JMusculoskeletNeuronalInteract2014;14(2):155-161
Original Article
Hylonome
Validity of an accelerometer as a vertical ground reaction
force measuring device in healthy children and
adolescents and in children and adolescents with
osteogenesis imperfecta type I
A. Pouliot-Laforte1,2,3*, L-N. Veilleux1,2*, F. Rauch1, M. Lemay2,3
1
ShrinersHospitalforChildren-CanadaandDepartementofPediatrics,McGillUniversity,Montréal,Québec,Canada;
MarieEnfantRehabilitationCenter,Sainte-JustineUniversityHospitalResearchCenter,Montréal,Québec,Canada;
3DepartmentofKinanthropology,UniversitéduQuébecàMontréal,Montréal,Québec,Canada
2
Abstract
Introduction: Verticalgroundreactionforces(vGRFs)arecloselyrelatedtobonestrengthanddevelopment.Itistherefore
relevanttoassesstheseforcesinbonedisordersaccompaniedwithmuscleweaknesssuchasinosteogenesisimperfectatypeI
(OItypeI).ThepurposeofthepresentstudywastoassessthevalidityofvGRFsderivedfromanaccelerometer.Methods: FourteenchildrenandadolescentswithadiagnosisofOItypeI(agerange:7to21;meanage[SD]:14.1[4.8]years;5males) and
fourteenhealthycontrols(agerange:6to21;meanage[SD]:12.5[4.2]years;5males)performedthreerepetitionsoffive
differentjumpandrisetestsonagroundreactionforceplate.Jumpsandrisesoutcomesweremeasuredsimultaneouslywiththe
groundreactionforceplateandanaccelerometer.Results: Pearsoncorrelationcoefficientswereover0.96(p<0.001)forthefive
tests.Thelimitsofagreementrepresentedbetween17and31%oftheaveragepeakforcemeasuredbybothdevices.Theaccelerometerisapromisingtooltoassessgroundreactionforcesineverydaylifesettingsandhasbeenshowntobesufficiently
sensitivetodetectmuscularweaknessinchildrenandadolescentwithOItypeI.
Keywords: Accelerometer,VerticalGroundReactionForces,Agreement,OsteogenesisImperfecta,HealthyChildrenandAdolescents
Introduction
Accelerometershavebecomeincreasinglypopulartoassess
physicalactivityinfree-livingconditions.Theyareoftenused
toestimatetheenergyexpenditureandhasbeenextensively
validatedinthisrespect1-4.However,theaccelerometerremains little investigated with regard to measuring vertical
groundreactionforces(vGRFs).Measuringthisparameteris
importantinbonedisordersasitrelatestothemechanicalload-
*Authorscontributedequallytotheproductionofthemanuscript.
Alloftheauthorshavenoconflictofinterest.
Corresponding author: Louis-Nicolas Veilleux, Shriners Hospital for Children,
1529 Cedar Avenue, Montréal, Québec, Canada H3G 1A6
E-mail: [email protected]
Edited by: J. Rittweger
Accepted 14 April 2014
ingoftheskeletonwhichstronglyinfluencesbonedevelopmentandregulation5.
Osteogenesisimperfecta(OI)isaheritabledisordercharacterizedbylowbonemassandincreasedbonefragilitythat
isusuallycausedbymutationsinoneofthetwogenesthatencodecollagentypeI,COL1A1 andCOL1A26.OItypeIisthe
mildestandmostcommonformofthedisorder7.Childrenand
adolescentswithOItypeItypicallyhavefewfunctionallimitationsbutsometimeshavemuscularweaknessofunclearetiology8-10. In that context, using a simple device like an
accelerometerwouldbehighlyvaluablefortheassessmentof
vGRFsandthereofthemechanicalloadingofthebones.This
isespeciallytrueforapopulationwithabonedisorderpresentingmuscleweaknesssuchasinOItypeI.
vGRFscanbeeasilyestimatedfromrawaccelerationprofilesthroughthesecondNewtonianlawofmotion(Force=
massXacceleration).Therefore,inadditiontoquantifyingenergyexpenditure,accelerometerscouldbeusedasaground
reactionforcemeasuringdeviceinfreelivingcondition.Re155
A.Pouliot-Laforteetal.:Accelerometer-derivedgroundreactionforce
lationshipsbetweencounts(acommonaccelerometeroutcome
variable) and vGRFs measured by a ground reaction force
platehavebeeninvestigated11-13 butthedefinitionof‘counts’
variesbetweenaccelerometerbrandandtherelationshipsare
task-specific12.
Afewstudieshaveassessedthedirectrelationshipbetween
vGRFsmeasuredbyaforceplateandanaccelerometer14-16.
Onlyoneofthesestudieswasdoneinapediatricpopulation14.
AccelerationsandvGRFsweremeasuredinthirty-five12to
14yearsoldadolescentsduringwalkingandrunning.TheauthorsshowedthatvGRFscorrelatewell(r2=0.97;p<0.001)
with actual vGRFs measured with a ground reaction force
plate.Theresultsofthisstudysuggestthatthisapproachis
validfortheinvestigationofsustainedandrepetitiveactivities
suchaswalkingandrunning.However,althoughwalkingand
runningtypifyafairproportionofchildren’severydaylife,anotherimportantproportionoftheseactivitiesconsistsinthe
productionofshortburstsofphysicalactivitysuchasthose
observedduringverticaljumpingorchair-rising17.Thesehigh
intensityactionsareknowntogeneratehighpeakforcesata
highratei.e.,generallylastinglessthan1s.AsnotedbyNeugebauerandcolleagues14,oneimportantlimitationoftheirstudy
isthatpeakresultantaccelerationwasaveragedoverarelativelylongperiodoftime(15sepochlength).Thisaveraging
methodisunlikelytoefficientlycapturethetransientaspect
ofjumpingandrisingmaneuvers.
Therefore, in the present study we suggest a simple approachinwhichinstantaccelerationdataareconvertedinto
vGRFsthroughthesecondNewtonianlawofmotion16.This
approachhasshownitspotentialinhealthyadultandelderly
populations,althoughithasnotyetbeeninvestigatedintypicallydevelopingchildrenandinchildrenwithmusculoskeletal
disorders.Healthychildrenandadolescentsaswellaschildren
andadolescentswithOItypeIwereaskedtoperformfivedifferentjumpandrisemanoeuvresonaportableforceplatform
whilewearinganaccelerometerontherighthip.Theselected
multipletwo-leggedhopping,themultipleone-leggedhopping,thesingletwo-leggedjump,theheel-riseandthechairrise tests have been validated in healthy children and
adolescents18 andinchildrenandadolescentwithOItypeI10.
Thespecificgoalofthisstudywastoassessthevalidityof
vGRFsmeasurementsderivedfromrawaccelerometerdatain
avarietyofactionsthatcloselyrepresentmovementsoccurring
inchildren’severydaylife17 andrequiringproductionofshort
buthighburstsofforce.Theselectedtestshavebeenshown
tocoveralargerangeofvGRFs19.
Participants and methods
Study population
FourteenchildrenandadolescentswithadiagnosisofOI
typeI(agerange:7to21;meanage[SD]:14.1[4.8]years;5
males)tookpartinthisstudy.Patientswererecruitedatthe
ShrinersHospitalforChildreninMontreal.Patientswereexcludediftheyhadanyfractureorsurgeryinthelowerlimbin
thepasttwelvemonths.Fourteenhealthychildrenandadoles156
cents(agerange:6to21;meanage[SD]:12.5[4.2]years;5
males)alsotookpartinthisstudy.Healthyparticipantswere
recruitedamongunaffectedsiblingsofpatients(i.e.,notpresentingclinicalsignsofOI),andchildrenofhospitalemployees. This study was approved by the Institutional Review
BoardoftheFacultyofMedicineofMcGillUniversity.Informedconsentwasprovidedbyparticipantsor,inminors,by
theirparents.Assentwasprovidedbyparticipantsaged7to
17years.
Measurement equipment
Accelerometer.TheGT3X+accelerometer(ActigraphTM,
LLC,Pensacola,FL,USA)wasusedtomeasureacceleration
inthreedimensions(vertical,antero-posteriorandmedio-lateral).TheGT3X+weighs19ganditsdimensionsare4.6cm
(height)X3.3cm(width)X1.5cm(thick).Theacceleration
datawassampledbya12bitanalogtodigitalconverteratrates
rangingfrom30Hzto100Hzandstoredinaraw,non-filtered
formatintheunitsofgravity(g’s).Thelargestaccelerationdetectable by the device is 6 g. In this study, the data were
recordedatafrequencyof60Hzandweredownloadedinraw
formattoalaptopwiththeActilife6.0softwarethroughaUSB
cable.
ForcePlate.TheLeonardoMechanograph® GroundReactionForcePlate(LeonardoMechanographyGRFP,Novotec
MedicalInc.,Pforzheim,Germany)wasusedtomeasureverticalgroundreactionforces18,19.Thesignalfromtheforcesensorswassampledatafrequencyof800Hzandwasanalyzed
usingtheLeonardoMechanographyGRFPResearchEdition®
software(LeonardoMechanographyGRFPResearchEdition
Software,version4.2-b05.53-RES;NovotecMedical).
Height was measured using a Harpenden stadiometer
(Holtain,Crymych,UK).Bodymasswasdeterminedusingthe
portableforceplatformLeonardoMechanographyGRFP.
Test procedure
Theexperimentercarefullypositionedtheaccelerometeron
thepatient’srightwaistslightlybehindtheanterioriliaccrest4.
Thedevicewasheldinplacedirectlyontheparticipant’sskin
withanelasticband.
Priortoeachtest,theexperimenterprovidedadescription
oftheprocedureandaphysicaldemonstrationofthetaskto
theparticipant.Theforceplatformwasadjustedtoindicatea
massofzerokgbeforeaparticipantsteppedontoit.Theparticipantstoodonthedeviceinanuprightpositionwiththeir
armsrelaxedontheirsides,onefootoneachsideoftheplatformatshoulderwidth.Bodymasswasrecordedoncetheparticipant stood still for at least 2 seconds. Following a
single-tonepitch,theparticipantperformedthetestmanoeuvre.Aftereachtrial,theparticipantremainedstillforatleast
2seconds.Adouble-tonepitchindicatedtheendofthetest.
Participantswereaskedtoperformfivedifferenttestsinthe
followingorder:multipletwo-leggedhopping,multipleoneleggedhopping,singletwo-leggedjump,heel-risetest,chairrisetest,asdescribedindetailelsewhere18.Threevalidtrials
wereperformedforeachtest.Atrialwasdefinedasasingle
A.Pouliot-Laforteetal.:Accelerometer-derivedgroundreactionforce
Gender(M/F)
Age(years)
Height(cm)
BodyMass(kg)
Control
OI type I
P
5/9
12.5(4.2)
152.4(21.6)
48.7(17.8)
5/9
14.1(4.8)
150.1(17.3)
47.9(20.9)
0.37
0.76
0.92
Results are shown as mean (SD).
Table 1. Anthropometricdata.
Control
OI
∆%
Multiple Two-Legged Hoppinga
GT3X
LeonardoGRFP
4.73(0.76)
4.63(0.79)
3.83(0.56)
3.88(0.49)
-24
-20
Multiple One-Legged Hopping-Right Legb
GT3X
LeonardoGRFP
2.95(0.39)
3.07(0.43)
2.26(0.27)
2.50(0.26)
-30
-23
Multiple One-Legged Hopping-Left Legb
GT3X
LeonardoGRFP
3.22(0.52)
3.02(0.48)
2.56(0.23)
2.44(0.23)
-26
-24
Single Two-Legged Jump
GT3X
LeonardoGRFP
2.43(0.48)
2.20(0.23)
2.29(0.30)
2.13(0.21)
-6
-3
Heel-Rise Test
GT3X
LeonardoGRFP
1.43(0.19)
1.50(0.16)
1.39(0.10)
1.45(0.10)
-3
-3
Chair-Rise Testc
GT3X
LeonardoGRFP
1.49(0.19)
1.57(0.14)
1.39(0.12)
1.43(0.17)
-7
-10
Results are shown as mean (SD). GT3X: Accelerometer data; Leonardo GRFP: force plate data; Significant differences between groups:
p=0.003; bp<0.001; cp=0.04.
a
Table 2. Peakforcemeasurementsrelativetobodyweightinpatientsandcontrolparticipantsasafunctionofthetestsandthemeasuringdevice.
jump(forthesingletwo-leggedjump)orasaseriesof5movements(heel-risetestandchair-risetest)or10consecutivevertical up-and-down movements (multiple two-legged and
multipleone-leggedhopping).Thecorrectexecutionofeach
trialwasvisuallyassessedbyanexperiencedexperimenter.At
theendofthefirsttestsession,theaccelerometerwasgiven
backtotheexperimenterandthedataweredownloadedtoa
laptop.
Data analysis
ToassessvalidityandreproducibilityoftheGT3X+forthe
measurementofverticalgroundreactionforces,rawverticalaccelerationsweretransformedintoforcewiththefollowingformula:
(1)vGRFs(kN)=|((a (g) *9.807(m/s2))*BM (kg))/1000|.
wherea istheinstantaccelerationandBM isthebodymassof
theparticipant.
TodeterminevalidityoftheGT3X+asaforce-measuring
device,thehighestforcevalueofeachofthethreetrialsfor
eachofthefivetestswereusedforfurtherstatisticalanalyses.
Statistical analyses
To assess validity, Pearson correlation between data
recordedbytheGT3Xaccelerometerandthegroundreaction
forceplatewasestablishedforcomparisontosimilarstudies12,13.BlandandAltmanplotsandlimitsofagreementwere
calculatedforeachgroupandforeachofthefiveteststodeterminethelevelofagreementbetweentheaccelerometerand
thegroundreactionforceplatemeasurements20.BlandandAltmananalyseswereperformedoneachgroupindependently.It
maybeassumedthatOIpatientsdonotengageinthesame
kindofphysicalactivitiesasdotheirhealthypeersduetofear
offractureandhigherfracturerisks.Therefore,OIchildren
maynothavethesamelevelofexpertiseashealthychildren
duetotheirlackofpractice.Asaresult,jumpingmovement
157
A.Pouliot-Laforteetal.:Accelerometer-derivedgroundreactionforce
Figure 1. BlandandAltmanplotsshowingthedifferencesbetweenpeakgroundreactionforces(Fmax;kN)asmeasuredbytheGT3XAccelerometer
andtheLeonardoGroundReactionForcePlateagainsttheaveragevalues(dotedline),with95%limitsofagreement(greyshadow)foreachof
thefiveclinicaltests.BlackcirclesrepresentdataofparticipantswithOItypeIandwhitefilledcirclesrepresentdataofcontrolparticipants.
patternsmaydifferbetweenOIpatientsandcontrols.This
couldleadtoincreasedinstabilityatthetrunkwhichinturn
mayinfluencerecordingsduetoincreasedmedio-lateraland
antero-posteriorgroundreactionforceproduction.
Peakforceofthreeconsecutivetrialswaskeptforthisanalysis.Therefore,theBlandandAltmanmethodforrepeatedmeasurementswasused21.Heteroscedasticitywaspresentinonetest
(S2LJ),thatis,theabsolutedifferenceswerecorrelatedwiththe
magnitudeofthemean(R<0.05).Underthiscircumstance,itis
suggestedthatdatashouldbelogtransformedtoremovethisrelationship.Therefore,dataofbothsystemswere(natural)log
transformedandthestandardapproachwasthenappliedon
thesedata.However,thisproceduredidnotallowremovingthe
relationshipandthereforethenormalapproachwasused.An
ANOVAcontrastingtwogroups(Controlvs.OItypeI)andtwo
devices(accelerometervs.forceplate)withrepeatedmeasures
onthelastfactorwascomputedtodetectanysystematicdifferencebetweengroupsanddevicesonabsolutepeakforceand
peakforcerelativetobodyweight18.Allstatisticalanalyseswere
158
performedusingPASW18® (SPSSInc.,Chicago,IL,USA).
Thep valuesaresetat0.05.
Results
PatientswithOItypeIdidnotdifferfromhealthyparticipantswithregardtoage,heightandbodymass(Table1).One
patientwasunabletogenerateenoughforcetoperformboth
multipleone-leggedhoppingtests(i.e.ontherightandleft
foot)whereasanotherpatientcouldnotperformthemultiple
one-leggedhoppingontheleftfootforthesamereason.The
ANOVAwithpeakforcerelativetobodyweightsetasthedependentvariable(Table2)revealedthatpatientswithOIproducedsignificantlylowerforcesthanhealthyparticipantsfor
themultipleone-legged(p<0.001)andtwo-leggedhopping
(p=0.003)testsaswellasforthechair-risingtest(p=0.04),regardlessofthedeviceusedforforcerecordings.Nodifferences
werefoundforthesingletwo-leggedjumpandfortheheelrisetestonbothdevices.
A.Pouliot-Laforteetal.:Accelerometer-derivedgroundreactionforce
GT3X
Leonardo GRFP
∆%
Multiple Two-Legged Hopping
Healthy
OItypeI
2.28(0.93)
1.72(0.73)
2.25(0.97)
1.75(0.75)
1
-1
Multiple One-Legged Hopping-Right Lega
Healthy
OItypeI
1.44(0.61)
1.02(0.48)
1.49(0.62)
1.13(0.48)
-3
-9
Multiple One-Legged Hopping-Left Legb
Healthy
OItypeI
1.58(0.69)
1.22(0.50)
1.48(0.66)
1.16(0.50)
7
5
Single Two-Legged Jumpb
Healthy
OItypeI
1.20(0.57)
1.09(0.53)
1.06(0.42)
1.01(0.48)
14
8
Heel-Rise Testa
Healthy
OItypeI
0.70(0.29)
0.65(0.30)
0.72(0.28)
0.68(0.31)
-3
-4
Chair-Rise Testd
Healthy
OItypeI
0.72(0.27)
0.65(0.30)
0.76(0.30)
0.67(0.31)
-5
-2
Results are shown as mean (SD). GT3X: Accelerometer data; Leonardo GRFP: force plate data; Significant differences between measuring
devices ap=0.003, bp=0.001, cp<0.001, dp=0.04.
Table 3. Absolutepeakforcemeasurements(kN)inpatientsandcontrolparticipantsasafunctionofthetestsandthemeasuringdevice.
TheresultsoftheBlandandAltmananalyses,morespecificallythelimitsofagreementpresentedinpercentoftheaverage
forcemeasurements,weresimilarbetweengroupsandwereas
follows:Multipletwo-leggedhopping:26%vs.33%;multiple
one-leggedhoppingoftherightleg:25vs.26%;multipleoneleggedhopping-leftleg:22vs.23%;singletwo-leggedjump:36
vs23%;heel-risetest:16vs15%andchair-risetest28vs.20%,
respectivelyforthehealthyparticipantsandOIpatients.Therefore,wedecidedtopoolalldataforsimplicitysake.TheBland
andAltmanplots(Figure1)illustratetheagreementbetween
forcemeasurementsderivedfromtheaccelerometerandthose
collectedbythegroundreactionforceplate.Thelimitsofagreementrepresented24%oftheaveragepeakforcemeasuredby
bothdevicesforthemultipletwo-leggedhopping,themultiple
one-leggedhoppingontherightlegandthechairrisetestand
23%31%,17%forthemultipleone-leggedhoppingontheleft
leg,thesingletwo-leggedjumpandtheheel-risetest,respectively.Correlationcoefficientsbetweenpeakgroundreaction
forcemeasuredbytheaccelerometerandtheGRFPwereof:0.97
(Multipletwo-leggedhopping),0.96(Multipleone-leggedhopping-Leftleg),0.97(Multipleone-legged-Rightleg),0.96(Single
two-legged jump, Heel-rise test), 0.99 (Chair-rise test; all
p’s<0.001)
TheANOVA(Table3)comparingtheaveragedpeakforceof
theaccelerometertothoseofthegroundreactionforceplate
showedthattheaccelerometerinbothgroupssignificantlyunderestimatedtheforcesmeasuredbytheforceplateby6%in
themultipleone-leggedhoppingontherightfoot(p=0.003),
andby4%intheheel-risetest(p<0.001)andthechair-risetest
(p=0.04)butsignificantlyoverestimatedtheseby6%inthemul-
tipleone-leggedhoppingtestontheleftfoot(p=0.04)andby
11%inthesingletwo-leggedjump(p=0.001).Forthemultiple
two-leggedhoppingtherewasnoobservableoverestimationor
underestimationoftheforcesbytheaccelerometer(p=1).
Discussion
ThisstudyonchildrenandadolescentswithOItypeIand
onhealthyparticipantsshowsmoderatetogoodagreementbetweenforcesderivedfromanaccelerometerandthosemeasuredbyagroundreactionforceplateduringtheexecutionof
fivedifferentmanoeuvres.Importantly,however,theobserved
differencesinvGRFsbetweentheOItypeIgroupandthecontrolgroupwereverysimilarwiththetwodevices.
Previousstudieshavereportedclosecorrelations,between
vGRFsmeasuredbyaforceplatformandaccelerometerderivedvGRFsduringaheel-risingtest(r=0.98)16,verticaljumping(r=0.90)15 andwalkingandrunning(r2=0.97)14.Thehigh
correlationsreportedinthecurrentstudyaresimilartothatobservedinthosepreviouslystatedstudiesandconcurtosuggest
thatmeasuringvGRFswithanaccelerometerisahighlyvalid
approachinchildrenandadolescents.However,correlation
analysisisastatisticallyinadequatewaytoestablishagreement
betweentwomethods20.Correlationsreflectthespreadofvalueswithinastudypopulationandthereforeprovideinformationaboutaspecificstudypopulationratherthanaboutthe
methodsthataretobecompared22.Theyalsoprovideameasureofthestrengthoftherelationshipbetweentwovariables
butfailtospecificallymeasureagreementbetweenthesetwo
159
A.Pouliot-Laforteetal.:Accelerometer-derivedgroundreactionforce
variables.Forexample,aperfectcorrelationcouldbeobserved
ifdatapointsweredistributedalonganystraightlinewhereas
perfectagreementwouldbeobservedonlyifdatapointswere
distributedalongthelineofequality.TheBlandandAltman
approachhasbeendevelopedtoavoidthesepitfalls20.Using
thisapproach,wefoundlimitsofagreementthatwerebetween
16%to24%oftheaverageforcemeasurementsin5ofthe6
testsandahigherlimitofagreementforthesingletwo-legged
jump(31%).Inouropinion,theseresultssuggestacceptable
agreementbetweenthetwomethods.However,itcontrasts
withtheveryhighcorrelationsthatweandothershavereportedandhighlightstheusefulnessoftheBlandandAltman
approachtocomplementthemoretraditionalcorrelationapproachinvalidationstudies.
Despitesomedifferencesinforcereadings,ourdatasuggeststhattheaccelerometerisapotentiallyvaluabletoolto
estimategroundreactionforcesineverydaylifesettings.The
smallover/underestimationofaveragedvGRFsbytheaccelerometer’sderivedforcessuggeststhatoveralongperiod
ofrecordings(i.e.manyrepetitions)forcemeasurementsare
quiteaccurate.ThattheaccelerometercouldreflectvGRFs
ofeverydaylifemovementsaccuratelyissupportedbythe
factthatthetestsusedinthecurrentstudyreflectawide
rangeofgroundreactionforces;from1.3to4.7timesbody
weight19. Most importantly, our data indicate that the accelerometerandtheforceplatefoundsimilardifferencesbetweentheOItypeIcohortandthecontrolgroup.Itmaybe
possibletouseanaccelerometerinstudiesthatcomparemuscleforcebetweentwogroupsofrelativelyfunctionalstudy
cohortswiththeteststhatwereevaluatedinthepresentstudy.
Therefore,theaccelerometercanbeasuitabletoolforassessinggroundreactionforcesduringtheteststhatwereevaluatedinthepresentstudy.
Theagreementbetweentheaccelerometerandtheforce
platemeasurementsmaybeinfluencedbymovementartifacts
oftheaccelerometer,asitisplaceddirectlyontheskin23.This
couldexplaintheaccelerometerforceoverestimationinthe
multipleone-leggedhoppingontheleftlegandinthesingle
two-leggedjump.Positioningtheaccelerometerontheright
hipmayhavealsoincreasedthediscrepancybetweenthetwo
methods,especiallyduringasymmetricmovements,suchas
one-leggedhopping23.Itmighthavebeenmoreappropriateto
placetheaccelerometeratornearthecenterofmass,i.e.atthe
lowerback16.Inthecurrentstudy,weoptedfortherighthip
placementbecausethissitehasbeenextensivelyvalidatedfor
theassessmentofenergyexpenditureinhomebasedsettings.
Finally,itwasnotedthattherewasaproportionalbiasinthe
singletwo-leggedjump,i.e.thatforceswereunderestimated
atlowlevelsofvGRFswhereastheywereoverestimatedat
higherlevelsofvGRFs.Thistestrequiresproductionofpronouncedsimultaneoushipandtrunkflexionsandextensions.
Becausetheaccelerometerwaswornatthehip,itispossible
thatthevelocityatwhichtheextensionsoccurpriortoliftoff
hadaninfluenceontheaccelerometersuchasdisplacement
artifactsorinclination.
160
Conclusions
Thegoalofthepresentstudywastodetermineifonecould
userawaccelerationdatatocomputedgroundreactionsforces
indifferentmaneuversmimickingafairproportionofchildren’severydaylifeactivities.Highcorrelationswerereported
aswellaslimitsofagreementrepresentingonaverage20%of
themeasuredforces.Inaddition,theaccelerometerderived
dataalloweddetectionofexpectedsignificantdifferencesbetweenbothstudypopulations.Accelerometryseemstobea
validtooltoestimatevGRFsineverydaylifesettingsintypicallydevelopingchildrenaswellasinchildrenwihOI.
Acknowledgements
This study was supported by the Shriners of North America, the Fonds
de la Recherche du Québec en Santé (FRQS), the Fondation Go and the
MENTOR-RSBO program supported by the Canadian Institute for Health
Research (CIHR) and the FRQ-S. Study sponsors had no involvement in
the writing of the manuscript; and in the decision to submit the manuscript
for publication.
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
CrouterSE,SchneiderPL,KarabulutM,BassettDR,Jr.
Validityof10electronicpedometersformeasuringsteps,
distance,andenergycost.MedSciSportsExerc2003;
35:1455-60.
EkelundU,SjostromM,YngveA,PoortvlietE,Nilsson
A,FrobergK,etal.Physicalactivityassessedbyactivity
monitoranddoublylabeledwaterinchildren.MedSci
SportsExerc2001;33:275-81.
NilssonA,BrageS,RiddochC,AnderssenSA,Sardinha
LB,WedderkoppN,etal.Comparisonofequationsfor
predictingenergyexpenditurefromaccelerometercounts
inchildren.ScandJMedSciSports2008;18:643-50.
PuyauMR,AdolphAL,VohraFA,ButteNF.Validation
andcalibrationofphysicalactivitymonitorsinchildren.
ObesRes2002;10:150-7.
RauchF,SchoenauE.Thedevelopingbone:slaveormasterofitscellsandmolecules?PediatrRes2001;50:309-14.
ForlinoA,CabralWA,BarnesAM,MariniJC.Newperspectivesonosteogenesisimperfecta.NatRevEndocrinol
2011;7:540-57.
BenAmorIM,RoughleyP,GlorieuxFH,RauchF.Skeletal clinical characteristics of osteogenesis imperfecta
causedbyhaploinsufficiencymutationsinCOL1A1.J
BoneMinerRes2013;28:2001-7.
EngelbertRH,BeemerFA,vanderGraafY,HeldersPJ.
Osteogenesisimperfectainchildhood:impairmentand
disability-afollow-upstudy.ArchPhysMedRehabil
1999;80:896-903.
TakkenT,TerlingenHC,HeldersPJ,PruijsH,Vander
Ent CK, Engelbert RH. Cardiopulmonary fitness and
musclestrengthinpatientswithosteogenesisimperfecta
typeI.JPediatr2004;145:813-8.
A.Pouliot-Laforteetal.:Accelerometer-derivedgroundreactionforce
10. VeilleuxLN,LemayM,Pouliot-LaforteA,CheungMS,
Glorieux FH, Rauch F. Muscle anatomy and dynamic
musclefunctioninosteogenesisimperfectatypeI.JClin
EndocrinolMetab2014;99:E356-62.
11. GarciaAW,LangenthalCR,Angulo-BarrosoRM,Gross
MM.Acomparisonofaccelerometersforpredictingenergyexpenditureandverticalgroundreactionforcein
school-agechildren.MeasurePhysEducExercSci2004;
8:119-44.
12. JanzKF,RaoS,BaumannHJ,SchultzJL.Measuringchildren’sverticalgroundreactionforceswithaccelerometry
duringwalking,running,andjumping:TheIowabonedevelopmentstudy.PediatrExercSci2003;15:34-43.
13. RowlandsAV,StilesVH.Accelerometercountsandraw
accelerationoutputinrelationtomechanicalloading.J
Biomech2012;45:448-54.
14. Neugebauer JM, Hawkins DA, Beckett L. Estimating
YouthLocomotionGroundReactionForcesUsinganAccelerometer-BasedActivityMonitor.PLoSONE2012;
7:1-7.
15. ElvinNG,ElvinAA,ArnoczkySP.Correlationbetween
groundreactionforceandtibialaccelerationinvertical
jumping.JAppliedBiomech2007;23:180.
16. SchmidS,HilfikerR,RadlingerL.Reliabilityandvalidityoftrunkaccelerometry-derivedperformancemeasure-
17.
18.
19.
20.
21.
22.
23.
mentsinastandardizedheel-risetestinelderlysubjects.
JRehabilResDev2011;48:1137-44.
VanPraaghE.Anaerobicfitnesstests:whatarewemeasuring?In:TomkinsonGR,OldsTS,editors.PediatricFitness:SecularTrendsandGeographicVariability.Basel:
Karger;2007.p.26-45.
Veilleux LN, Rauch F. Reproducibility of jumping
mechanographyinhealthychildrenandadults.JMusculoskeletNeuronalInteract2010;10:256-66.
VeilleuxLN,RauchF,LemayM,BallazL.Agreement
betweenverticalgroundreactionforceandgroundreactionforcevectorinfivecommonclinicaltests.JMusculoskeletNeuronalInteract2012;12:219-23.
BlandJM,AltmanDG.Statisticalmethodsforassessing
agreementbetweentwomethodsofclinicalmeasurement.
Lancet1986;1:307-10.
BlandJM,AltmanDG.Measuringagreementinmethod
comparisonstudies.StatMethodsMedRes1999;8:135-60.
deVetHC,TerweeCB,KnolDL,BouterLM.Whento
useagreementversusreliabilitymeasures.JClinEpidemiol2006;59:1033-9.
HornerFE,SladeJ,BilzonJL.Theeffectofanatomical
placementandtrunkadiposityonthereliabilityandvalidityoftriaxialaccelerometeroutputduringtreadmillexercise.JPhysActHealth2013;10:1193-200.
161