Epoch length and the physical activity bout analysis: An accelerometry research issue.

0.23 MB·English

by  Ayabe, Makoto

Ayabeetal.BMCResearchNotes2013,6:20
http://www.biomedcentral.com/1756-0500/6/20
RESEARCH ARTICLE Open Access
Epoch length and the physical activity bout
analysis: An accelerometry research issue
Makoto Ayabe1*, Hideaki Kumahara1,2, Kazuhiro Morimura1 and Hiroaki Tanaka1
Abstract
Background: The purpose of the present investigation was to compare the bouts ofdaily physical activity (PA)
determined by threedifferent accelerometerepochlengths under free-living conditions.
Methods: One hundredthirty-four adults (50±7years) wore an accelerometer (Lifecorder)for 7consecutive days
under free-living conditions in order to determine the time spent in physical activity of light intensity(LPA),
moderate intensity (MPA), vigorous intensity (VPA), moderate to vigorous intensity (MVPA), and the total physical
activity (TPA;sum ofLPA, MPA and VPA). Additionally, allPA was divided according to the bout durations
(sporadic, >3 min, >5 min, and>10 min). These indicesof PA were analyzed using threedifferent epoch lengths
(4sec, 20 sec and 60 sec) derived from the accelerometer.
Results: The LPA significantly increased in association with increases in theepoch length (48.7±15.9 to 178.7±
62.6 min/day, p<0.05). The amount of sporadic VPAdetermined by the shortest epoch length (2.9±5.2min/day)
was significantly longer thanthetwo remaining epochlengths (1.1±2.4to 0.9±2.5min/day,p<0.05). The times
ofthe MVPA bouts lasting longer than 3 minutes determined using the 4-second epochlength (2.6±5.4 to 7.7±
10.0 min/day) were significantly shorterthan those determined using the other two settings (6.5±10.5 to 13.8±
13.8 min/day, p<0.05). The frequenciesof theMVPA bouts lasting longer than10 minutes determined using the
4-second epoch length (0.2±0.3 bouts/day) were significantlylowerthan those determined using the other two
settings (0.3±0.4 bouts/day, p<0.05).
Conclusion: The epoch length setting of the accelerometer affects theestimationof the PA bouts under free-living
conditions inmiddle-aged to older adults.
Keywords: Accelerometer, Pedometer, Pedometry, Methodology, Accuracy,Validity
Background count toward the 30 minute minimum by performing
The American College of Sports Medicine and the boutsofactivitylasting10minutesormore[1].
American Heart Association recommend that all healthy Strathetal.[4]showedthatboutsofPAlasting≥10min-
adults 18 to 65 years of age engage in moderate-intensity utes may constitute a more time-efficient strategy to
aerobic (endurance) physical activity (MPA) or vigorous- decrease body mass index and waist circumference. In
intensity aerobic physical activity (VPA) in order to addition,accumulatingMVPAinshorterboutsmightbea
promote and maintain health [1]. Since several consensus beneficial starting point for individuals to increase their
statements also recommend engaging in moderate to vig- PA levels and decrease their body mass index and waist
orous intensity physical activity (MVPA) [2,3], spending a circumference. Furthermore, sustained volitional activity
longer time performing MVPA is widely accepted to be a (i.e.,≥10mininduration)mightplayanimportantrolefor
goal for a healthy lifestyle. Additionally, moderate- the long-term maintenance of weight loss [5]. These find-
intensityaerobicactivity,whichisgenerallyequivalenttoa ings indicate that participating in MVPA lasting>10 min
brisk walk that noticeably accelerates the heart rate, can might be more beneficial than accumulating sporadic
MVPA.
*Correspondence:[email protected] However,withregardtoboutsofMVPAlasting>10min-
1FacultyofHealthandSportsScience,FukuokaUniversity,8-19-1Nanakuma,
utes, no consistent definition has been used to evaluate
Jonan-ku,Fukuoka814-0180,Japan
Fulllistofauthorinformationisavailableattheendofthearticle
©2013Ayabeetal.;licenseeBioMedCentralLtd.ThisisanOpenAccessarticledistributedunderthetermsoftheCreative
CommonsAttributionLicense(http://creativecommons.org/licenses/by/2.0),whichpermitsunrestricteduse,distribution,and
reproductioninanymedium,providedtheoriginalworkisproperlycited.
Ayabeetal.BMCResearchNotes2013,6:20 Page2of7
http://www.biomedcentral.com/1756-0500/6/20
activity monitor data, and the traditional use of 1 minute to the measurements were performed between March
recordingintervalsseemstobethedefactoboutdefinition andApril.
[6]. Accelerometers function by integrating a filtered digi-
tizedaccelerationsignaloverauser-specifiedtimeinterval, Physicalactivityassessments
commonly referred to as an epoch [7]. The usual acceler- During the course of the present investigation, all parti-
ometer stored magnitude of accelerations at fixed record- cipants wore a pedometer with a uni-axial accelerometer
ing intervals (1 sec, 4 sec, 15 sec, or 60 sec or longer) is (Lifecorder Ex 4-sec version, Kenz, Nagoya, Japan; LC)
called an “epoch” [8]. Atthe endof eachepoch,the some under free-livingconditions. TheLCisa small andlight-
index of physical activity is calculated; this process is weight (60 mm×46 mm×26 mm, 42 g) device that is
repeated until data collection is completed [8,9]. Recent equipped with a uni-axial piezo-electronic accelerometer
studiesexaminedtheappropriateepochlengthtoestimate that samples vertical acceleration ranging between
thesporadicMVPA[7,8,10],andtheuseofashortertime 0.06 G and 1.94 G at 32 Hz. Based on the magnitude
samplingintervalmightreducethemisclassificationerrors and frequency of accelerations, the LC determines the
of PA estimates [8]. However, there have been no studies level of movement intensity every 4 seconds on a scale
that have directly compared the bouts of PA determined of 1 (minimal intensity of movement) to 9 (maximal in-
usingdifferent epochlengths under free-livingconditions. tensity of movement). As shown in a previous investiga-
Since most walking bouts last 30 sec or less [11,12], we tion [9], the intensity levels closely approximated the
hypothesized that a shorter epoch length may be more metabolic equivalents (METs) of the activities being per-
suitable to reflect the “real” bout duration under free- formed. Previous investigations have demonstrated that
livingconditions. the LC is suitable for research purposes [9,13,14]. The
Therefore, the purpose of the present investigation was LC has the potential limitation, similar to other types of
to examine the effects of the epoch length lasting from activity monitors, in that they too often remain vulner-
4 seconds to 1 minute on the evaluation of bouts of PA able to problems from external vibrations and unmeas-
lasting longer than 3, 5 and 10 minutes under free-living ured sources of energy expenditure such as hill climbing
conditionsinmiddle-toolder-agedadults. and isometric activity in daily life [15]. After reading the
instructions regarding the general care of these activity
monitors, the participants wore the LC for 10 days con-
Methods tinuously, except while sleeping or bathing. We used
Subjects seven continuous days (the 4th day to the 10th day, five
One hundred and thirty-four adults, aged from 32 to weekdays and two weekend days) for data the analysis
72 yr., participated in the present investigation. All because the 1st to the 10th days included non-wearing
volunteers were recruited from Fukuoka city (Fukuoka, time and the 2nd day was used to familiarize the sub-
Japan) by the newspaper and web site. All volunteers liv- jects with the LC. All subjects documented when they
ing independently and were free of debilitating chronic removed the LC die for personal reasons. The LC was
diseases. Furthermore, none of the participants habit- placed on the left anterior mid-line of the thigh on the
ually participated in any sports activities. The character- waist band of the participant’s clothing. The present in-
istics of the participants are shown in Table 1 (Table 1). vestigation placed the LC on the left side in order to
After an explanation of the study requirements, each prevent the LC from being dropped while the subjects
subject read and signed a consent form. The ethics com- inserted or removed their belts. The LC was sealed, so
mittee of Fukuoka University approved all procedures that the participants could not gain access to the PA
used in the present investigation. All procedures related measurements during the experimental period. After the
data collection period was completed, the participants
returned the LC by mail to the investigators so that the
Table1Characteristicsofparticipants stored data could be uploaded to a personal computer
for the analysis. To beincluded inthe analysis,each par-
All Female Male
ticipant had to have at least seven days of sufficiently
N 134 42 92
wearingthe device(>10hours)[7].
Age(years) 50±7 50±7 50±8
Height(cm) 165.0±8.1 156.2±4.9 169.1±5.8* Accelerometer’sdataanalysis
Bodyweight(kg) 66.7±12.4 54.7±8.4 72.3±9.8* All accelerometer’s out puts were averaged over 7 days,
Bodymassindex(kg/m2) 23.3±6.9 28.5±6.9 21.0±5.5* andweeklyvaluesweredefinedasindividualsrepresented.
Sincethepresentinvestigationwasfocusedontheacceler-
Waistcircumference 88.2±8.8 84.3±9.7 90.0±7.8*
ometer’s algorithms, the inter individual variability, such
Dataareexpressedasthemeanswiththestandarddeviation(mean±SD).
*Significantlydifferentbetweenmalesandfemales(p<0.01). astheday-to-dayvariabilityorhourlyvariability,werenot
Ayabeetal.BMCResearchNotes2013,6:20 Page3of7
http://www.biomedcentral.com/1756-0500/6/20
evaluated according to the previous validation studies metabolic syndrome in previous studies [16,18,19]. Fur-
[8,14]. In the present investigation, the epoch length was thermore,theboutsofPAlastinglongerthan5mincom-
alsosetat20secand60sec,inadditiontotheusual4sec prised just 0.14 percent of all walking bouts under free-
setting. The reasons why we chose these three epoch living conditions [11,12]. Finally, the PA bouts lasting
lengths are as follows: first, the 4 sec and 60 sec epoch longer than 10 min have been widely recommended by
lengths were used in previous studies [6,16,17]. Addition- consensusstatements[1](Figure1).
ally, as the LC detects the intensity of PA every 4 sec, the
epoch length had to be selected from a multiple of 4 sec, Statistics
and the epoch length must be a common divisor of the The data were expressed as the means with standard
bout duration (3 min, 5 min and 10 min). Therefore, we deviations. The potential gender-related differences in
chose20secasthethirdepochlength. the obtained data were analyzed by the unpaired t-test.
ThetimespentinlightintensityPA(LPA),moderatein- Aone-wayANOVAandScheffe’stestwereusedtocom-
tensityPA(MPA),vigorousintensityPA(VPA),andmod- pare the PA levels for the 3 different epoch lengths. The
erate+vigorous intensity PA (MVPA) was determined. magnitude of the differences between the time and fre-
Furthermore, the sum of the time for LPA, MPA, and quency of PA for the 4 sec epoch length compared to
VPA was defined as the total PA (TPA). Thereafter, based those of the other two epoch lengths was analyzed by
on the time series data of the three epoch lengths, the paired t-tests. Statistical significance was set at P<0.05.
TPAandMVPAweredividedaccordingtotheboutdura- All statistical analyses were performed using the Stat-
tions (Sporadic, < 3 min, < 5 min, and<10 min, respect- View software program (version 5.0.1, SAS Institute,
ively). We defined a “bout” as PA that was maintained Cary, NC,USA)(Figure 2).
completely at an intensity level of 1 or higher for at least
3minutes,5minutesor10minutes(TPA3,TPA5,TPA10, Results
andTPA3,TPA5,TPA10,respectively).Similarly,thedaily Theheight,bodyweight,bodymassindex,andwaist cir-
timespentfortheMVPAlastingatleast3minutes,5min- cumference were significantly different between males
utesand10minuteswereassessed(MVPA3,MVPA5,and and females (p<0.01, Table 1). As none of the acceler-
MVPA10, respectively). The reason for using these dura- ometer outputs significantly differed between genders,
tionswasasfollows:First,onlythe3minboutsofPAhad thePAdata areshownasthe combinedresultsformales
a significant impact on the post-prandial blood lipids and and females (Tables 2 and 3). Table 2 shows the time
Figure1Accelerometeroutputsasdeterminedusingthreedifferentepochlengths.EachillustrationshowstheMETvaluesfrom14:00to
14:30inoneoftheparticipants.Theupper,middleandlowerillustrationsindicatetheMETvaluesdeterminedbytheepochlengthsof4,20and
60seconds,respectively.Fortheepochlengthof4seconds,noPAboutslasting>10minutesweredetected.Fortheepochlengthof20seconds,
onePAboutlasting>10minuteswasdetected(14:03to14:15).Fortheepochlengthof60seconds,oneMVPAboutlasting>10minuteswas
detected(14:03to14:15).
Ayabeetal.BMCResearchNotes2013,6:20 Page4of7
http://www.biomedcentral.com/1756-0500/6/20
Figure2Differencesinthetimeandfrequencyofphysicalactivityboutscomparedwiththosedeterminedusinga4secepochlength.
TheY-axisshowsthedifferencesbetweenthetimeorthefrequencyspentonthephysicalactivityboutdeterminedbythe4secepochlength
andthatbythe20secor60secepochlength,respectively.TPA3,TPA5,TPA10;totalphysicalactivitylastinglongerthan3min,5minand
10min.MVPA3,MVPA5,MVPA10;moderatetovigorousintensityphysicalactivitylastinglongerthan3min,5minand10min.
spent in sporadic PA as determined by the three differ- Figure 2 shows the differences in the times and fre-
ent accelerometer epoch lengths. The LPA significantly quencies of the bouts of TPA and MVPA between those
increased with the increase in epoch length (p<0.05). determined using the 4-second epoch length and those
Theamount ofsporadicVPAdetermined bythe shortest determined using the 20- and 60-second epoch lengths.
epoch length (4 sec) was significantly longer compared The magnitudes of the differences in the times and
with the two remaining epoch lengths (p<0.05). Table 3 frequencies of the bouts of TPA determined using the
shows the time and frequency for the PA bouts as deter- 60-second epoch length were significantly larger than
mined using the three different accelerometer epoch those determined using the 20-second epoch length
lengths. Regardless of the intensities and bout durations, (p<0.001).Incontrast,themagnitudesofthedifferencesin
the frequency and the time for the PA bouts determined the times and frequencies of the bouts of MVPA did not
using the 4 sec epoch length were significantly shorter differsignificantly,exceptforthetimeoftheMVPA .
3M
compared with the other two settings (p<0.05). Add-
itionally,thefrequencyandthetimeforTPAdetermined
using the 20 sec epoch length were significantly smaller Discussion
compared with that determined using the 60 sec epoch Thepresentinvestigationis,tothebestofourknowledge,
length (p<0.05). the first study to examine the role of an accelerometer’s
epoch length (4 sec, 20 sec and 60 sec) in the evaluation
Table2Accelerometer’sepochlengthandthetimespent
ofPAboutswithregardtoaccelerometer-basedPAassess-
insporadicphysicalactivity
mentinmiddle-toolder-agedindividuals.Mainfindingof
Epochlength thepresentinvestigationisthat,regardlessoftheintensity
4sec 20sec 60sec and bout duration categories, the frequency and time for
LPA(min/day) 48.7±15.9 104.2±36.1** 178.7±62.6**,‡ the PA bouts were significantly lower when the shortest
MPA(min/day) 30.0±14.9 33.2±16.2 28.5±15.9*,† epoch length (4 sec) was used compared with the longer
VPA(min/day) 2.9±5.2 1.1±2.4** 0.9±2.5** epochlengths(20secand60sec).Furthermore,whenthe
MVPA(min/day) 32.9±16.5 34.3±17.1 29.4±16.9*,† analysis was limited to the TPA, the frequency and time
for the PA bouts were significantly lower as the epoch
Dataareexpressedasthemeanswiththestandarddeviation(mean±SD).
length increased. Although previous investigation exam-
LPA;lightintensityphysicalactivity,MPA;moderateintensityphysicalactivity,
VPA;vigorousintensityphysicalactivity,MVPA;moderatetovigorousintensity inedtheepochlengthonthe“sporadicMVPAestimation”
physicalactivity.
[7,8,10], the present investigation firstly showed that the
*,**Significantlydifferentcomparedwith4sec(*p<0.05,**p<0.01).†,
‡Significantlydifferentcomparedwith20sec(†p<0.05,‡p<0.01). epoch length affects the “MVPA bouts estimation”. These
Ayabeetal.BMCResearchNotes2013,6:20 Page5of7
http://www.biomedcentral.com/1756-0500/6/20
Table3Frequencyandtimeinboutsofphysicalactivityby3differentaccelerometer’sepochlengths
Epochlength
4sec 20sec 60sec
Frequency(bouts/day) TPA3 3.3±1.8 5.1±2.0** 9.0±3.3**,‡
TPA5 1.5±1.1 2.7±1.4** 5.6±2.1**,‡
TPA10 0.4±0.5 0.8±0.8** 2.2±1.1**,‡
MVPA3 1.1±1.3 1.8±1.3** 1.2±0.9**,‡
MVPA5 0.5±0.8 0.9±0.8** 0.7±0.7**
MVPA10 0.2±0.3 0.3±0.4** 0.3±0.4**
Time(min/day) TPA3 22.0±15.1 41.3±21.6** 101.8±37.6**,‡
TPA5 14.8±13.2 30.0±20.3** 80.2±33.2**,‡
TPA10 6.9±9.9 15.9±17.0** 47.7±27.8**,‡
MVPA3 7.7±10.0 14.3±12.8** 13.8±13.8**
MVPA5 5.2±8.1 10.3±11.2** 10.6±12.8**
MVPA10 2.6±5.4 5.4±8.8** 6.5±10.5**
Dataareexpressedasthemeanswiththestandarddeviation(mean±SD).TPA3,TPA5,andTPA10;Boutsofphysicalactivitylastinglongerthan3min,5minand
10min.MVPA3,MVPA5,andMVPA10;Boutsofmoderatetovigorousintensityphysicalactivitylastinglongerthan3min,5minand10min.
**Significantlydifferentcomparedwith4sec(**p<0.01).†,‡Significantlydifferentcomparedwith20sec(†p<0.05,‡p<0.01).
results indicate that the accelerometer’s epoch length with regard to the PA bouts, it is of interest that the rela-
affects the estimation of the PA bouts under free-living tionships between the accelerometer’s epoch length and
conditionsinmiddle-agedtoolderadults.The20-and60- the amount of PA were inverted (Table 3). The frequency
secondepochlengthsresultedinlargeramountsofMVPA and time for the PA bouts were significantly smaller for
bouts compared with the 4-second epoch length. Thus, theshorterepochlengthsettingcomparedwiththelonger
caution should be exercised regarding the accelerometer’s epoch length settings. The mean daily time for MVPA
10
data settingwhendiscussingthePAboutsreportedinpre- under the shortest epoch length setting was lessthan half
vious publications. Furthermore, the best epoch length to ofthatunderthelongerepochlengthsettings.
estimateMVPAboutsremainsunclear. Themain explanation for the differences in the amount
One of the original findings of the present investigation of PA among the three different epoch lengths is consid-
is that, with regard to the accelerometer-based PA assess- eredtobeduetotwodifferentfactors.First,theboutdur-
ment, a shorter epoch length results in smaller amounts ationofdailyPAisveryshortunderfree-livingconditions
of PA bouts compared with the longer epoch lengths [12,16,17]. Orendurff et al. previously indicated that 60%
under free-living conditions in middle aged to older of all walking bouts lasted less than 30 seconds, and 81%
adults. Although several previous investigations have of all walkingbouts lasted1 minuteor less, and 2 minute
examined the effects of an accelerometer’s epoch length walking bouts represented just 1% of all walking bouts
on the estimation of sporadic MVPA [6,8,20], the present [12]. Second, the shorter epoch has a higher sensitivity to
investigation provides the first demonstration that an detect the changes in the human PA, such as the transi-
accelerometer’s epoch length affects the estimation of the tion to PA from rest, and/or the switch from walking to
frequencyandtimeofPAbouts.Apreviousstudyshowed jogging, etc., compared with the longer epoch length.
thatashorterepochlengthwasmoresuitableforestimat- Underthe1minepochlength,forexample,ifpeoplewalk
ingthetime spent in sporadicMVPA, whereMVPAwere forfivesecondsthenrestfor55seconds,this1minwould
significantly higher and light-intensity activity was signifi- be categorized as LPA or MVPA rather than a resting
cantly lower when presented as 10 sec epochs compared period. Similarly, if people take a short break during con-
with that using a 60 sec epoch length [8]. Similar results tinuousjogging,thisminutewouldbecategorizedasMPA
arealsofoundinadolescentsandchildrenwhereasignifi- ratherthan VPA, regardless of the presence or absence of
cant epoch effect was seen for time spent in VPA, LPA, thebreak.Inotherwords,withregardtothe60secepoch
and rest in the child and adolescent samples and MVPA length,theintensityofPAislesslikelytoresultinaclassi-
in the child sample [10]. These findings are consistent fication of inactivity or VPA, because the various types of
with the results of the present investigation, where the PA, includingtheinactivity, areaveragedover1minperi-
daily time spent in sporadic MPA and VPA based on the ods. As a result, the accelerometer might detect a larger
shortestepochlength(4sec)weresignificantlylongerthan number of PA bouts. In contrast, a shorter epoch length
thatbasedonthelongerepochlengths(Table2).However, can more sensitively detect the changes in the type and
Ayabeetal.BMCResearchNotes2013,6:20 Page6of7
http://www.biomedcentral.com/1756-0500/6/20
intensity of PA, as well as the resting period (brief inter- intermittent PA, how to assess the activity level during
ruptions, etc.). Thereafter, the PA bouts were significantly theseshortinterruptions,andwhatlengthofinterruptions
smaller for the shorter epoch length compared with the canbeaccepted,etc.However,arecentpublication,which
longerepochlength. summarized accelerometer-based research, showed that
There is large variability in MVPA bouts lasting longer the longer epoch duration (1 minute) is the most com-
than 10 minutes. In the present investigation, the mean monly used epoch length [6]. In addition, the time for
MVPA lasting>10 minutes was 2.6±5.4 min day-1 under MVPAboutslasting>10minutesbasedonthe60-second
the 4-second epoch length. Similarly, the time for MVPA epoch length is also significantly associated with health
boutswas2.0mindayintheUSNHANESStudy[21],0.4 outcomes using this length [4,5]. Additionally, using a
to 0.6 min day-1 in British, Italian and French volunteers longer epoch length can reduce the data to be saved, so
[22] and 0.0 to 5.2 min day-1 in a Japanese study [23]. In the results usinga longer epoch length would be superior
contrast, some investigators have reported much longer with regard to data treatment, and would allow for a
MVPA bouts of>30 min day-1 in overweight and obese longer monitoring period by saving the data storage cap-
individuals[24],19.2±18.6minday-1inoverweightparti- acity. Therefore, each epoch length has both advantages
cipantsand25.8±23.4minday-1innormal-weightindivi- and disadvantages. Future research should first develop a
duals [5]. The differences in the amounts of MVPA bouts methodtoquantifythePAboutsonthebasisofadetailed
seem to exceed the acceptable range and are caused not analysis, thereafter, the contribution of the different PA
onlybytheepochlength,includinghuman-behavioralele- bouts to the health outcomes should be examined based
ments, but also by the data analysis process. In other on the quantitative assessment of the PA bouts based on
words, the procedures used to assess and quantify the PA themostvalidepochlength.
bouts must evolve in order to determine how to manage Thereareseverallimitationstothepresentinvestigation
briefinterruptions(i.e.breaks of<1minute) ofPA.Inthe that should be considered when interpreting the results.
present investigation, all MVPA bouts satisfied the inclu- First,thetype(uni-axial)andposition(waist)oftheaccel-
sioncriteria,wheretheintensityofPAwasabove3METs erometer used in the present investigation does not allow
for that lasting at least 10 minutes. In contrast, some for the collection of upper body activity, and thus may
investigators have attempted to improve the sensitivity of underestimatethe total PA. Additionally,astheLC hasan
detectingMVPAbouts.Forexample,toallowforthebrief originalalgorithmforPAintensityclassification,theresults
periodsofrestcommonduringcertainactivities,i.e.paus- of the present investigation should be confirmed using
ing for a water break while playing basketball, the criteria other activity monitors. Second, the participants evaluated
usedtodefineaboutrequiresarunningaverageof70%of in the present investigation were primarily middle-aged,
the counts to be above the cut-off point [21]. Similarly, non-active males and females. Furthermore, the partici-
interruptionsof1or2minutesbelow3METsareallowed pants all lived in urban areas, and buses and trains were
during consecutive MVPA lasting>10 minutes to define theirprimarymeansoftransportation.Thus,itislimitedto
bouts of MVPA [25]. We cannot conclusively define the generalize the present investigation to the heterogeneous
best epoch length to estimate PA bouts because we were individuals, therefore, the results of the present investiga-
unable to compare the accelerometer’s outputs with the tion should be confirmed in active individuals and hetero-
“goldstandardmethodforassessingPAbouts.”Becauseit geneousvolunteers,includingyoungchildren.
is difficult to assess the bout duration of PA under free-
living conditions, using video tape recordings and/or dir- Conclusion
ect observation is the gold standard method for assessing In summary, the present investigation was the first to
PAboutsinexperimentalstudies. examine the role of the accelerometer’s epoch length on
However,thepreviousandpresentinvestigationsappar- the evaluation of the PA bouts with regard to the
entlyindicatethatashorterepochlengthmaybesuperior accelerometer-based PA assessment. The frequency and
to estimate the PA bouts compared to a longer epoch time for the PA bouts were significantly lower as the
length [8,26]. Furthermore, recent publications showed epoch length increased. These results indicate that the
that, although PA bouts are relatively rare under free- accelerometer’s epoch length affects the estimation of the
living conditions, the MVPA lasting>1 min was signifi- PA bouts under free-living conditions in middle-aged to
cantly associated with the blood lipid status and age in older adults. Thus, caution should be exercised when
previous studies [16,17]. In contrast, with regard to the selecting the accelerometer’s data setting, and when dis-
1 min epoch length, it is doubtful whether it can evaluate cussingtheresultsofpreviousstudiesaboutthePAbouts.
“real” PA bouts under free-living conditions, due to the The best epoch length to estimate MVPA bouts remains
lack of accurate treatment of the interruptions in activity. unclear, and will need to be assessed in future studies. In
Further investigations will be required to determine how ordertogeneralizetheresultsofthepresentinvestigation,
to divide the continuous PA with interruptions and the future studies should examine the effects of the epoch
Ayabeetal.BMCResearchNotes2013,6:20 Page7of7
http://www.biomedcentral.com/1756-0500/6/20
length in heterogeneous volunteers using several types of 15. ShephardRJ,AoyagiY:Measurementofhumanenergyexpenditure,with
activitymonitors. particularreferencetofieldstudies:anhistoricalperspective.EurJAppl
Physiol2011,112:2785–2815.
16. AyabeM,KumaharaH,MorimuraK,IshiiK,SakaneN,TanakaH:Veryshort
Competinginterests
boutsofnon-exercisephysicalactivityassociatedwithmetabolic
Noneoftheauthorshaveanyprofessionalrelationshipwithcompaniesor
syndromeunderfree-livingconditionsinJapanesefemaleadults.EurJ
manufacturesthatwillbenefitfromtheresultsofthepresentstudy.
ApplPhysiol2012,112(10):3525–3532.
Authors’contributions 17. AyabeM,AokiJ,KumaharaH,TanakaH:Age-relateddifferencesindaily
physicalactivitydividedbyboutduration:Preliminaryfindingsinfemale
Planning:MA,HT,Conducting:MA,HK,KM,Datatreatment:MA,KM, conveniencesamples.JSportsSci2012,30(7):709–713.
Reporting:MA,HK,HT.Allauthorsreadandapprovedthefinalmanuscript.
18. MiyashitaM,BurnsSF,StenselDJ:Accumulatingshortboutsofbrisk
walkingreducespostprandialplasmatriacylglycerolconcentrationsand
Acknowledgement
restingbloodpressureinhealthyyoungmen.AmJClinNutr2008,
PartofthisstudywassupportedbyFukuokaUniversity(GlobalFUprogram, 88(5):1225–1231.
andFukuokaUniversityInstituteforPhysicalActivity).
19. AyabeM,KumaharaH,MorimuraK,IshiiK,SakaneN,TanakaH:
Accumulationofshortboutsofnon-exercisedailyphysicalactivityis
Authordetails
associatedwithlowervisceralfatinJapanesefemaleadults.IntJSports
1FacultyofHealthandSportsScience,FukuokaUniversity,8-19-1Nanakuma, Med2013,34:62–67.
Jonan-ku,Fukuoka814-0180,Japan.2FacultyofNutritionalSciences,
20. HeilDP,BennettGG,BondKS,WebsterMD,WolinKY:Influenceofactivity
NakamuraGakuenUniversity,Fukuoka,Japan.
monitorlocationandboutdurationonfree-livingphysicalactivity.ResQ
ExercSport2009,80(3):424–433.
Received:5October2012Accepted:8January2013
21. MetzgerJS,CatellierDJ,EvensonKR,TreuthMS,RosamondWD,Siega-Riz
Published:18January2013
AM:PatternsofobjectivelymeasuredphysicalactivityintheUnited
States.MedSciSportsExerc2008,40(4):630–638.
References
22. DavisMG,FoxKR:Physicalactivitypatternsassessedbyaccelerometryin
1. HaskellWL,LeeIM,PateRR,PowellKE,BlairSN,FranklinBA,MaceraCA, olderpeople.EurJApplPhysiol2007,100(5):581–589.
HeathGW,ThompsonPD,BaumanA:Physicalactivityandpublichealth:
23. AyabeM,AokiJ,KumaharaH,TanakaH:Durationandfrequencyofdaily
updatedrecommendationforadultsfromtheAmericanCollegeof
physicalactivityandachievementofexerciseandphysicalactivity
SportsMedicineandtheAmericanHeartAssociation.MedSciSportsExerc
referenceforhealthpromotion2006.JapJPhysFitnessandSportsMed
2007,39(8):1423–1434. 2008,57(5):577–586.
2. SarisWH,BlairSN,vanBaakMA,EatonSB,DaviesPS,DiPietroL,Fogelholm
24. YoungDR,JeromeGJ,ChenC,LaferriereD,VollmerWM:Patternsof
M,RissanenA,SchoellerD,SwinburnB,etal:Howmuchphysicalactivity
physicalactivityamongoverweightandobeseadults.PrevChronicDis
isenoughtopreventunhealthyweightgain?OutcomeoftheIASO1st
2009,6(3):A90.
StockConferenceandconsensusstatement.ObesRev2003,4(2):101–114.
25. HagstromerM,TroianoRP,SjostromM,BerriganD:Levelsandpatternsof
3. Ishikawa-TakataK,TabataI:Exerciseandphysicalactivityreferencefor objectivelyassessedphysicalactivity–acomparisonbetweenSweden
healthpromotion2006(EPAR2006).JEpidemiol2007,17(5):177. andtheUnitedStates.AmJEpidemiol2010,171(10):1055–1064.
4. StrathSJ,HollemanRG,RonisDL,SwartzAM,RichardsonCR:Objective
26. McClainJJ,AbrahamTL,BrusseauTAJr,Tudor-LockeC:Epochlengthand
physicalactivityaccumulationinboutsandnonboutsandrelationto
accelerometeroutputsinchildren:comparisontodirectobservation.
markersofobesityinUSadults.PrevChronicDis2008,5(4):A131. MedSciSportsExerc2008,40(12):2080–2087.
5. CatenacciVA,GrunwaldGK,IngebrigtsenJP,JakicicJM,McDermottMD,
PhelanS,WingRR,HillJO,WyattHR:PhysicalActivityPatternsUsing
doi:10.1186/1756-0500-6-20
AccelerometryintheNationalWeightControlRegistry.Obesity
Citethisarticleas:Ayabeetal.:Epochlengthandthephysicalactivity
(SilverSpring)2011,6(19):1163–1170.
boutanalysis:Anaccelerometryresearchissue.BMCResearchNotes2013
6. HeilDP,BrageS,RothneyMP:ModelingPhysicalActivityOutcomesfrom 6:20.
WearableMonitors.MedSciSportsExerc2012,44(1S):S50–S60.
7. TrostSG,McIverKL,PateRR:Conductingaccelerometer-basedactivity
assessmentsinfield-basedresearch.MedSciSportsExerc2005,
37(11Suppl):S531–S543.
8. GabrielKP,McClainJJ,SchmidKK,StortiKL,HighRR,UnderwoodDA,Kuller
LH,KriskaAM:Issuesinaccelerometermethodology:theroleofepoch
lengthonestimatesofphysicalactivityandrelationshipswithhealth
outcomesinoverweight,post-menopausalwomen.IntJBehavNutrPhys
Act2010,7:53.
9. KumaharaH,SchutzY,AyabeM,YoshiokaM,YoshitakeY,ShindoM,IshiiK,
TanakaH:Theuseofuniaxialaccelerometryfortheassessmentof
physical-activity-relatedenergyexpenditure:avalidationstudyagainst
whole-bodyindirectcalorimetry.BrJNutr2004,91(2):235–243.
10. EdwardsonCL,GorelyT:Epochlengthanditseffectonphysicalactivity
intensity.MedSciSportsExerc2010,42(5):928–934. Submit your next manuscript to BioMed Central
11. AyabeM,KumaharaH,AokiJ,TanakaH:Assessmentoffrequencyand and take full advantage of:
boutsdurationofmoderateintensityphysicalactivitybyan
accelerometerin4-secintervalunder-freelivingcondition.Journalof
JapanSocietyfortheStudyofObesity2007,13(2):197–200. • Convenient online submission
12. OrendurffMS,SchoenJA,BernatzGC,SegalAD,KluteGK:Howhumans • Thorough peer review
walk:boutduration,stepsperbout,andrestduration.JRehabilResDev • No space constraints or color figure charges
2008,45(7):1077–1089.
• Immediate publication on acceptance
13. AyabeM,AokiJ,IshiiK,TakayamaK,TanakaH:Pedometeraccuracyduring
stairclimbingandbenchsteppingexercises.JSportsSci&Med2008, • Inclusion in PubMed, CAS, Scopus and Google Scholar
7(2):249–254.
• Research which is freely available for redistribution
14. AyabeM,IshiiK,TakayamaK,AokiJ,TanakaH:Comparisonofinterdevice
measurementdifferenceofpedometersinyoungerandolderadults.
BrJSportsMed2010,44(2):95–99. Submit your manuscript at
www.biomedcentral.com/submit