7 Pages
English

How many days of monitoring predict physical activity and sedentary behaviour in older adults?

-

Gain access to the library to view online
Learn more

Description

The number of days of pedometer or accelerometer data needed to reliably assess physical activity (PA) is important for research that examines the relationship with health. While this important research has been completed in young to middle-aged adults, data is lacking in older adults. Further, data determining the number of days of self-reports PA data is also void. The purpose of this study was to examine the number of days needed to predict habitual PA and sedentary behaviour across pedometer, accelerometer, and physical activity log (PA log) data in older adults. Methods Participants (52 older men and women; age = 69.3 ± 7.4 years, range= 55-86 years) wore a Yamax Digiwalker SW-200 pedometer and an ActiGraph 7164 accelerometer while completing a PA log for 21 consecutive days. Mean differences each instrument and intensity between days of the week were examined using separate repeated measures analysis of variance for with pairwise comparisons. Spearman-Brown Prophecy Formulae based on Intraclass Correlations of .80, .85, .90 and .95 were used to predict the number of days of accelerometer or pedometer wear or PA log daily records needed to represent total PA, light PA, moderate-to-vigorous PA, and sedentary behaviour. Results Results of this study showed that three days of accelerometer data, four days of pedometer data, or four days of completing PA logs are needed to accurately predict PA levels in older adults. When examining time spent in specific intensities of PA, fewer days of data are needed for accurate prediction of time spent in that activity for ActiGraph but more for the PA log. To accurately predict average daily time spent in sedentary behaviour, five days of ActiGraph data are needed. Conclusions The number days of objective (pedometer and ActiGraph) and subjective (PA log) data needed to accurately estimate daily PA in older adults was relatively consistent. Despite no statistical differences between days for total PA by the pedometer and ActiGraph, the magnitude of differences between days suggests that day of the week cannot be completely ignored in the design and analysis of PA studies that involve < 7-day monitoring protocols for these instruments. More days of accelerometer data were needed to determine typical sedentary behaviour than PA level in this population of older adults.

Subjects

Informations

Published by
Published 01 January 2011
Reads 516
Language English

Hart
etal
.
InternationalJournalofBehavioralNutritionandPhysicalActivity
2011,
8
:62
http://www.ijbnpa.org/content/8/1/62

RESEARCH

OpenAccess

Howmanydaysofmonitoringpredictphysical
activityandsedentarybehaviourinolderadults?
TeresaLHart
1
,AnnMSwartz
2
,SusanECashin
2
andScottJStrath
2*

Abstract
Background:
Thenumberofdaysofpedometeroraccelerometerdataneededtoreliablyassessphysicalactivity
(PA)isimportantforresearchthatexaminestherelationshipwithhealth.Whilethisimportantresearchhasbeen
completedinyoungtomiddle-agedadults,dataislackinginolderadults.Further,datadeterminingthenumber
ofdaysofself-reportsPAdataisalsovoid.Thepurposeofthisstudywastoexaminethenumberofdaysneeded
topredicthabitualPAandsedentarybehaviouracrosspedometer,accelerometer,andphysicalactivitylog(PAlog)
datainolderadults.
Methods:
Participants(52oldermenandwomen;age=69.3±7.4years,range=55-86years)woreaYamax
DigiwalkerSW-200pedometerandanActiGraph7164accelerometerwhilecompletingaPAlogfor21consecutive
days.Meandifferenceseachinstrumentandintensitybetweendaysoftheweekwereexaminedusingseparate
repeatedmeasuresanalysisofvarianceforwithpairwisecomparisons.Spearman-BrownProphecyFormulaebased
onIntraclassCorrelationsof.80,.85,.90and.95wereusedtopredictthenumberofdaysofaccelerometeror
pedometerwearorPAlogdailyrecordsneededtorepresenttotalPA,lightPA,moderate-to-vigorousPA,and
sedentarybehaviour.
Results:
Resultsofthisstudyshowedthatthreedaysofaccelerometerdata,fourdaysofpedometerdata,orfour
daysofcompletingPAlogsareneededtoaccuratelypredictPAlevelsinolderadults.Whenexaminingtimespent
inspecificintensitiesofPA,fewerdaysofdataareneededforaccuratepredictionoftimespentinthatactivityfor
ActiGraphbutmoreforthePAlog.Toaccuratelypredictaveragedailytimespentinsedentarybehaviour,fivedays
ofActiGraphdataareneeded.
Conclusions:
Thenumberdaysofobjective(pedometerandActiGraph)andsubjective(PAlog)dataneededto
accuratelyestimatedailyPAinolderadultswasrelativelyconsistent.Despitenostatisticaldifferencesbetweendays
fortotalPAbythepedometerandActiGraph,themagnitudeofdifferencesbetweendayssuggeststhatdayofthe
weekcannotbecompletelyignoredinthedesignandanalysisofPAstudiesthatinvolve<7-daymonitoring
protocolsfortheseinstruments.Moredaysofaccelerometerdatawereneededtodeterminetypicalsedentary
behaviourthanPAlevelinthispopulationofolderadults.

Background
easytoadministerhoweveraresubjecttoerrorand
Physicalactivity(PA)isasporadicandcomplexbeha-recallbias[3].ObjectivemeasuresofPAandsedentary
viourtomeasureandissubjecttointer-andintra-indi-behaviour,suchaspedometersandaccelerometers,have
vidualvariability[1].Ithasalsobeensuggestedthatshownpromisewhenusedtoassesshabitualbehaviour.
sedentarybehaviourisanimportant,independentbeha-Determiningthenumberofdaystoreliablyassesshabi-
viourtoaccountforduetoitsrelationwithhealth[2].tualPAandsedentarybehaviourandminimizingparti-
Self-reportmethodstoassessPAandsedentarybeha-cipantburdenremainsachallenge.
vioursuchaslogs,questionnaires,orsurveys,areoftenThenumberofdaystoreliablypredicthabitualPA
behaviourforyoungandmiddle-agedadultshasbeen
*Correspondence:sstrath@uwm.edu
examinedusingpedometersandaccelerometers.Aspart
2
DepartmentofHumanMovementSciences,UniversityofWisconsin-
ofayear-longpedometerself-monitoringstudyinadults
Milwaukee,Milwaukee,Wisconsin,USA
Fulllistofauthorinformationisavailableattheendofthearticle
(meanage=38±10years),itwasdeterminedthatfive
©2011Hartetal;licenseeBioMedCentralLtd.ThisisanOpenAccessarticledistributedunderthetermsoftheCreativeCommons
AttributionLicense(http://creativecommons.org/licenses/by/2.0),whichpermitsunrestricteduse,distribution,andreproductionin
anymedium,providedtheoriginalworkisproperlycited.

Hart
etal
.
InternationalJournalofBehavioralNutritionandPhysicalActivity
2011,
8
:62
http://www.ijbnpa.org/content/8/1/62

Page2of7

consecutivedaysorsixrandomlyselecteddaysofdatalimited.Thereforethepurposeofthisstudywasto
wereneededtoproduceanintraclasscorrelation(ICC)examinethenumberofdataacquisitiondaysneededto
of0.80[4].Datafromsevendaysofconsecutiveped-reliablypredictPAandsedentarybehaviourusingaped-
ometermonitoringinadultmen(meanage=49.1±ometer,accelerometer,orPAloginanolderadult
16.2years)andwomen(meanage=44.8±16.9years)population.Secondly,weaimtoprovideanindicationof
suggestedthatthreedaysofmonitoringproducedanwhetherestimatesofPAandsedentarybehaviourvary
ICCof0.80orgreater[5].Similarresultswerereporteddependingonwhichdaysoftheweekareexamined..
whenassessingPAandsedentarybehaviourusinganTheresultsfromthisstudywillprovideusefulinforma-
accelerometerinmiddle-agedadults.ToreliablypredicttionregardingtheuseofPAassessmentmethodology
21daysofmonitoring,itwassuggestedthatthreetofortheolderadultpopulation.
fourdaysofaccelerometermonitoringwereneededto
achieve80%reliabilityfortotalPAaswellasmoderate
Methods
andvigorousintensityPA,andsevendaysofmonitoring
ParticipantsandProcedures
wereneededtopredictsedentarybehaviour[6].Participantsincluded52oldermenandwomen(mean±
Together,thesedatasuggestthataminimumofthreestandarddeviationage=69.3±7.4years,range=55-86
daysofobjectivemonitoringareneededtoreliablypre-years)whowererecruitedaspartofalargerongoing
dictPAbehaviour,whilesevendaysofobjectivemoni-trialexaminingobjectivelydeterminedphysicalactivity
toringareneededtopredicttimespentinsedentaryprofilesofcommunitydwellingolderadults.Participants
behaviourinayoungtomiddle-agedpopulation.woreaYamaxDigiwalkerSW-200(YamasaCorporation,
DespitethepopularityofestimatingPAandsedentaryTokyo,Japan)pedometerandanActiGraphmodel7164
behaviourstodeterminePAorSBprevalenceorrela-(formerlyCSIandMTI;ActiGraphLLC,Pensacola,FL)
tionshipswithvariousaspectsofhealth,thereremainaaccelerometerconcurrentlywhilecompletingaphysical
numberofgapsintheliteraturefocusingonthenumberactivitylog(PAlog)duringallwakinghours(excluding
ofdataacquisitiondaysneededtoreliablypredictPAshoweringandswimming)for21consecutivedays.
andsedentarybehaviours.First,whilestudieshaveTable1containsdemographicinformationforallparti-
reportedconsistentresultswithregardstonumberofcipants.Ethicalapprovalforthisstudywasgrantedby
daysofmonitoringusingobjectivemethodsofPAtheUniversityofWisconsin-MilwaukeeInstitutional
assessmentinayoungtomiddle-agedpopulation,dataReviewBoard.
onolderadultsislacking.Second,thenumberofdays
ofdataacquisitionneededtoreliablypredictPAbeha-
Instruments
viourfromsubjectivemethodsisvoidintheliterature.
ActiGraphAccelerometer
Finally,comprehensive,concurrentcomparisonsacrossTheActiGraph(model7164)usedinthisstudyisoneof
subjectiveandobjectivemethodsofPAandsedentarythemostwidelyusedaccelerometersinPAresearch.
behaviourassessmentwithinthesamepopulationareThissamemodelhasbeenusedforobjectivePA

Table1Participantdemographicsandphysicalactivitybehaviourdata
All(N=52)
Age(years)
69.3(7.4)

BodyMassIndex(kg/m
2
)

WaistCircumference(cm)

RestingSystolicBloodPressure(mmHg)

RestingDiastolicBloodPressure(mmHg)

Accelerometer(counts/day)

PALog(MET-min/day)

27.0(23.9,29.3)

94.5(83.9,101.0)

128(14)

7(7)9

250550(116260)

1107(358)

Pedometer(steps/day)
5589(3941,8971)
Note.Dataarepresentedasmean(standarddeviation)ormedian(interquartilerange).

Male(n=13)
68.7(10.2)

27.0(24.6,30.2)

100.0(92.0,111.0)

127(13)

74(9)

278430(152809)

1147(496)

8971(3894,9260)

Female(n=39)
69.6(6.3)

27.0(23.6,29.4)

90.3(81.6,100.4)

128(14)

78(8)

241257(102040)

1094(307)

5490(3892,7592)

Hart
etal
.
InternationalJournalofBehavioralNutritionandPhysicalActivity
2011,
8
:62
http://www.ijbnpa.org/content/8/1/62

monitoringintheNationalHealthandNutritionExami-
nationSurvey(NHANES)[7].Detailedtechnicalspecifi-
cationfortheActiGraphareprovidedelsewhere[8,9].
TheprimaryoutputsfromtheActiGraphareactivity
counts,whichrepresentrawaccelerationsthathave
beenfiltered,digitized,integratedandrescaled.Detected
activitycountsaresummedovereachepoch(i.e.,com-
monlyaminuteinlengthforadults)[10].Thesumof
theactivitycountsinagivenepochisrelatedtoactivity
intensityandcanbecategorized(e.g.,sedentary,light,
moderate,vigorous)basedonvalidatedactivitycount
cutpoints[8,9].Forthecurrentstudy,theActiGraph
waswornontherightsideofthewaistonanelastic
belt,accordingtomanufacturerspecifications.
YamaxDigiwalkerSW-200Pedometer
TheYamaxDigiwalkerSW-200isanelectronicped-
ometerwithahorizontal,spring-suspendedleverarm
whichmovesupanddownwithverticalaccelerationsof
thehip.Whenaccelerationsare

0.35g,theleverarm
makesanelectricalcontactandonestepisrecorded.
TheYamaxdisplaysthenumberofstepstakenduringa
givenperiodoftimeduringwhichthemonitorisworn
withadisplayoutputrangeof0-99,999steps.The
Yamaxattachestothewaistlineofpantsortoabeltat
themidlineofeitherthigh.TheYamaxDigiwalkeris
consideredtobethecriterionpedometerforfree-living
PAresearchstudies[11].
PhysicalActivityLog
ParticipantsrecordedactivitiesonthePAlogwhichhas
beenusedbyothersinpaststudies[12]attheendof
eachday.Alistofactivitieswasprovidedforthepartici-
pantonthedailylogsheetandthebroadcategories
included:householdactivities,lawn/gardenactivities,
volunteer/occupational,careofothers,transportation,
walking,dancing,sports,conditioning,andinactivity.
Alongwithprovidinginformationregardingthetypeof
activityperformed,participantswerealsoaskedtorecord
howlongeachactivitywasperformed.METvalueswere
assignedtoeachactivity[13]andmultipliedbythenum-
berofminuteseachactivityperformedresultinginMET-
min.AllPAlogswerecheckedbyaresearcherand
reviewedwitheachparticipanttoensurecompleteness.
BecausethePAlogdidnothaveposturaldetermination
components,wecombinedsedentarybehaviourswith
lightintensityPAintooneintensitycategory.Total
MET-min/daywerecalculatedforeachdayinadditionto
MET-min/dayineachofthetwointensitycategories(i.e.,
sed/lightandmoderate-to-vigorous).
DataTreatmentandStatisticalAnalysis
TodeterminetotalPAperformedeachday,totalactivity
countsfromtheActiGraph,totalnumberofsteps/day
fromtheYamax,andtotalMET-min/dayfromthePA
logwereutilizedinanalyses.ActiGraphdatawere

Page3of7

screenedfornon-weartimeusing60consecutivezero
counts/minute.ActiGraphcutpointsof0-50counts/min
wereconsideredtobesedentarybehaviour(basedon
cutpointssuggestedbyEsligeretal.[14]andCrouteret
al.[15]andfoundtoberepresentativeofsitting/lying
behaviour[16]),51-759counts/minwereconsideredto
belightPA[17],760-1951counts/minwereconsidered
tobemoderatelifestylePA,and

1952counts/minwere
consideredtobemoderate-tovigorous-intensityPA[8].
ForthePAlog,activitieswithMETvaluesrangingfrom
1-2.99wereconsideredsed/light,

3wereconsidered
moderate-tovigorous-intensityPA[18].Theresulting
dataforthePAlogwastotalMET-min/day,andMET-
min/dayineachintensitycategory.Non-normallydis-
tributeddataweretransformedusinglog+1forinfer-
entialanalyses.Descriptivestatisticswerepresentedin
theirrawform(i.e.,nottransformed).
Toaddressthemainpurposeofthispaper,Spearman-
BrownProphecyFormulasbasedonICCforall21days
andareliabilityof.80wereusedtopredictthenumber
ofdaysofcompletedataneededtorepresenttotalPA
(Yamaxsteps,totalActiGraphcounts,andtotalPAlog
MET-min/day),sedentarybehaviour(totalminutesfrom
accelerometerwith

50activitycounts/min),light-
intensity(ActiGraphmin/dayandPAlogMET-min/day),
moderate-intensity(ActiGraphmin/day),vigorous-inten-
sity(ActiGraphmin/day),andmoderate-tovigorous-
intensity(ActiGraphmin/dayandPAlogMET-min/
day)PA.Theseanalyseswererepeatedtocalculatea
reliabilityof0.85,0.9,and0.95.
Toaddressthesecondaryaimofthispaper,arepeated
measuresanalysisofvariance(RMANOVA)with
posthoc
pairwisecomparisonswherenecessarywasusedtodeter-
minebetweendaydifferencesinmeanPAlevel(i.e.,the
sevendaysoftheweek)foreachintensityandforallinstru-
ments.ToderivedailymeanPAestimatesfromthe21
days,datafromeachdayoftheweekwasaveraged(e.g.,
themeanofthethreeMondaysincludedinthemonitoring
periodwereaveragedtodeterminetheaveragePAbeha-
viouronMonday).AnalyseswerecompletedusingSAS
version9.1(Cary,IL)andSPSSversion17(Chicago,IL).
Statisticalsignificanceforallanalyseswassetatp<.05.
Results
Themeanwear/completiontimewas823.1±103.3
minutes/day.TotalPAfromtheActiGraph,Yamaxand
PAlogwereallnon-normallydistributedandtherefore
logtransformedforinferentialanalysis.Alllogtransfor-
mationsresultedinnormaldistributions.
NumberofDaysofCompleteDataNeededtoPredictPA
Behaviour
Thenumberofdaysofcompletedataneededtopredict
21daysoftotalPAbehaviourinthisolderadult

Hart
etal
.
InternationalJournalofBehavioralNutritionandPhysicalActivity
2011,
8
:62Page4of7
http://www.ijbnpa.org/content/8/1/62
Table2Numberofcompletedataacquisitiondaysneededtopredict21daysofphysicalactivitybehaviorfrom
objectiveandsubjectivemethodologiesasdeterminedbySpearmanBrownProphecyFormula
ActiGraph7164YamaxSW-200Physicalactivitylog
ReliabilityValue0.800.850.900.950.800.850.900.950.800.850.900.95
TotalPA
3461346102145818
SedentaryBehavior
571121
––––––––
Light-IntensityPA
34714
––––
14202121
Moderate-IntensityLifestylePA
23510
––––––––
MVPA
23510
––––
571221
Note.Physicalactivity(PA);moderate-tovigorous-intensityphysicalactivity(MVPA)
Note.ActiGraphcutpointsusedforthisanalysiswere0-50counts/min=sedentarybehavior;51-759counts/min=lightintensityphysicalactivity;760-1951=
moderatelifestylephysicalactivity;

1952=moderate-to-vigorousphysicalactivity(Crouteretal.,2006;Matthews,2005;Freedsonetal.,1998).ForthePAlog,
LightPA=activitiesrangingfrom1-2.99METsandMVPA=activities

3METs(Pateetal.,1995).
Note.Physicalactivityloglightintensityphysicalactivityisinclusiveofsedentarybehavior
populationwasconsistent,rangingfromthreedaysActiGraphandfortheYamaxshowednosignificantdif-
(ActiGraph)tofourdays(YamaxandPAlog;Table2).ferencesintotalPAlevelbetweendaysoftheweek(p=
Moredaysofcompletedata(fivedays)wereneededto.17forboththeActiGraphandtheYamax).Therewas
reliablypredictsedentarybehaviourfromtheActiGraphasignificantdifferenceintotalPAlevelobserved
comparedthenumberofdaysneededtopredicttimebetweendaysoftheweekbythePAlog(p=.04).
spentinlight-(ActiGraph:3days),moderate-(Acti-Resultsofthe
posthoc
pairwisecomparisonsaredis-
Graph:2days),andvigorous-intensityPA(ActiGraph:2playedonTable3.
days).Further,theseresultsshowtheself-reportofPADataforActiGraphvariableslightintensityPAand
behaviourrequiresmoredaysofcompletedata(14dayssedentarybehaviourwerenormallydistributed;allother
forlight-intensityPAand5formoderate-tovigorous-ActiGraphvariableswerenon-normalandlogtrans-
intensityPA),comparedwiththeActiGraph(3daysforformedforinferentialanalysis.Table4showsthe
lightintensityPAand4daysformoderate-tovigorous-means(95%CI)forActiGraph-determinedsedentary
intensityPA),toreliablypredict21daysoflight-inten-behaviour,lightPA,moderatelifestylePA,andmoder-
sityandmoderate-tovigorous-intensityPAbehaviourate-tovigorous-intensityPAforeachdayoftheweek.
(Table2).DataforallPAlogvariableswerenon-normallydistrib-
utedandlogtransformedforinferentialanalysis.The
WhichdaysoftheweekcanbeusedtopredicttotalPA
means(95%CI)forPAlog-estimatedMET-min/dayin
behaviour?
sed/light,moderate-tovigorous-intensityPAarealso
TotalPAfromtheActiGraph,YamaxandPAlogbydaypresentedinTable4.
oftheweekarepresentedinTable3asthemeanvalueResultsoftheRMANOVAshowednosignificantdif-
forthatdayoftheweek(95%confidenceinterval;95%ferencesbetweendaysoftheweekforsedentarybeha-
CI)basedonthe21daysofmonitoringperformed(e.g.,viourandlightPAbasedonActiGraphdata(p=.48
dataforMondayrepresentstheaverageofthethreeand.58,respectively;Table4).However,whenexamin-
Mondaysthatweremonitoredduringthe21monitoringingthedatafromthePAlog,resultsshowedsignificant
period).ResultsfromtheindividualRMANOVAforthedifferencesbetweendaysoftheweekforMET-min/day
Table3AveragetotaldailyphysicalactivitylevelasassessedbyActiGraph,Yamax,andPhysicalActivityLog
SundayMondayTuesdayWednesdayThursdayFridaySaturday
ActiGraphTotal
211205236379218298233436229008223605230597
ActivityCounts/day
(150350,(171532,(155461,(153418,(142342,(138209,(139701,
325184)344185)316880)318170)346925)303186)297831)
YamaxTotalSteps/day
5790618660226783603051505431
(3412,8836)(4037,9249)(3403,8663)(3618,8850)(3921,9167)(3692,8330)(3701,7795)
PALogTotalMET-min/
980971992103697210501112
day
(713,1353)
g
(803,1284)(768,1287)
g
(841,1322)
c
(754,1281)
b,g
(896,1300)
a
(918,1331)
a,c
Note.Datapresentedrepresenttheaveragevalueforeachdayoftheweekcalculatedfromoriginal21daysofdata.Dataarepresentedasmedian(interquartile
range).
Note.Superscript

a

representsdayswhicharesignificantlydifferentfromSunday;

b

representsdifferentfromMonday;

c

representsdifferentfromTuesday;

d

representsdifferentfromWednesday;

e

representsdifferentfromThursday;

f

representsdifferentfromFriday;and

g

representsdifferentfromSaturday.
SignificantdifferencesdeterminedusingRepeatedMeasuresAnalysisofVariance(logtransformationwasusedondataforalldaysforanalysiswheredatawere
non-normallydistributed).p<.05wasusedforstatisticalsignificance.

Hart
etal
.
InternationalJournalofBehavioralNutritionandPhysicalActivity
2011,
8
:62
http://www.ijbnpa.org/content/8/1/62

Page5of7

Table4DailyestimatedtimespentinphysicalactivityintensitiesbyActiGraph(minutes/day)andphysicalactivitylog
(MET-min/day)
SundayMondayTuesdayWednesdayThursdayFridaySaturday
(
S
m
e
i
d
n
e
/d
nt
a
a
y)
r
2
yBehavior*
(3924.72,4.4756.7)(3824.71,2.4341.9)(3894.12,1.4153.1)(406.443,7.4167.7)(3954.92,7.4859.7)(3904.62,3.4155.6)(3814.61,2.4944.3)
2LightPA*
(min/day)(2853.40,9.3333.1)(2913.11,3.3535.8)(293.361,23.631.6)(2913.21,2.3333.4)(2923.01,4.3537.9)(2913.21,3.3435.6)(3013.02,4.3147.1)
Sed/LightPA

713.2781.1710.4682.1692.5749.2735.0
(
MET-min/day)
3
(493.1,971.3)(596.6,983.8)
a
(556.3,942.5)
a
(559.7,1075.6)
a
(569.7,925.3)(550.8,1008.8)
a
(568.2,1083.3)
a
PMAo

d
(
e
m
r
i
a
n
t
/
e
d
L
a
i
y
f
)
e
2
style
(35.6,6942..56)
c,d,e,f
(45.3,1715.03.9)
c,e,f,g
(48.16,8.938.7)
a
(40.26,01.004.7)
b
(64.9,799.59.0)
a,b
(36.4,617.008.5)
a,c
(37.16,9.12153.4)
c
MVPA

(min/day)
2
12.315.713.715.012.013.014.7
(4.5,25.3)(6.5,39.0)(6.5,25.7)(7.7,31.5)(5.2,30.0)(6.8,29.8)(7.5,27.8)
MVPA

143.5249.1227.0223.0282.0278.0270.9
(
MET-min/day)
3
(57.8,363.9)(78.1,474.4)(92.5,404.3)(77.6,329.7)(159.5,515.8)(112.8,478.0)(97.5,455.0)
Note.Datapresentedre

presenttheaveragevalueforeachdayoftheweekcalculatedfromoriginal21daysofdata.Dataarepresentedas*mean(95%
confid
2
enceinterval)or
3
median(interquartilerange).
Note.,ActiGraphdata;,PhysicalActivityLogdata
Note.Physicalactivity(PA);moderate-to-vigorousphysicalactivity(MVPA).ActiGraphcutpointsusedforthisanalysiswere0-50counts/min=sedentarybehavior;
51-759counts/min=lightintensityphysicalactivity;760-1951=moderatelifestylephysicalactivity;

1952=moderate-to-vigorousphysicalactivity(Crouteretal.
[15];Matthews[17];Freedsonetal.[8]).
Note.ForthePAlog,sedentarybehavior/lightintensityphysicalactivity(Sed/LightPA);moderate-tovigorousintensityphysicalactivity(MVPA).Sed/LightPA=
activitiesrangingfrom1-2.99METsandMVPA=activities

3METs.(Pateetal.[18])
rNeoptree.sSeunptserdsicfrfieprte

nat

frreopmresWeendtsnedsadyasy;w

hei

crheparreesseingtnsifidcifafnetrleyndtifffreormenTthfruorsmdaSyu;n

fd

arye;pr

be

serentpsredsieffnetrsendtifffreoremntFrfirdoamy;Manodnd

ag

y;re

cp

rreespernetssednitfsfedrieffnetrfernotmfrSoamtuTrduaeys.day;

d

†nSiognn-infiocramntalldyiffdeirsetrnicbeustedde).teprm<i.n0e5dwuasisnugsReedpfeoartsetdatiMsteiacsalurseigsnAifnicaalynscise.ofVariance(logtransformationwasusedondataforalldaysforanalysiswheredatawere

insed/lightPA(p=.004).
Posthoc
pairwisecompari-(dependingontheintensity)ofcompletedataacquisi-
sonsshowedalldaysweresignificantlydifferentfromtionareneededtopredictintensityspecificPAbeha-
SundayexceptforThursday(Table4).viourfromaPAlog.Sedentarybehaviour,capturedin
SignificantdifferencesbetweendaysoftheweekwerethisstudyonlybytheActiGraph,required5daysof
presentwhenestimatingmoderate-intensitylifestylePAcompletedatatoreliablypredicttotaltimespentin
(p=.01)andmoderate-tovigorous-intensityPA(p=sedentarybehaviour.
.011)usingdatafromtheActiGraph.Noconsistentpat-Theresultsofthisstudyaresimilartopreviouslypub-
ternsofdifferencebetweendayscanbeidentifiedfromlishedfindingsusingobjectiveassessmenttools.Astudy
the
posthoc
pairwisecomparisons(Table4).Examina-of365daysofpedometerdataacquisitioninadults
tionofPAlevelbylogdatashowednosignificantdiffer-(meanage=38.0±9.9years)concludedthatfiveconse-
encebetweendaysoftheweekforMET-min/dayincutivedaysorsixrandomdayswereneededtoreliably
moderate-tovigorous-intensityPA(p=.06;Table4).estimatehabitualPA,howevernospecificdaysofthe
weekwererecommended[4].Tudor-Lockeetal.[5]col-
Discussion
lectedsevendaysofpedometerdatainadults(meanage
Thisstudyexaminedthenumberofcompletedata=49.1±16.2years)andsuggestedthatthreedaysof
acquisitiondaysneededtoreliablyestimate21daysofpedometermonitoringweresufficienttoreliablyesti-
PAandsedentarybehaviourusingobjectiveinstrumentsmatehabitualPA.Further,significantdifferencesinPA
(ActiGraphaccelerometerandYamaxDigiwalkerped-levelbetweendaysoftheweekwerereportedbyTudor-
ometer)andasubjectiveinstrument(PAlog)inanLockeetal[5]withtheprimarydifferencebeingSun-
olderadultpopulation.Resultsofthisstudysuggestday.Inspiteofthesedailydifferences,theauthorssug-
threetofourdaysofcompletedataarerequiredbyallgestedanythreedaysofmonitoringwouldbesufficient
instrumentstoreliablypredicttotalPAbehaviour.[5].Roweandcolleagues[19]examineddatafromolder
Whentheoutcomevariableofinterestistimespentinadults(meanage=74.0±9.5years)whoworeaped-
aspecificintensityofphysicalactivity,thenumberofometerfor7days.Resultsshowednodifferencesin
daysneededtoreliablypredictPAbehaviourisdepen-steps/daybetweendaysoftheweek,withtheexception
dentontheinstrumentused.TwotothreedaysofdataofSundayandauthorsconcludedthattwodaysofdata
acquisitionfromtheActiGraphwillprovidereliableesti-collectionwereneededtoreliablypredictsteps/day[19].
matesoflight-,moderatelifestyle-,andmoderate-toOurresultsaddtothiscurrentliteraturesetbysuggest-
vigorous-intensityPA,whereasfiveto14daysingthataminimumoffourdaysofcompletepedometer

Hart
etal
.
InternationalJournalofBehavioralNutritionandPhysicalActivity
2011,
8
:62
http://www.ijbnpa.org/content/8/1/62

dataareneededtoreliablyestimatetotalPAlevelinan
olderadultpopulation.However,despitethelackofsta-
tisticaldifferenceinsteps/daybetweendays,itshouldbe
notedthatthereisaclinicallysignificantdifferencein
thenumberofstepstakenperdaywhenexaminedby
dayoftheweek.Forinstance,theaveragedailysteps
takenduringthe21daymonitoringperiodonFriday
were5312,while6601stepsweretakenonSunday.This
differenceequatestoalmost1300stepsperday,orover
1/2mile[20-22].Whenlookingattheefficacyofaped-
ometerintervention,anincreaseof1000stepsperday
maybeclinicallysignificant,basedonthebaselinesteps
taken.Therefore,theauthorsrecommendthatthe
impactofdayoftheweekbeconsideredindesigning
andanalyzingPAstudies,forexampleinplanningfor
appropriatecoverageofeachdayoftheweekinthe
overallgroupifa<7-dayprotocolisused,andtocon-
trolforpossibledayoftheweekeffectsbymatchingor
statisticaladjustmentwhenrepeatedassessmentsare
used(e.g.ininterventionstudies).
WhenpredictingintensityspecificPAbehaviourusing
objectivemonitoring,itisapparentthatthenumberof
daysofcompletedataneededtoreliablypredictbeha-
viourarelessastheintensityofPAincreases.Moderate-
tovigorous-intensityPAisgenerallyplanned,predict-
able,andlessvariableinolderadults,therebyrequiring
fewerdaysofcompletedatatoreliablypredictthebeha-
viour[19].However,sedentarybehaviourandlight-
intensityPAarelesspredictableonaday-to-daybasis,
andthereforerequiremoremonitoringdaystoreliably
predictthebehaviour.Similartotheresultsofthecur-
rentstudythatsuggesttwo-tothree-daysofcomplete
dataacquisitionareneededtoreliablypredictPAlevel,
Roweandcolleagues[19]concludedthattwodaysof
datacollectionwereneededtoreliablypredictseven-
daysofmoderate-tovigorous-intensityPAinanolder
adultpopulation[18].Further,Matthewsetal.[6]
reportedthreetofourdaysofmonitoringwereneeded
topredict21daysofmoderate-intensitylifestyle(i.e.,
500-1951counts/min),andmoderate-tovigorous-inten-
sity(i.e.,

1952counts/min)PAinmiddle-agedmen
andwomen[6].Whenexaminingsedentarybehaviour,
thenumberofdaysneededtoreliablypredictbehaviour
increases.Matthewsetal.[6]suggestedsevendaysof
monitoringtoreliablyassessinactivity(i.e.,0-499activ-
itycounts/minute)comparedwithfivedaysofcomplete
dataneededtoreliablypredictsedentarybehaviour(i.e.,
0-50activitycounts/minute)inthecurrentstudy.
Togetherthesedatasuggestthatmoredaysofdata
acquisitionareneededtoreliablyestimatesedentary
behaviourcomparedwithPAbehaviourofatleasta
moderateintensity.
Resultsfromthisstudyandpublishedliteraturesug-
gestthatfewerdaysofmonitoringareneededwhen

Page6of7

objectivePAassessmenttoolsareused,comparedwith
subjectivePAassessmenttools.Resultsshowedlittle
day-to-daydifferencesbetweendaysoftheweekas
assessedbybothapedometerandaccelerometer,but
largervarianceswereseeninthePAlogdata,especially
whenexaminingsedentaryandlightintensitybehaviours
(uptoa146METmin/daydifferencebetweendays).
Previousresearchhasdemonstratedthatindividualsare
betterabletorecallPAofmoderate-tovigorous-inten-
sity,comparedwiththoseoflesserintensities[23],due
totheubiquitousnatureofsedentaryandlightintensity
activities,andthefactthatmanymoderate-andvigor-
ous-intensityactivitiesaregenerallyplannedandpurpo-
seful[24].Therefore,resultsofthisstudysupport
previouspublishedconclusionssuggestingmoredaysof
monitoringareneededtoreliablypredictPAfromself-
reportinstruments.Ourdataaddstothecurrentlitera-
turebymakingconcurrentassessmentsofthenumber
ofdaysofdataacquisitionneededtoreliablypredictPA
behaviourbysubjectiveandobjectivePAmethodologies.
Seven-daymonitoringprotocols,whichareoften
employedinstudiesthatexaminePAaseitheranout-
comevariable,orasanindependentvariable,would
achievereliabilityofICCsapproximately0.85to0.90for
allmeasuresofPAandsedentarybytheaccelerometer;
approximately0.85forthepedometerandapproxi-
mately0.85-0.90fortotalorMVPAbythePAlog,
given100%compliance.A7-daymonitoringperiod
couldreliablypredictlightintensityactivitybytheaccel-
erometerbutnotbythelog.Additionally,thesevenday
monitoringperiodavoidsissuesrelatedtoday-to-day
variationofPAbehaviour.Ourdatasupportthecontin-
ueduseofa7-dayPAmonitoringperiodtoreliablypre-
dictPAandsedentarybehaviour.
Thisstudyprovidesusefulinformationregardingthe
utilizationofmultiplemethodsofPAassessmentinthe
field,howeverithassomelimitations.First,thispopula-
tionwasfairlyhomogeneousinageandhealthstatus
whichmaycauseathreattoexternalvalidity.Addition-
ally,therewererelativelyfewmalescomparedto
femaleswhichmayhaveresultedfromvolunteerbias.
However,thistypeofhomogeneoussamplewasalso
seeninrelatedliterature[4,5,19].Also,therewasonlya
meanof1.3minutesofvigorousintensityPAas
detectedbytheaccelerometerwhichresultedinthe
combinationofmoderateandvigorousintensityPAfor
theanalysis.ThispracticeisconsistentwithTroianoet
al.[7].Thedatareportedinthispaperfocusedonrelia-
bilitywhenusingsomecommoncutpoints(whichwere
notdesignedspecificallyforanolderpopulation),how-
evergiventhesimilarityintherequirednumberofdays
ofmonitoringformoderate-lifestyle(760-1951counts/
min)andmoderate-to-vigorous(1952+counts/min)PA
wouldsuggestthattheseresultsarenotundulyaffected

Hart
etal
.
InternationalJournalofBehavioralNutritionandPhysicalActivity
2011,
8
:62
http://www.ijbnpa.org/content/8/1/62

bychoiceofcutpoints.Similarly,thelowamountof
MET-min/dayseeninvigorousintensityPAbythePA
logresultedinthecombiningofthemoderateandvig-
orousintensityPAcategories.Finally,therewasno
powercalculationforthisstudy.However,thesample
sizeofthecurrentstudyisinthemid-rangeofsamples
inrelatedliterature[4,5,19].
Conclusions
Thisstudyconcurrentlyexaminedthenumberofdays
ofmonitoringtoreliablyestimatehabitualPAand
sedentarybehaviourfromobjectiveandsubjectivePA
assessmentmethods.Basedontheresultsofthisstudy,
3-4daysofmonitoringareneededtoassesshabitualPA
regardlessofwhichinstrumentisselected.Despiteno
statisticaldifferencesbetweendaysfortotalPAbythe
pedometerandActiGraph,themagnitudeofdifferences
betweendayssuggeststhatdayoftheweekcannotbe
completelyignoredinthedesignandanalysisofPAstu-
diesthatinvolve<7-daymonitoringprotocolsforthese
instruments.IfspecificintensitiesofPAareanoutcome
ofinterest,additionaldaysmaybeneededforthePA
logbutnottheaccelerometer.Forsedentarybehaviour,
any5daysofmonitoringwillprovideareliableestimate
ofbehaviour.
Acknowledgements
ThisworkwaspartiallysupportedbyaCareerDevelopmentAward(Strath)
fromtheNationalInstituteofAging(K01AG025962).
Authordetails
1
DepartmentofHealthSciences,ArizonaStateUniversity,Phoenix,Arizona,
USA.
2
DepartmentofHumanMovementSciences,UniversityofWisconsin-
Milwaukee,Milwaukee,Wisconsin,USA.
Authors

contributions
THwasinvolvedwithdataanalysisandmanuscriptpreparation.SSandAS
wereinvolvedwithdatacollectionandmanuscriptpreparation.SCwas
involvedwithdataanalysis.Allauthorsreadandapprovedthefinal
manuscript.
Competinginterests
Theauthorsdeclarethattheyhavenocompetinginterests.
Received:26May2010Accepted:16June2011Published:16June2011
References
1.BaranowskiT,MasseLC,RaganB,WelkG:
Howmanydayswasthat?
We

restillnotsure,butwe

reaskingthequestionbetter

.
MedSciSports
Exerc
2008,
40
:S544-549.
2.PateRR,O

NeillJR,LobeloF:
Theevolvingdefinitionof

sedentary

.
Exerc
SportSciRev
2008,
36
:173-178.
3.MatthewsCE,WelkGJ:
UseofSelf-ReportInstrumentstoAssessPhysical
Activity.
In
PhysicalActivityAssessmentsforHealth-RelatedResearch.
Edited
by:WelkGJ.Champaign,IL:HumanKinetics;2002:107-121.
4.KangM,BassettDR,BarreiraTV,Tudor-LockeC,AinsworthB,ReisJP,
StrathS,SwartzA:
Howmanydaysareenough?Astudyof365daysof
pedometermonitoring.
ResQExercSport
2009,
80
:445-453.
5.Tudor-LockeC,BurkettL,ReisJP,AinsworthBE,MaceraCA,WilsonDK:
How
manydaysofpedometermonitoringpredictweeklyphysicalactivityin
adults?
Prevmed
2005,
40
:293-298.

Page7of7

6.MatthewsCE,AinsworthBE,ThompsonRW,BassettDRJr:
Sourcesof
varianceindailyphysicalactivitylevelsasmeasuredbyan
accelerometer.
MedSciSportsExerc
2002,
34
:1376-1381.
7.TroianoRP,BerriganD,DoddKW,MasseLC,TilertT,McDowellM:
Physical
activityintheUnitedStatesmeasuredbyaccelerometer.
MedSciSports
Exerc
2008,
40
:181-188.
8.FreedsonP,MelansonE,SirardJ:
CalibrationoftheComputerScienceand
Applications,Inc.accelerometer.
MedSciSportsExerc
1998,
30
:777-781.
9.MatthewsCE,ChenKY,FreedsonPS,BuchowskiMS,BeechBM,PateRR,
TroianoRP:
AmountoftimespentinsedentarybehaviorsintheUnited
States,2003-2004.
AmJEpidemiol
2008,
167
:875-881.
10.ChenKY,BassettDRJr:
Thetechnologyofaccelerometry-basedactivity
monitors:currentandfuture.
MedSciSportsExerc
2005,
37
:S490-500.
11.SchneiderPL,CrouterSE,BassettDR:
Pedometermeasuresoffree-living
physicalactivity:comparisonof13models.
MedSciSportsExerc
2004,
36
:331-335.
12.AinsworthBE,BassettDRJr,StrathSJ,SwartzAM,O

BrienWL,
ThompsonRW,JonesDA,MaceraCA,KimseyCD:
Comparisonofthree
methodsformeasuringthetimespentinphysicalactivity.
MedSciSports
Exerc
2000,
32
:S457-464.
13.AinsworthBE,HaskellWL,WhittMC,IrwinML,SwartzAM,StrathSJ,
O

BrienWL,BassettDRJr,SchmitzKH,EmplaincourtPO,JacobsDRJr,
LeonAS:
Compendiumofphysicalactivities:anupdateofactivitycodes
andMETintensities.
MedSciSportsExerc
2000,
32
:S498-504.
14.EsligerDW,CopelandJL,BarnesJD,TremblayMS:
Standardizingand
optimizingtheuseofaccelerometerdataforfree-livingphysicalactivity
monitoring.
JPhysActHealth
2005,
3
:366-383.
15.CrouterSE,ClowersKG,BassettDRJr:
Anovelmethodforusing
accelerometerdatatopredictenergyexpenditure.
JApplPhysiol
2006,
100
:1324-1331.
16.HartTL,McClainJJ,Tudor-LockeC:
Controlledandfree-livingevaluation
ofobjectivemeasuresofsedentarybehaviors.
JPhysActHealth
[http://
journals.humankinetics.com/jpah-in-press/jpah-in-press/controlled-and-free-
living-evaluation-of-objective-measures-of-sedentary-and-active-behaviors].
17.MatthewsCE:
Calibrationofacclerometeroutputforadults.
MedSci
SportsExerc
2005,
37
:S512-S522.
18.PateRR,PrattM,BlairSN,HaskellWL,MaceraCA,BouchardC,BuchnerD,
EttingerW,HeathGW,KingAC,KriskaA,LeonAS,MarcusBH,MorrisJ,
PaffenbargerRSJr,PatrickK,PollockML,RippeJM,SallisJ,WilmoreJH:
Physicalactivityandpublichealth:arecommendationfromtheCenters
forDiseaseControlandPreventionandtheAmericanCollegeofSports
Medicine.
JAMA
1995,
273
:402-407.
19.RoweDA,KembleCD,RobinsonTS,MaharMT:
Dailywalkinginolder
adults:day-to-dayvariabilityandcriterion-referencedvalidityoftotal
dailystepcounts.
JPhysActHealth
2007,
4
:434-446.
20.BassettDJr,StrathS:
Useofpedometersinassessingphysicalactivity.
In
Physicalactivityassessmentsforhealth-relatedresearch.
Editedby:WelkG,
DaleD.Champaign,IL:HumanKinetics;2002:163-77.
21.BassettDJr,AinsworthB,LeggettS,MathienC,MainJ,HunterD,
DuncanGE:
Accuracyoffiveelectronicpedometersformeasuring
distancewalked.
MedSciSportsExerc
1996,
28
:1071-7.
22.WelkGJ,DifferdingJA,ThompsonRW,BlairSN,DziuraJ,HartP:
Theutility
oftheDigi-walkerstepcountertoassessdailyphysicalactivitypatterns.
MedSciSportsExerc
2000,
32
:S481-8.
23.StrathSJ,BassettDRJr,SwartzAM:
ComparisonoftheCollegeAlumnus
QuestionnairePhysicalActivityIndexwithobjectivemonitoring.
Annals
ofEpidemiology
2004,
14
:409-415.
24.WarehamNJ,RennieKL:
Theassessmentofphysicalactivityinindividuals
andpopulations:Whytrytobemorepreciseabouthowphysical
activityisassessed?
IntJObes
1998,
22
:S30-S38.
doi:10.1186/1479-5868-8-62
Citethisarticleas:
Hart
etal
.:
Howmanydaysofmonitoringpredict
physicalactivityandsedentarybehaviourinolderadults?
International
JournalofBehavioralNutritionandPhysicalActivity
2011
8
:62.