patient sample.Journal of the American Optometry Association 1999;70:301–308.

5. Padula WV, et al. Evaluating and treating visual dysfunction.In: Zasler ND, editor. Brain injury medicine. New York:Demos Publishing; 2007.

6. Suh M, et al. Deficits in predictive smooth pursuit after mildtraumatic brain injury. Neuroscience Letters 2006;401:108–113.

7. Cavanagh P, Mather G. Motion: the long and short of it. SpatVis 1989;4:103–129.

8. Habak C, Faubert J. Larger effect of aging on the perceptionof higher-order stimuli. Vision Research 2000;40:943–950.

9. Wilson HR, Ferrera VP, Yo C. A psychophysically motivatedmodel for two-dimensional motion perception. VisionNeuroscience 1992;9:79–97.

10. Ledgeway T. Adaptation to second-order motion results in amotion after effect for directionally-ambiguous test stimuli.Vision Research 1994;34:2879–2889.

11. Bertone A, Faubert J. Motion sensitivity of first- and second-order translating, expanding/contracting and rotation pat-terns. Investigative Ophtalmology and Visual Science1999;40:2245.

12. Clifford CW, Vaina LM. A computational model of selectivedeficits in first and second-order motion processing. VisionResearch 1999;39:113–130.

13. Satz P, et al. Mild head injury in children and adolescents: areview of studies (1970–1995). Psychological Bulletin1997;122:107–131.

14. Frencham KA, Fox AM, Maybery MT. Neuropsychologicalstudies of mild traumatic brain injury: a meta-analytic reviewof research since 1995. Journal of Clinical and ExperimentalNeuropsychology 2005;27:334–351.

15. Binder LM, Rohling ML, Larrabee GJ. A review of mild headtrauma. Part I: meta-analytic review of neuropsychologicalstudies. Journal of Clinical and ExperimentalNeuropsychology 1997;19:421–431.

16. Gagnon I, et al. Children show decreased dynamic balanceafter mild traumatic brain injury. Archives of PhysicalMedicine and Rehabilitation 2004;85:444–452.

17. Slobounov S, et al. Alteration of postural responses to visualfield motion in mild traumatic brain injury. Neurosurgery2006;59:134–139; discussion 134–139.

18. Slobounov S, Sebastianelli W, Simon R. Neurophysiologicaland behavioral concomitants of mild brain injury in collegiateathletes. Clinical Neurophysiology 2002;113:185–193.

19. Slobounov S, Sebastianelli W, Moss R. Alteration of posture-related cortical potentials in mild traumatic brain injury.Neuroscience Letters 2005;383:251–255.

20. Geurts ACH, Knoop JA, Van Limbeek J. Is postural controlassociated with mental functioning in the persistent post-concussion syndrome? Archives of Physical Medicine andRehabilitation 1999;80:144–149.

21. Geurts ACH, et al. Identification of static anddynamic postural instability following traumatic braininjury. Archives of Physical Medicine and Rehabilitation1996;77:639–644.

22. Gagnon I, et al. Motor performance following a mildtraumatic brain injury in children: an exploratory study.Brain Injury 1998;12:843–853.

23. Gagnon I, et al. Visuomotor response time in children with amild traumatic brain injury. Journal of Head TraumaRehabilitation 2004;19:391–404.

24. Lachapelle J, et al. Texture segregation in traumaticbrain injury—a VEP study. Vision Research 2004;44:2835–2842.

25. Lachapelle J, Ptito A, McKerral M. Electrophysiologicalmarkers of impaired visual processing in mild to moderateTBI. 2007; submitted.

26. Nishida S, Ledgeway T, Edwards M. Dual multiple-scaleprocessing for motion in the human visual system. VisionResearch 1997;37:2685–2698.

27. Chubb C, Sperling G. Drift-balanced random stimuli: ageneral basis for studying non-Fourier motion perception.Journal of the Optical Society of America A 1988;5:1986–2007.

28. Bertone A, et al. Motion perception in autism: a‘complex’ issue. Journal of Cognitive Neuroscience2003;15:218–225.

29. Bertone A, et al. Enhanced and diminished visuo-spatialinformation processing in autism depends on stimuluscomplexity. Brain 2005;128:2430–2441.

30. Kogan CS, et al. Integrative cortical dysfunction andpervasive motion perception deficit in fragile X syndrome.Neurology 2004;63:1634–1639.

31. Wurtz RH. Optic flow: a brain region devoted to optic flowanalysis? Current Biology 1998;8:R554–R556.

32. Gibson JJ. The perception of the visual world. Boston:Houghton Mifflin; 1950. p xii,235.

33. Slobounov S, Sebastianelli W, Moss NE. Alteration ofposture-related cortical potentials in mild traumatic braininjury. Neuroscience Letters 2005;383:251–255.

34. Medecine, M.T.B.I.C.o.t.H.I.I.S.I.G.A.C.o.R., Definition ofmild traumatic brain injury. Journal of Head TraumaRehabilitation 1993;8:86–87.

35. Carroll LJ, et al. Methodological issues and researchrecommendations for mild traumatic brain injury: theWHO Collaborating Centre Task Force on Mild TraumaticBrain Injury. Journal of Rehabilitation Medicine 2004;43(Suppl):113–125.

36. McCrory P, et al. Summary and agreement statement of the2nd International Conference on Concussion in Sport,Prague 2004. British Journal of Sports Medicine2005;39(Suppl1):i78–i86.

Mild traumatic brain injury induces prolonged visual processing deficits in children 667

Downloaded By: [Canadian Research Knowledge Network] At: 18:35 28 October 2008

37. Habak C, Casanova C, Faubert J. Central and peripheralinteractions in the perception of optic flow. Vision Research2002;42:2843–2852.

38. King NS, et al. The Rivermead Post Concussion SymptomsQuestionnaire: a measure of symptoms commonly experi-enced after head injury and its reliability. Journal ofNeurology 1995;242:587–592.

39. Swaine BR, Friedman DS. Activity restrictions as part of thedischarge management for children with a traumatic headinjury. Journal of Head Trauma Rehabilitation 2001;16:292–301.

40. Levin HS, et al. Neurobehavioral outcome following minorhead injury: a three-center study. Journal of Neurosurgery1987;66:234–243.

41. Weibull W. A statistical distribution function of wideapplicability. Journal of Applied Mechanics 1951;18:292–297.

42. Smith AT, et al. The processing of first- and second-ordermotion in human visual cortex assessed by functionalmagnetic resonance imaging (fMRI). Journal ofNeuroscience 1998;18:3816–3830.

43. Karwatsky P, et al. Defining the nature of motionperception deficits in glaucoma using simple and complexmotion stimuli. Optometry and Visual Science 2006;83:466–472.

44. Glovinsky Y, Quigley HA, Dunkelberger GR. Retinal gang-lion cell loss is size dependent in experimental glaucoma.Investigative Ophthalmology and Visual Science 1991;32:484–491.

45. Quigley HA, Dunkelberger GR, Green WR. Chronic humanglaucoma causing selectively greater loss of large optic nervefibers. Ophthalmology 1988;95:357–363.

46. Shabana N, et al. Motion perception in glaucoma patients:a review. Surveys in Ophthalmology 2003;48:92–106.

47. Giaschi DE, et al. Loss of sensitivity to motion-definedform in patients with primary open-angle glaucoma and

ocular hypertension. Journal of the Optic Society of AmericaA: Optic Image Science Vision 1996;13:707–715.

48. Bosworth CF, Sample PA, Weinreb RN. Perimetric motionthresholds are elevated in glaucoma suspects and glaucomapatients. Vision Research 1997;37:1989–1997.

49. Tetewsky SJ, Duffy CJ. Visual loss and getting lost inAlzheimer’s disease. Neurology 1999;52:958–965.

50. O’Brien HL, et al. Visual mechanisms of spatial disorientationin Alzheimer’s disease. Cerebral Cortex 2001;11:1083–1092.

51. Wrightson P, McGinn V, Gronwall D. Mild head injuryin preschool children: evidence that it can be associatedwith a persisting cognitive defect. Journal of Neurology,Neurosurgery and Psychiatry 1995;59:375–380.

52. Bohnen NI, Jolles J, Twijnstra A. Neuropsychological deficitsin patients with persistent symptoms six months after mildhead injury. Neurosurgery 1992;30:692–695; discussion695–696.

53. Bohnen NI, et al. Late neurobehavioural symptoms after mildhead injury. Brain Injury 1995;9:27–33.

54. Potter DD, et al. Changes in event-related potentials in athree-stimulus auditory oddball task after mild head injury.Neuropsychologia 2001;39:1464–1472.

55. Chan RC. Attentional deficits in patients with persistingpostconcussive complaints: a general deficit or specificcomponent deficit? Journal of Clinical and ExperimentalNeuropsychology 2002;24:1081–1093.

56. Sterr A, et al. Are mild head injuries as mild as we think?Neurobehavioral concomitants of chronic post-concussionsyndrome. BMC Neurology 2006;6:7.

57. Gosselin N, et al. Neurophysiological anomalies in sympto-matic and asymptomatic concussed athletes. Neurosurgery2006;58:1151–1161, discussion 1151–1161.

58. Gagnon I, et al. Exploring children’s self-efficacy related tophysical activity performance after a mild traumatic braininjury. Journal of Head Trauma Rehabilitation 2005;20:436–449.

668 O. Brosseau-Lachaine et al.

Downloaded By: [Canadian Research Knowledge Network] At: 18:35 28 October 2008