www.jped.com.br
ORIGINAL
ARTICLE
Predictive
capacity
of
different
bioelectrical
impedance
analysis
devices,
with
and
without
protocol,
in
the
evaluation
of
adolescents
夽
Vivian
Siqueira
Santos
Gonc
¸alves
a,∗,
Eliane
Rodrigues
de
Faria
b,
Sylvia
do
Carmo
Castro
Franceschini
a,
Silvia
Eloiza
Priore
aaDepartmentofNutritionandHealth,UniversidadeFederaldeVic¸osa,Vic¸osa,MG,Brazil bUniversidadeFederaldoEspíritoSanto,Alegre,ES,Brazil
Received22October2012;accepted13March2013 Availableonline13September2013
KEYWORDS Adolescenthealth; Bodycomposition; Electricimpedance; Obesity Abstract
Objective: thisstudywasperformedtodeterminethepredictivecapacityoffourdifferent bio-electricalimpedanceanalysis(BIA)devicesintheassessmentofadolescents,withandwithout aprotocol.
Methods: across-sectional study wasperformed with215adolescents aged10 to14 years, ofbothgenders,evaluated throughanthropometryandbodycompositionbydual energy X-rayabsorptiometry(DXA)andbyfourdifferentBIAdevices,withandwithoutaprotocol.The followingtestswereused:Kolmogorov-Smirnov’s,chi-squared,Student’storMann-Whitney’s, Kruskal-Wallis’s,Wilcoxon’s,andkappaindex.TheROCcurveswereconstructedandthe sen-sitivity,specificity,andpositiveandnegativepredictivevalueswerecalculated.
Results: of the 215 adolescents, 44.2% had excessive body fat. The tetrapolar BIA device equippedwitheighttactile electrodesshowed moresensitivityandresultsthatwerecloser tothoseobtainedbyDXA(areaundertheROCcurve[AUC]=0.964withprotocolandAUC= 0.973withoutprotocol,p<0.001),aswellasgreateragreement(k=0.67withprotocoland k=0.63withoutprotocol,p<0.001).Theevaluationwithoutprotocolwassimilartothatby DXAinmostinvestigatedsituations(p>0.05).
夽 Pleasecitethisarticleas:Gonc¸alvesVS,FariaER,FranceschiniSC,PrioreSE.Predictivecapacityofdifferentbioelectricalimpedance
analysisdevices,withandwithoutprotocol,intheevaluationofadolescents.JPediatr(RioJ).2013;89:567---74.
∗Correspondingauthor.
E-mail:vivian.goncalves@ufv.br(V.S.S.Gonc¸alves).
0021-7557©2013SociedadeBrasileiradePediatria.PublishedbyElsevierEditoraLtda. http://dx.doi.org/10.1016/j.jped.2013.03.023
Conclusion: BIAiscapableofpredictingalterationsinadolescents’bodycomposition.Whenit isimpossibletoperformtheassessmentwithaprotocol,itsresultsmaybeusefulinpopulation studies.
©2013SociedadeBrasileiradePediatria.PublishedbyElsevierEditoraLtda.
PALAVRAS-CHAVE
Saúdedo adolescente;
Composic¸ãocorporal; Impedânciaelétrica; Obesidade
Capacidadepreditivadediferentesequipamentosdebioimpedânciaelétrica,come sempreparoprévio,naavaliac¸ãodeadolescentes
Resumo
Objetivo: determinaracapacidadepreditivadequatroequipamentosdistintosde bioimpedân-ciaelétrica(BIA)naavaliac¸ãodeadolescentes,comesemarealizac¸ãodeprotocolo. Métodos: estudo transversalrealizado com215 adolescentesde10 a14anos, deambos os sexos,avaliadosatravésdaantropometriaedacomposic¸ãocorporalpeloDEXAeporquatro equipamentosdistintosdeBIA,comesemprotocolo.Foramutilizadosostestesestatísticos: Kolmogorov-Smirnov,doQui-quadrado,t-StudentouMann-Whitney,Kruskal-Wallis,Wilcoxon eÍndiceKappa.ForamconstruídascurvasROCecalculadososvaloresdesensibilidade, especi-ficidadeepreditivospositivoenegativo.
Resultados: dosadolescentes,44,2%apresentaramexcessodegorduracorporal.ABIA tetrap-olar,equipadacomoitoeletrodostáteis,demonstrou-semaissensívelecomresultadosmais próximosaoDEXA(AUC=0,964comprotocoloeAUC=0,973semprotocolo,p<0,001), apre-sentando,também,maiorconcordância(k=0,67comprotocolo,ek=0,63semprotocolo,p< 0,001).Aavaliac¸ãosemprotocolofoisemelhanteaoDEXAnamaioriadassituac¸õesinvestigadas (p>0,05).
Conclusão: aBIAéuminstrumentocapazdepredizerdistrofiasrelacionadasàgorduracorporal deadolescentes.Naimpossibilidadederealizac¸ãodoprotocolo,seusresultadospodemserúteis emestudospopulacionais.
©2013SociedadeBrasileiradePediatria.PublicadoporElsevierEditoraLtda.
Introduction
TheWorldHealthOrganization (WHO)definesadolescence astheperiodfrom10to19years,characterizedbyintense physical, psychological, and social changes. Rapidgrowth and nutritionalvulnerability also characterize this phase, whenthereisconsolidationof eatinghabits, which,when adequate,canbecomeaprotectivefactorforobesity, car-diovasculardisease,andmetabolicdisordersinadulthood.1
According to the household budget survey (HBS) con-ductedbytheBrazilianInstituteofGeographyandStatistics (InstitutoBrasileirodeGeografiaeEstatística-IBGE)inthe years 2008 and 2009, excess body weight was identified inapproximately 20%of the adolescent populationof the metropolitanareasofBrazil.2
When weightincreases due toexcess body fat, it can leadtoadolescentobesity,whichhasbeenconsidereda pre-dictorofriskforcardiovasculardisease,diabetesmellitus, dyslipidemia,andhypertension.Therefore,itisrelevantto adequatelyestimatebodycompositioninadolescence,asit isaperiodofgreatchange,mainlyasaresultofpuberty.3---6
Itmustbeemphasized,however,thatthereisaneedfor methodsusedinbodycompositiondeterminationthat are practical,fast,andeasytoperform,withthepossibilityof beingapplicabletoseveralworkingconditions,includingin population-basedstudies in the field.Among these meth-ods,bioelectrical impedance analysis(BIA) is highlighted, asithasallthesecharacteristicsatarelativelylowcost,in additiontoitsportabilityandnoninvasiveness.4,5,7---9
The use of previous preparations (protocols) for stan-dardization of variables that affect body hydration is a recommendationtoperformBIA.9---11 However,itsusemay
be restricted by lack of adherence or difficulty tofollow theserequirementsbytheadolescent.
Given the importance of accurately determining body compositionandthebroaduseofBIA,thisstudyaimedto determinethepredictivecapacityoffourdifferentdevices intheevaluationofadolescentswithandwithoutaprotocol.
Methods
Sample
This was an epidemiological, cross-sectional study, with a population of 215 adolescents of both genders, aged between10yearsto14years,11months,selectedbysimple randomsamplingfromallpublicandprivateschoolsinthe agerangeofinterest,locatedinurbanandruralareasofthe city ofVic¸osa,stateof MinasGerais,Brazil.The following inclusioncriteriawereused:interestinparticipatinginthe study;absenceofprostheticsand/orpacemakers;absence of chronic diseases or use of continuous medication that couldinterferewithbodyhydration;andadherencetothe recommendedprotocoltoundergoBIA.
Sampleselectionwasbasedonthetotalnumberof ado-lescents in the city at the age of interest in 2010.12 The
samplewascalculatedusingEpiInfosoftware,release6.04 forcross-sectionalstudies,consideringatotalpopulationof
Este é um artigo Open Access sob a licença de CC BY-NC-ND
5,754individuals,theexpectedfrequencyofexcessbodyfat of17.5%,13 andvariabilityof5%,totaling214individuals,
withaconfidencelevelof95%.
Thesampledrawwasconductedamongallwhometthe inclusioncriteriaandreturnedthesignedinformedconsent, respecting the proportionality of the number of students thateachschoolhadineachagegroup.Whenthe adoles-centdidnotwanttoparticipateorabandonedthestudy,a newdrawwasmadetoreplacehim/her.
The project wasapprovedby the EthicsCommitteeon HumanResearchofUniversidadeFederaldeVic¸osa (proto-col.N.0140/2010);adolescentsandtheirparentssignedthe consentform,preparedinaccordancewithstandards estab-lishedby196/96ResolutionoftheNationalHealthCouncil.
Anthropometric
assessment
Weight was measured on a digital scale witha maximum capacityof150kgandasensitivityof50g,whereasheight wasmeasuredusingaportablestadiometerwithan exten-sionof 2.13m and0.1cmresolution.Measurements were made in duplicate, allowing the use of the mean values betweenthetwomeasurements.Incaseswherethe differ-enceexceeded0.5cm,newmeasurementswereperformed. Thebodymassindex(BMI)/ageandheight/ageindiceswere calculated tocharacterize the population, usingas refer-encethecutoffs(Z-scores),establishedbytheWHO.14
Body
composition
The body fat percentage (BF%) was measured by a DXA equipment (Lunar Prodigy Advance DXA System - analysis version: 13,31, GE Healthcare) and estimated by the fol-lowingBIAequipment:tetrapolarhorizontalBiodynamics®,
model 450 (BIA1), tetrapolar verticalTanita®,model
BC-558 (BIA 2), tetrapolar vertical Biospace®, equipped with
eight tactile electrodes, InBody® 230 model (BIA 3), and
bipolarverticalTanita®,model2220(BIA4).BF%was
ana-lyzedaccordingtotheclassificationproposedbyLohman15,
consideringasexcessbodyfatvalues≥20%formalesand≥ 25%forfemales,andaslowbodyfatpercentage<10%for malesand <15% forfemales. Assessment byBIA was per-formed in twostages, following theprotocol proposedby Barbosa16 andlater,withinanaverage timeperiodof9±4
days,withouttheprotocol.
The protocol involved previous preparation aiming to standardizethehydrationstatustoundergotheBIA assess-mentandconsistedofthefollowing:beatleastsevendays afterthelastmenstrualperiodandsevendaysbeforethe next; undergo complete fasting in the previous 12hours; refrain from physical exercises in the previous 12hours; noalcoholconsumptionintheprevious48hours;nouseof diureticsforatleastsevendaysbeforetheassessment;and urination 30minutes before the assessment. Adolescents werealsoaskedtoremovemetalobjectssuchasearrings, rings,watches,andothers,whichcouldinterferewiththe passageofelectricalcurrent.
Statistical
analysis
The Kolmogorov-Smirnovnormality test wasperformed to determinevariabledistribution(parametricornot)andthus
choose the most appropriate statistical test to evaluate data.Parameterswithnormaldistributionwereexpressed asmeanandstandarddeviation;thosewithnon-normal dis-tributionwereexpressedasmedianandrange.
Thechi-squaredtestwasusedtocompareprevalence.In caseofcontinuousvariables,Student’storMann-Whitney’s testswereperformedtocomparetwo,andKruskal-Wallis’s test wasused tocompare threeor more;Wilcoxon’s test wasusedwhentwoevaluationswereperformedbythesame individualsusingthesameequipment(withandwithout pro-tocol).
Thekappaindexwasusedtodeterminetheagreement between the assessments by BIA and DXA, classifying it according to the criteria by Landis and Koch17, with the
followingconcordances:from0to0.19:poor;0.2to0.39: weak;0.4-0.59:moderate;0.6to0.79:substantial;0.8to 1.0:almostperfect.
ROCcurveswereconstructedtoverifythecapacityofBIA inpredictingexcessbodyfatwhencomparedtoDXA.Areas underthecurve (AUC)werecalculatedwiththeir respec-tive 95% confidence intervals. The null hypothesis would beaccepted withan AUCvalue ≤ 0.50.Sensitivity (prob-abilityof atest beingpositive, if thereis an alteration), specificity (probability of a test being negative, if there isnoalteration),positive predictive values(proportionof true positivesamong all individuals who tested positive), andnegativepredictivevalues(proportionoftruenegatives amongallindividualswhotestednegative)werecalculated foreachdevice,withandwithoutprotocol;excessbodyfat wasconsideredasthealteredvariable.18
The database was created usingMicrosoft OfficeExcel 2007withduplicateentries;statistical analyseswere per-formed using the SIGMA STAT software, release 3.2 and MEDCALCstatisticalsoftware,release12.2.1.0.Significance levelwassetatp<0.05.
Results
Subjects’characteristics
Atotalof 215adolescents(ofwhom53.5%,n=115,were females),participatedinthestudy,presenting the follow-ing median values for age, weight, and BMI: 11.9 years (range: 10.1 to 14.9), 42.2kg (range: 25.1 to 92.8), and 18.0kg/m2 (range: 12.5 to 33.8). The mean height was
151.6±10.0cm. There was no difference regarding these parametersbetweengenders(p>0.05).
Regardingthenutritionalstatus,2.8%(n= 6)hadshort statureforage,and3.3%(n=7)hadlowBMIforage,16.7% (n=36)wereoverweight,and8.4%(n=18)wereobese.
DXAassessmentshowedaprevalenceof44.2%(n=95)of excessBF%and13.5%(n=29)oflowBF%.
Comparisonofassessmentswithandwithout protocol
Table1showstheprevalenceoflowBF%,normalBF%,and high BF% measured by DXA and estimated by BIA, with andwithoutprotocol. Itwasobserved thattheevaluation carried out by all BIA devices with a protocol identified more adolescents with high BF% than without protocol.
Table1 NutritionalstatusbypercentageofbodyfatinadolescentsassessedbyDXAandfourbioelectricalimpedancedevices, withandwithoutprotocol.
Device Nutritionalstatusbypercentageofbodyfat
LowBF% NormalBF% HighBF%
% 95%CI % 95%CI % 95%CI DXA 13.5 9.6-18.7 42.3 35.9-49.0 44.2 37.7-50.1 BIA1 Withprotocol 6.5a 3.9-10.6 55.8a 49.1-62.3 37.7 31.5-44.3 Withoutprotocol 9.3 6.1-13.9 54.0a 47.3-60.5 36.7 30.6-43.4 BIA2 Withprotocol 0.0a 0.0-1.8 46.5 40.0-53.2 53.5 46.8-60.0 Withoutprotocol 0.0a 0.0-1.8 53.5a 46.8-60.0 46.5 40.0-53.2 BIA3 WithProtocol 9.3 6.1-13.9 50.2 43.6-56.9 40.5 34.1-47.1 Withoutprotocol 8.8 5.7-13.4 52.6a 45.9-59.1 38.6 32.4-45.3 BIA4 WithProtocol 6.5a 3.9-10.6 59.5a 52.9-65.9 34.0a 28.0-40.5 Withoutprotocol 10.7 7.2-15.5 56.3a 49.6-62.7 33.0a 27.1-39.6
BIA1,biodynamicshorizontaltetrapolar device,model450;BIA2,tanitaverticaltetrapolardevice,modelBC-558;BIA3,biospace
equippedwitheighttactileelectrodes,modelInBody230;BIA4,Tanitaverticalbipolardevice,model2220;BF%,percentageofbody fat;CI,confidenceinterval;DXA,dual-energyX-rayabsorptiometry.
Chi-squaredtest:BIAvs.DXA.
ap<0.05.
Regardingthe increase in BF%,BIA 4 wasthe only device that underestimated the prevalence (p < 0.05), whereas the others were similar to DXA in both assessments (p > 0.05). It is noteworthy that BIA 3 showed prevalence moresimilartoDXAin allsituations(p >0.05) exceptfor normal BF% without protocol, where it overestimated it (p<0.05).
When compared,all BIA devices hadsimilarvalues for body fat in kg (BF) when compared to DXA in both with and without protocol assessments (p > 0.05), considering thetotalpopulation.Regardingthestratificationbygender, onlythe malegender, assessed byBIA 2, washigherthan DXA(p=0.011withprotocolandp=0.017without proto-col).The comparisonofBIAdevicesbygenderis shownin
Table2.
Itwasobserved that,forfemales,the protocoldidnot influenceanyoftheassessments,whereasformales,BIA2 and3alsoshowedsimilarvaluesinbothsituations(p>0.5). Itisnoteworthy,however,thatBIA3didnotdifferfromDXA, contrarytothatoccurredwithBIA2,whichoverestimated BFinbothsituations(p<0.05).
AgreementbetweentheassessmentsbyBIAandby DXA
When analyzing the agreement between DXA evaluations andbyeachoftheBIAdevices(Table3),significancewere observedamongallofthem(p<0.001).However,BIA3again
Table 2 Body fat determined by different electrical bioimpedance devices, with and without protocol, by gender, in
adolescents.
BF(kg) Female(n=115) p Male(n=100) p
Withprotocol Withoutprotocol Withprotocol Withoutprotocol
Md(Range) Md(Range) Md(Range) Md(Range)
BIA1 9.4(3.2-35.0) 9.6(2.4-23.6) 0.385 6.9(2.4-28.3) 6.9(2.2-28.2) 0.027
BIA2 10.8(5.6-26.7) 11.1(5.5-24.6) 0.173 7.4(2.9-31.6) 7.1(3.2-30.8) 0.259
BIA3 9.5(2.7-30.4) 9.6(3.0-30.7) 0.252 6.7(1.6-33.8) 6.7(1.3-35.1) 0.599
BIA4 9.8(2.3-27.8) 9.9(2.4-28.0) 0.512 6.2(1.0-31.8) 5.7(1.9-30) <0.001
BIA1,biodynamicshorizontaltetrapolar device,model450;BIA2,tanitaverticaltetrapolardevice,modelBC-558;BIA3,biospace equippedwitheighttactileelectrodes,modelInBody230;BIA4,tanitaverticalbipolardevice,model2220;BF%,bodyfat;Md,median. Wilcoxontest(BIAwithvs.BIAwithoutprotocol).
Table3 Agreementbetweenbodyfatassessmentin ado-lescents performed by different bioelectrical impedance devicesandDXA.
Kappaindex(n=215) BIA1 Withprotocol Total 0.52a Female 0.52a Male 0.50a Withoutprotocol Total 0.59a Female 0.54a Male 0.64a BIA2 Withprotocol Total 0.49a Female 0.38a Male 0.57a Withoutprotocol Total 0.51a Female 0.46a Male 0.53a BIA3 Withprotocol Total 0.67a Female 0.62a Male 0.70a Withoutprotocol Total 0.62a Female 0.58a Male 0.65a BIA4 Withprotocol Total 0.53a Female 0.46a Male 0.62a Withoutprotocol Total 0.52a Female 0.43a Male 0.61a
BIA 1, biodynamics horizontal tetrapolar device, model 450; BIA2,tanitaverticaltetrapolardevice, modelBC-558;BIA3, biospaceequippedwitheighttactileelectrodes,modelInBody 230;BIA4,tanitaverticalbipolardevice,model2220; Kappa index:BIAvs.DXA.
a p<0.001.
showedbetterresults,withastrongagreementforthetwo assessmentsinbothgenders.
TheotherBIAdevicesalsoshowedstrongagreementfor themalegender.Forthetotal populationandthe female gender,agreementwasconsideredmoderate.
BIA 2 presented the worst results; the female gender, withprotocol,wasconsideredtheweakestamongall ana-lyzeddata.
Predictivecapacityofelectricalbioimpedance, withandwithoutprotocol
Table4showsthe AUC,sensitivity, andspecificity,aswell aspositive andnegativepredictive valuesforeach device and gender, at the assessments with and without proto-col, obtained after creating the ROC curves, considering theexcess BF. Itwas observedthat BIA 3, without proto-col,showedthehighest areasforthetotalpopulationand forbothgenders,afterstratification.
TheAUCs werealsocomparedforevaluationswithand withoutprotocol,performedforeach genderandforeach deviceandnodifferenceswereobserved(p>0.05).
Discussion
Thenutritionalstatus oftheadolescentsstudied followed the trendindicated by the HBS, witha low frequency of malnutritionandhigherprevalenceofoverweight. Approxi-mately25%oftheadolescentsinthecitywereoverweight, higherthanthenationalprevalence(20.5%)andthatfound inanotherstudywithadolescentsfromthestate ofMinas Gerais (20.1%), but within the range found in the South-eastern region of Brazil (20% to 27%).2,19 Comparing the
prevalenceofalterationsinBF%(43%)andBMI/age(25%),it isclearthattheindexfailedtoidentifyseveraladolescents whoalready hadthese alterations, confirming the impor-tanceofmethods topredict excess BFeveninthose who presentnormalweight.
ItisworthmentioningthattheBMIcriteriashouldnotbe usedalone.Adolescentswithanadequate BMImayhavea highBF%andmayeventuallyhaverisksofmorbiditysimilar tothosewithhighBMI,especiallyinfemales20---24,
highlight-ingtheneedforBF%assessmentinordertoidentifypossible riskfactorsforhealth.
ExcessBFmayberelatedtogenetic,metabolic, physio-logical,andlifestylecomponents,suchassedentarylifestyle andpoor eatinghabits. Itis associated withinsulin resis-tance,dyslipidemia,andmetabolicsyndrome,someofthe risk factors for cardiovascular disease that have already beenidentifiedinadolescentsattheagegroupanalyzedin thisstudy,corroboratingtheimportanceofmonitoringthese youngindividuals.20,21,23,25,26
BIA deviceswereable topredict increasesin BF%,but showeddistinctcharacteristicswhentheprotocolinfluence wasanalyzed.Ingeneral,whencomparedwithDXA,itwas observedthat bothassessmentsbehaved similarlyin rela-tiontotheprevalenceofexcessBF,andonlyBIA4showed a different result (p < 0.05). BIA 3 was shown to be the moststabledevice,differingfromDXAinonlyonesituation (normalBFwithoutprotocol),butsimilartoDXAregarding theprevalenceofexcessorlowBF%,withandwithoutthe protocol.Therefore,for aprevalence study,BIA3appears tobe themost adequate device, while BIA 4 is the least recommended.
BIA1,3,and4weresimilartoDXAwhencomparingBFin kgforbothgendersatthetwoassessments,indicatingthey areadequate for use. BIA 2, when evaluating thefemale gender,couldalsobeconsidereduseful.Thesedevicesare portable,easy touseandtransport,anddespitetheprice variationamongthem,aremuchmoreaffordablethanDXA.
Table4 PredictivecapacityofexcessbodyfatindifferentelectricalbioimpedancedevicescomparedtoDXA,bygender,in theevaluationofadolescents,2011.
AUC SD 95%CI Sensit% Specif% PPV% NPV%
BIA1 Withprotocol Total 0.938a 0.0144 0.897-0.967 91.6 79.2 77.7 92.2 Female 0.926a 0.0220 0.862-0.966 72.1 96.3 95.7 75.4 Male 0.957a 0.0173 0.897-0.987 91.2 87.9 79.5 95.1 Withoutprotocol Total 0.946a 0.0142 0.907-0.972 85.3 92.5 90.0 88.8 Female 0.931a 0.0226 0.869-0.970 91.8 83.3 86.2 90.0 Male 0.966a 0.0172 0.909-0.992 91.2 92.4 86.1 95.3 BIA2 Withprotocol Total 0.916a 0.018 0.871-0.950 89.5 78.3 76.6 90.4 Female 0.919a 0.0239 0.853-0.962 75.4 94.4 93.9 77.3 Male 0.980a 0.0101 0.929-0.997 100.0 86.4 79.1 100.0 Withoutprotocol Total 0.894a 0.0215 0.845-0.932 80.0 85.3 81.7 84.4 Female 0.895a 0.0286 0.824-0.944 88.5 77.8 81.8 85.7 Male 0.945a 0.0266 0.880-0.981 85.3 92.4 85.3 92.4 BIA3 Withprotocol Total 0.964a 0.0129 0.929-0.984 96.8 86.7 85.2 97.2 Female 0.951a 0.0237 0.894-0.983 98.4 87.1 89.6 97.9 Male 0.968a 0.0144 0.917-0.992 93.4 92.6 93.4 92.6 Withoutprotocol Total 0.973a 0.0086 0.942-0.991 88.4 95.8 94.4 91.3 Female 0.984a 0.0100 0.935-0.999 94.1 96.9 94.1 97.0 Male 0.986a 0.0077 0.939-0.999 97.6 92.4 86.8 98.4 BIA4 Withprotocol Total 0.932a 0.0156 0.890-0.962 83.2 88.3 84.9 86.9 Female 0.907a 0.0261 0.838-0.953 88.5 75.9 80.6 85.4 Male 0.970a 0.0134 0.914-0.994 97.1 81.2 73.3 98.2 Withoutprotocol Total 0.929a 0.0161 0.886-0.959 85.3 85.9 82.7 88.0 Female 0.899a 0.0277 0.829-0.947 75.4 88.9 88.5 76.2 Male 0.980a 0.0106 0.930-0.998 91.2 95.5 91.2 95.5
AUC,areaunderthecurve;BIA1,biodynamicshorizontaltetrapolardevice,model450;BIA2,tanitaverticaltetrapolardevice,model BC-558;BIA3,biospaceequippedwitheighttactileelectrodes,modelInBody230;BIA4,tanitaverticalbipolardevice,model2220;CI, confidenceinterval;NPV,negativepredictivevalue;PPV,positivepredictivevalue;SD,standarddeviation;Sensit,Sensitivity;Specif, specificity.
ROCcurve:BIAvs.DXA.
ap<0.001.
Formales,however,BIA2wasnotshowntobeanaccurate option.
The resultsshowed thatthe protocoldid notinfluence resultsinfemaleadolescents,which,togetherwiththe sim-ilarityto DXA,demonstrates that the assessment without protocolcouldbeusedforfemalesusinganyofthedevices. For males, although devices 1 and 4 were influenced by theprotocol,bothassessmentsweresimilartothatofDXA. Thus,itissuggestedthatwhenusingthesedevices,oneof thetwoformsofassessmentshouldbestandardized.
Theprotocolusedaimedtostandardizefactorsthatmay influence BIA assessment accuracy, mainly related to the
state of hydration, such as beverage consumption, men-strual period,and physical activity.Guidelines related to thetechnicalaspectsofthedevices,whichwereprovided bythemanufacturers,werealsoobserved.16
BIA is based on the passage a low-intensity electric current through the body of the individual; impedance, resistance, reactance, and phase angle values are deter-mined,throughwhichbodycompositionisestimated.These valuesarestronglyrelatedtobodyhydration,aswaterisa goodconductorofelectricity,whilefatisnot.Ifthetissues areinatypicalconditionsofhydration,themethodaccuracy iscompromised.10,27
The protocol, however, can compromise adherence of adolescentsinpopulationstudies,asitrequireseffortand interest to follow the established requirements. It was observed,however,thatsomeoftheanalyzeddevices, par-ticularlyBIA3,hadgoodresultsevenwithouttheproposed standardization, suggesting an alternative when it is not possibletoconducttheprotocol.
Therehavebeennostudiesintheliteraturethatverified the influence of the protocol on the use of BIA in ado-lescents. However, some studies investigated differences betweenassessmentsaftertheconsumptionoffood,oneof theitemsthatcomprisedtheprotocol usedinthepresent study.
Vilac¸a et al.,28 when evaluating 41 elderly Brazilian
males,useddataobtainedbytetrapolarBIAandcompared to DXA,after fasting and after eating a meal. No differ-enceswereobservedbetweenthemeasurements(p>0.05). Theseresultscorroboratethosefoundinthepresentstudy, confirmingtheusefulnessoftheassessmentwithout proto-col.
Conversely, Gallagheretal.,11 whenstudying the
influ-ence of meals with different compositions onthe results estimated by BIA in 28Australian adultsof both genders, reportedthattherewasasignificantvariationinimpedance and consequently, on BF estimate, after consumption of meals. Similar results were demonstrated by Slinde and Rossander-Hulthen,10 when they evaluated healthy adults
by tetrapolarBIA method, beforeandafterthe consump-tionof threestandardizedmeals duringa 24-hourperiod. The authors concludedthat the impedance measurement decreasedapproximatelytwotofourhoursafterameal(p <0.05),causingvariationofupto8.8%(women)and9.9% (men)inBF%,underestimatingit.
The twostudies presented opposite resultstothoseof thepresentstudy,showingtheinfluenceoftheprotocolon theassessments,butbecausetheywerenotcomparedwith DXA,itisimpossibletoknowwhethertheassessmentafter consumptionofmealswouldalsobehelpful.
Regarding the agreement of assessments with BIA and DXA,BIAhadthreewithbetterresults,andalthoughnone of the devices had a kappa index > 0.8 (almost per-fectagreement)17,whenanalyzedtogetherwiththeother results,itwasobservedthattheresultsconfirmedthe pos-sibilityofusingtheassessmentwithoutprotocol.
The ROCcurve analysisshowedagaintheusefulnessof BIA in the absence of a protocol. There was no differ-encebetween areas withandwithout protocol for anyof the devices,indicating the capacity of thisassessment in predictingBF%increase,asallconstructedcurveswere sig-nificant(p<0.001).
BIA3showedthegreatestareasforthegeneraland strat-ifiedpopulation.Itisobservedthatthegeneralpopulation andthefemalegender hadhighersensitivitywhen adopt-ingtheprotocol,whichdemonstratesitscapacitytodetect agreaternumberofadolescentswithexcessBF.Asforthe malegender, the highest sensitivity was demonstratedat theevaluationwithouttheprotocol,whichagainhighlights itsusefulness.
This devicehasatetrapolar system,whichdiffersfrom the others,since it hasby eighttactile electrodes andis multifrequency.Thecombinationofthesefactorsappearsto ensuremoresensitivitywhenestimatingbodycompositionin
adolescents,whiletheprotocoldidnotinfluencetheresults inanyoftheanalyzedsituations.
TheotherBIAdevices,whichareoflowercostandmore availableforhealthservicesandthatalsoshowedmoderate sensitivity,specificity,andpositiveandnegativepredictive values,canbeusedwithcautionatthepopulationlevel,in theabsenceofmoresensitivemethods.
The assessment ofbody compositionofany adolescent performed by methods that arenot considered to bethe ‘‘goldstandard’’ shouldbemade considering thepossible errorsandshouldnotannultheimportanceandtheneedfor preventionactivitiesand/orcontrolofexcessBF,whatever theresults.
Conclusion
Based on the results, it was concluded that electrical bioimpedance has good predictive capacity to estimate excessBFinadolescents,andthatwhenitisnotpossibleto performtheprotocol, theresultsarealsosimilartothose ofDXA,thusallowingitsuseinpopulationstudies.
The tetrapolar device with eight tactile electrodes showedthehighest sensitivityandthebestresultsforthe overallpopulationandforthefemalegender,withprotocol, andforthemalegender,withoutprotocol.
Ahighprevalenceofadolescentswithhighpercentageof BFwasobserved,whichsuggeststheimportanceofspecific healthcareprogramsinthispopulation,aimingtocorrect dystrophiesandpreventcardiovascularandmetabolic dis-ordersinadulthood.
Due to the widespread use of electric bioimpedance devices,studieswithotheragegroups,inthepresenceand absenceof a protocol, arerequired to confirm its impor-tanceandindicatethereliabilityoftheresultsfortheentire population.
Funding
Financialsupportwasreceivedfromthefollowingagencies: Fundac¸ãodeAmparoàPesquisadoEstadodeMinasGerais (FAPEMIG)/ProcessNo.APQ-01618-10andConselhoNacional deDesenvolvimentoCientíficoeTecnológico(CNPq)/Process No.485986/2011-6.
Conflicts
of
interest
Theauthorsdeclarenoconflictsofinterest.
References
1.Organizac¸ãoMundialdaSaúde-OMS.Nutritioninadolescence -issuesandchallengesforthehealthsectorWHO.Geneva:WHO; 2005.
2.InstitutoBrasileirodeGeografiaeEstatística-IBGE.Pesquisa deorc¸amentosfamiliares2008-2009:despesas,rendimentose condic¸õesdevida.RiodeJaneiro;2010.
3.Kim HA, Lee Y, Kwon HS, Lee SH, Jung MH, Han K, et al. Gender differences in the association of insulin resistance withmetabolicriskfactorsamongKoreanadolescents:Korea NationalHealthandNutritionExaminationSurvey2008-2010. DiabetesResClinPract.2013;99:54---62.
4.JaegerAS,BarónMA.Usodelabioimpedanciaeléctricaparala estimacióndelacomposicióncorporalenni˜nosyadolescentes. AnVenezNutr.2009;22:105.
5.KimH,KimCH,KimDW,ParkM,ParkHS,MinSS,etal. Exter-nalcross-validationofbioelectricalimpedanceanalysisforthe assessment of body composition in Korean adults. NutrRes Pract.2011;5:246---52.
6.SiervogelRM,DemerathEW,SchubertC,RemsbergKE,Chumlea WC,Sun S,et al.Pubertyand bodycomposition. HormRes. 2003;60:36---45.
7.LinaresCL,CianguraC,BouillotJ-L,CoupayeM,DeclèvesX, PoitouC,etal.Validityofleg-to-legbioelectricalimpedance analysistoestimatebodyfatinobesity.ObesSurg.2011;21:917.
8.RodriguesMN,SilvaSC,MonteiroWD,FarinattiPTV.Estimativa dagorduracorporalatravésdeequipamentosde bioimpedân-cia, dobras cutâneas e pesagemhidrostática. Rev Bras Med Esporte.2001;7:125---31.
9.EickembergM,OliveiraCC,RorizAKC,SampaioLR. Bioimpedân-ciaelétricaesuaaplicac¸ãoemavaliac¸ãonutricional.RevNutr. 2011;24:873---82.
10.SlindeF,Rossander-HulthenL.Bioelectricalimpedance:effect of3identicalmealsondiurnalimpedancevariationand calcu-lationofbodycomposition.AmJClinNutr.2001;74:474---8.
11.GallagherM,WalkerKZ,O’DeaK.Theinfluenceofabreakfast mealontheassessmentofbodycompositionusingbioelectrical impedance.EurJClinNutr.1998;52:94---7.
12.Brasil --- Ministério da Saúde --- DATASUS Dados epidemiológi-cosVic¸osa2010.[cited03Mar].Availablefrom:http://tabnet. datasus.gov.br/cgi/deftohtm.exe?ibge/cnv/popmg.def. 13.Cardoso ChavesO,doCarmo CastroFranceschiniS,Machado
RochaRibeiroS,FerreiraRochaSant’AnaL,Garc¸ondeFariaC, EloizaPrioreS.Comparisonofthebiochemical,anthropometric andbodycompositionvariablesbetweenadolescentsfrom10to 13yearsoldandtheirparents.NutrHosp.2012;27:1127---33.
14.deOnisM,OnyangoAW,BorghiE,SiyamA,NishidaC,Siekmann J. Developmentofa WHO growth reference for school-aged childrenand adolescents. BullWorld Health Organ.2007;85: 660---7.
15.Lohman TG. Assessing fat distribution In: Advances in body compositionassessment:currentissuesinexercisescience. Illi-nois:HumanKinetics.Champaign.1992:57---63.
16.BarbosaKBF. Consumoalimentaremarcadoresde riscopara a síndrome metabólica em adolescentes do sexo feminino: comparac¸ãoentreinstrumentosdeinquéritodietético [Disser-tation].Vic¸osaMG:UniversidadeFederaldeVic¸osa;2006.
17.LandisJR,KochGG.Themeasurementofobserveragreement forcategoricaldata.Biometrics.1977;33:159---74.
18.Martinez EZ, Louzada-Neto F,Pereira BB. A curva ROC para testesdiagnósticos.CadernodeSaúdeColetiva.2003;11:7---31.
19.Coelho LG, Cândido AP, Machado-Coelho GL, de Freitas SN. Association between nutritional status, food habits and physical activity level in schoolchildren. J Pediatr (Rio J). 2012;88:406---12.
20.FariaER,FranceschiniSdoC,PeluzioMdoC,Sant’AnaLF, Pri-oreSE.Correlac¸ãoentrevariáveisde composic¸ãocorporale metabólicaemadolescentesdosexofeminino.ArqBrasCardiol. 2009;93:119---27.
21.SerranoHM,CarvalhoGQ,PereiraPF,PeluzioMdoC, Frances-chini Sdo C, Priore SE. Composic¸ão corpórea, alterac¸ões bioquímicaseclínicasdeadolescentescomexcessode adiposi-dade.ArqBrasCardiol.2010;95:464---72.
22.CarvalhoGQ,PereiraPF,SerranoHM,FranceschiniSdoC,de PaulaSO,PrioreSE,etal.Peripheralexpressionofinflammatory markersinoverweightfemaleadolescentsandeutrophicfemale adolescentswithahigh percentageofbodyfat.ApplPhysiol NutrMetab.2010;35:1---7.
23.PereiraPF,SerranoHM,CarvalhoGQ,LamounierJA,PeluzioMdo C,FranceschiniSdoC,etal.Bodyfatlocationand cardiovas-culardiseaseriskfactorsinoverweightfemaleadolescentsand eutrophicfemaleadolescentswithahighpercentageofbody fat.CardiolYoung.2012;22:162---9.
24.VieiraPR,FariaE,FariaF,SperandioN,AraújoC,StofelesR, et al. Fatores associadosà adiposidadeem adolescentes do sexofemininoeutróficascomadequadoeelevadopercentual degorduracorporal:elaborac¸ãodeummodeloderisco.Arch LatinoamNutr.2011;61:279---87.
25.Gonc¸alvesVSS,ChavesOC,RibeiroSMR,Sant’anaLF, Frances-chiniSdoC,PrioreSE.Disponibilidadedomiciliardelipídeos paraconsumoesuarelac¸ãocomoslipídeosséricosde adoles-centes.RevPaulPediatr.2012;30:229---36.
26.CostaRF, SantosNS,GoldraichNP, BarskiTF,deAndrade KS, KruelLF.Metabolicsyndromeinobeseadolescents:a compar-ison ofthreedifferent diagnostic criteria.J Pediatr (RioJ). 2012;88:303---9.
27.NationalInstitutesofHealthTechnologyAssessmentConference Statement,12.Nutrition.:Bioelectricalimpedanceanalysisin bodycompositionmeasurement.;1994.p.1---35.
28.VilacaKH,FerriolliE,LimaNK,PaulaFJ, MorigutiJC.Effect offluidandfoodintakeonthebodycompositionevaluationof elderlypersons.JNutrHealthAging.2009;13:183---6.
