Rev. paul. pediatr. vol.34 número3

REVISTA

PAULISTA

DE

PEDIATRIA

www.rpped.com.br

ORIGINAL

ARTICLE

Influence

of

glycemic

index

and

glycemic

load

of

the

diet

on

the

risk

of

overweight

and

adiposity

in

childhood

Kellen

Cristine

Silva

_,

_Luciana

_Neri

_Nobre

_,

Sofia

Emanuelle

de

Castro

Ferreira

Vicente

_,

_Lidiane

_Lopes

_Moreira

_,

Angelina

do

Carmo

Lessa

_,

_Joel

_Alves

_Lamounier

a_{UniversidadeFederaldoTocantins,Palmas,TO,Brazil}

b_{UniversidadeFederaldosValesdoJequitinhonhaeMucuri,Diamantina,MG,Brazil} c_{UniversidadeFederaldeSãoPaulo,SãoPaulo,SP,Brazil}

d_{UniversidadeFederaldeMinasGerais,BeloHorizonte,MG,Brazil} e_{UniversidadeFederaldeSãoJoãoDelRey,SãoJoãoDelRey,MG,Brazil}

Received12August2015;accepted25December2015 Availableonline10February2016

KEYWORDS

Glycemicindex; Overweight; Children; Adiposity; Carbohydrate

Abstract

Objective: Toinvestigatetheassociationbetweentheglycemicindexandtheglycemicloadof thedietwiththeriskofoverweightandhighadiposityinchildrenwith5yearsofage. Methods: Cross-sectionalstudynestedinacohortof232childrenbornandlivinginDiamantina (MG,Brazil).Parentsand/orguardiansprovidedthefoodintakedata,usingasemiquantitative foodfrequencyquestionnaire,pasthistoryandsocioeconomicconditions.Anthropometricand fatnessdatawerecollectedfromthechildren.Thedietaryglycemicindexandtheglycemic load werecalculatedfromthefoodintake.Theglycemicindexandglycemicloadeffecton overweightandadiposityinchildrenwasassessedbythePoissonregression(p<0.05).

Results: Theprevalenceofoverweightbybodymassindexwas17.3%,andhighadipositywas observedin3.4%and6.9%bytricepsskinfoldandsubscapularskinfold,respectively.No dif-ferencewas reportedbetweenthemean bodymassindex,tricepsskinfold andsubscapular skinfoldaccordingtotheglycemicindexandglycemicloadtertiles;however,theoverweight group presented ahigher carbohydrateintake(p=0.04).No associationwas found between glycemicindexandglycemicloadwithoverweightandadiposityamongthechildrenassessed. Conclusions: Theglycemicindexandglycemicloadofthedietwerenotidentifiedasriskfactors foroverweightandadiposityinthiscross-sectionalstudy.

©2016SociedadedePediatriadeS˜aoPaulo.PublishedbyElsevierEditoraLtda.Thisisanopen accessarticleundertheCCBYlicense(http://creativecommons.org/licenses/by/4.0/).

∗_{Correspondingauthor.}

E-mail:[email protected](K.C.Silva). http://dx.doi.org/10.1016/j.rppede.2015.12.009

PALAVRAS-CHAVE

Índiceglicêmico; Sobrepeso; Crianc¸as; Adiposidade; Carboidrato

Influênciadoíndiceglicêmicoecargaglicêmicadadietasobreoriscodesobrepeso eadiposidadenainfância

Resumo

Objetivo: Investigaraassociac¸ãoentreoíndiceglicêmicoeacargaglicêmicadadietasobreo riscodesobrepesoeadiposidadeemcrianc¸asdecincoanosdeidade.

Métodos: Estudotransversalaninhadoaumacoortede232crianc¸asnascidaseresidentesem Diamantina(MG,Brasil).Ospaise/ouresponsáveisforneceramosdadosdeconsumoalimentar, utilizandoumquestionáriosemi-quantitativodefrequênciaalimentar,históricodopacientee condic¸õessocioeconômicas.Osdadosantropométricosegorduracorporalforamcoletadosdas crianc¸as.Oíndiceglicêmicodadietaeacargaglicêmicaforamcalculadosapartirdaingestão dealimentos.Oefeitodoíndiceglicêmicoedacargaglicêmicanosobrepesoeadiposidadedas crianc¸asfoiavaliadoatravésdaregressãodePoisson(p<0,05).

Resultados: Aprevalênciadesobrepesopeloíndicedemassacorporalfoide17,3%,e adiposi-dadeelevadafoiobservadaem3,4%e6,9%atravésdapregacutâneadotrícepsepregacutânea subescapular,respectivamente.Nãohouvediferenc¸aentreamédiadeíndicedemassa corpo-ral,pregacutâneadotrícepsepregacutâneasubescapulardeacordocomostercisdeíndice glicêmicoecargaglicêmica;noentanto,ogrupocomsobrepesoapresentoumaioringestãode carboidratos(p=0,04).Nãofoiencontradaassociac¸ãoentreíndiceglicêmicoecargaglicêmica comsobrepesoeadiposidadeentreascrianc¸asavaliadas.

Conclusões: Oíndiceglicêmicoecargaglicêmicadadietanãoforamidentificadoscomofatores deriscoparasobrepesoeadiposidadenesseestudotransversal.

©2016SociedadedePediatriadeS˜aoPaulo.PublicadoporElsevierEditoraLtda.Esteéumartigo OpenAccesssobalicençaCCBY(https://creativecommons.org/licenses/by/4.0/deed.pt).

Introduction

TheBrazilianpopulation,particularlychildren,has experi-encedanutritionaltransitioncharacterizedbyadecreasein malnutritionandacorrespondingincreaseintheprevalence ofoverweightandobesity.1,2_{Correspondingtothisprocess,}

adropintheconsumptionofcereals,beansandtuberswas observedwitha subsequentincreasedintakeof processed foods,likesandwiches,cookies,snacksandsweets.2

Thischangeintheconsumptionpatternpredisposesthe intakeofahigh-refinedcarbohydratediet,characterizedby ahighGlycemicIndex(GI)andahighGlycemicLoad(GL). TheGIisdefinedastheincrementalareaundertheglycemic responsecurvepostconsumptionof25or50gofthe carbohy-dratesavailableinafood.Thisparameterisexpressedasa percentagerelativetotheglycemicresponseofareference food(glucoseorwhitebread)which,thus,reflectsthe qual-ityofthecarbohydrate.3_{TheGListheproductoftheGIof}

thefoodanditsavailablecarbohydratecontent;therefore, itrepresentsboththequalityandquantityofcarbohydrate anditsabilitytoraisethebloodglucose.4

HighGIandGLdietsarequicklydigested,absorbedand transformedintoglucose.Theseprocessesaccelerateinsulin and glucose fluctuations, resulting in the early return of hunger, causing excessive caloric consumption. However, low GI and GL diets provide a slow and gradual insulin andglucosereleaseinthebloodstream,therebypromoting increased fat oxidation, reducing lipogenesis and, conse-quently, increasing satiety and resulting in reduced food intake.5

Regardingthesefactors,someresearchershavestudied theassociation between the glycemicindex andglycemic load and the risk of overweight in children.6---11 _Nielsen

etal.10 _{andMurakamietal.}11 _{reportedthathighglycemic}

loadandindexdietsincreasetheriskofoverweightamong children and adolescents; however, it has not been con-firmedbyotherauthors.6---9

Consideringtheeffectoftheglycemicindexandglycemic loadonhungerandhigh-energyintake,thisstudyproposed to investigate the association between these parameters and the risk of overweight/obesity and high adiposity in children.

Method

This cross-sectional study was nested in a cohort of live births between September 2004 and July 2005 in Dia-mantina,Brazil.The objectivewastomonitorgrowthand development in the first year of life.12 _{Details regarding}

the cohort selection and cross-sectional study have been describedearlierelsewhere.13

Thepresentstudyinvolved5year-old(±5months) chil-dren ofboth gendersfromthecohortmentioned above.12

DatacollectionwasdonebetweenJuly2009andJuly2010. Eachpreschoolerwasvisitedintherespectivehome. Inter-viewsanddatacollectioncommencedonlyaftertheparents signed the consentform permittingtheir child to partici-pateinthestudy.Theresearchersweretrainedpriorindata collectiontoavoidmeasurement errors.Fournutritionists collected thedata. As across-sectional study nested in a cohort,thesamplepowerwascalculatedpos-hocusingthe parameter riskoverweight/obeseinrelation to3rdtertile for glycemicloadadjustedfor energyobtainedbyPoisson regression, which was1.20. Using thestatistical software ‘‘G*Power’’,85%powerwasobtained.

and Mucuri-UFVJM, protocol number 039/08. Parents or guardianssigned an informedconsent formtoparticipate inthestudy.

Thenutritionalstatuswasassessedbybodymass index-age (BMI/age) and adiposity was assessed by the triceps skinfold (TSF) and subscapular skinfold (SSF). Weight was measured using a portable, digital, electronic Kratos®

scale, witha maximum capacity of 150kg and 50g incre-ments; height was measured on a portable stadiometer (Alturaexata®_{), with 0.1cm scale accuracy, according to}

Jellife14 _{recommendations.Skinfoldsweremeasured using}

acompass checkskinfold Lange® _{SkinfoldCaliper scaleof}

upto60mmwith±1mmaccuracy,ontherightsideofthe body,basedonthestandardizationofLohmanetal.15_Three

measurementsat each anatomical point wereperformed, andtheaverageusedintheanalyses.

Thez-score<−2childrenidentifiedwithunderweight, z-score≥−1 andz-score≤+1 normal weight,z-score>+1 and z-score≤+2overweight,z-score>+2obesityaccordingtothe BMI/age.16,17 _{The subjects were grouped into two}

cate-gories for statistical analysis, underweight/normal weight were classified as normal weight and those with over-weight/obesityasoverweight.The z-score>+2.0 identified children withelevated adiposity (TSF/age, SBF/age).16 _To

identify the z-score of the children, the Software WHO Anthro and WHO Anthro plus versions 3.0.1 and 1.0.3, respectively,(WHO,Geneva).Thetwosoftwareswereused becausesome childrenhadalreadycrossed5 yearsofage whenassessed.

The mothers of the children were also subjected to weight and height evaluations to obtain their body mass index.ABMI≥25kg/m2_{wasconsideredhigh.}18

Thedataofbothmothersandchildrenwereassessedon asingleoccasion,inthemorning,utilizingthefacilitiesof theFederalUniversityofJequitinhonhaandMucuri.

Childrenidentifiedwithsome typeofnutritional disor-der received nutritionalguidance but this wasperformed onlyafterallthedatawerecollected.Theseguidelineswere directedatthechildrenandtheirfamilies.

Theusualintakeofpreschoolchildrenwasassessedbya Food-Frequency Quantitative Questionnaire (FFQQ) devel-oped by Sales et al.19 _{adapted to suit this study after a}

pilottest.TheparentsorguardiansfilledintheFFQQ dur-ingthehomevisit.Toincreasetheaccuracyofdetermining the amount offood the childreningested, aphoto album offoodservingsandeatingutensilswasused.Toobtainthe dailyintakeof eachnutrient,weemployedthe methodol-ogyproposedbyWillet20_{;thesumoftheportionweightswas}

multipliedbythefrequencyofconsumption.The chemical compositionofthefoodwasdeterminedusingtheDietPRO softwareversion5i(AgromídiaSoftwareLtda,Brazil).

Toobtainthedietaryglycemicindex,theGIofafoodwas initiallymultipliedbyitsdailycontributiontothe carbohy-dratesavailableineachserving,afterwhich,theseproducts were added up, using the equation proposed by Wolever etal.21 _{The dietary glycemic loadwasobtained from the}

productofthedietaryGIandthetotal available carbohy-drate intakedividedby 100.21 _{The available carbohydrate}

wasobtainedbysubtractingthefibercontentfromthetotal carbohydratesavailableinthefood.

TheGIofeachfoodwasidentifiedusingtheTableofFood CompositionoftheUniversityofSãoPaulo(TBCA/USP)22_;in

theabsenceofthisTable,weusedthevalueslistedinthe International Table of Glycemic Indexand Load glucose.4

WhenthefoodwasnotidentifiedinanyTable,theGIwas determinedbyusingtheGIofafoodwithsimilarnutritional characteristics,followingtheidenticalorderofreference. Thefoodusedasastandardwasglucose,GI=100.

To better understand the relationship between dietary GIandGLandoverweight/obesityandadiposity,theserates wereadjustedbythedietaryenergytomonitortheeffectof caloricintakeinthisrelationship.Finally,tothisweapplied the residual nutrient method proposed by Willet et al.23

The residualnutrient intakeof the nutrient wasadjusted bytheenergycalculatedbyaddingtheresidueofalinear regressionmodel. Thetotal energyintakewasconsidered theindependentvariableandtheabsolutenutrientvalueof thedependentvariable.Theestimatedenergyrequirement foreachchildwascalculatedbytheformulassuggestedby theInstituteofMedicine(IOM).24

Consideringthatothervariables,besidesdiet,can influ-enceoverweight/obesityandadiposityinpreschoolers,we alsodecided toevaluate the socioeconomicstatus of the children(percapitafamilyincomeinminimumwages),total breastfeedingperiod,maternal education(yearsof study) and overweight mothers. Parents or caregivers answered astructured questionnaire regarding these variables. The dataonthebreastfeedingpracticewasobtained prospec-tivelyfromthecohortstudy,whenthechildrenwerevisited everymonthathome,thereby,reducingthepossibilitythat the mother’smemory bias could have occurred regarding theeatinghabitsoftheirchildren.Tertilecutoffswereused tocategorizetheaveragemealglycemicloadandglycemic indexforsubsequent statisticalanalyses. The relationship betweenthecharacteristics ofthechildrenandtertilesof dietaryglycemicindexandglycemicloadwasexploredby ANOVA (parametric variables) and by Kruskal---Wallis test (nonparametricvariables).The association between nutri-tionalcompositiondietsandnutritionalstatuswasassessed byt-testandMann---WhitneyU-test.Poissonregressionwas appliedtodeterminetheinfluenceoftheglycemicindexand glycemicloadontheconditionofoverweightin preschool-ers.Initially,abivariateanalysiswasdoneandthevariables withp-value<0.20wereselectedformultivariateanalysis. Thesignificancelevelwasp<0.05.

All statistical analyses were performed using the Sta-tisticalPackagefor SocialSciences(SPSS)version19.0 for Windows(SPSSInc.,Chicago,IL,USA).Ethicalapproval(ref. no.ETIC545/08)wasobtainedfromtheFederalUniversity ofMinasGerais.

Results

Thedistributionofunderweight,normalweight,overweight and obesity was 7 (3.0%), 185 (79.7%), 38 (16.4%) and 2 (0.9%),respectively.Theprevalenceofoverweight/obesity was17.3%, with16.2% (n=23)in boysand 18.9%(n=17) in girls. High adiposity prevalence was 3.5% for the triceps skinfoldand6.9%forthesubscapularskinfold.

Table1 Anthropometric,socioeconomicanddietarycharacteristicsofchildreninDiamantina,Brazil,accordingtothetertiles ofdietaryglycemicindex.

Variables Dietaryglycemicindex

1◦_Tertile

(n=77)

2◦_Tertile

(n=78)

3◦ _Tertile

(n=77)

p-value

BMIchild(kg/m2_{,median±SD) 15.5(1.8) 15.4(1.6) 15.3(1.5) 0.574}a

Tricipitalskinfold(mm,median±SD) 8.9(3.3) 8.1(2.4) 8.3(2.7) 0.514a

Subscapularskinfold(mm,median±SD) 6.1(2.5) 5.5(1.9) 5.4(1.7) 0.601a

Daysofbreastfeeding(median±SD) 271.2(115.8) 254.9(126.7) 267.2(123.4) 0.596a

GLadjustedenergy(%,mean±SD) 151.6(45.9) 165.6(35.4) 185.5(34.2) <0.001b

IGadjustedenergy(%,median±SD) 64.0(3.6) 66.8(3.9) 70.7(4.0) <0.001a

Energyintake(g/day,mean±SD) 1528(575) 1695(641) 1528(613) 0.146b

Carbohydrateintakec_{(g/day,median±SD) 226.4(45.2) 235.5(24.9) 240.8(25.6) 0.001}a

Lipidintakec_{(g/day,mean±SD) 63.3(14.9) 61.1(8.0) 59.1(8.9) 0.058}b

Proteinintakec_{(g/day,median±SD) 50.4(12.9) 47.7(9.1) 43.0(8.4) <0.001}a

Fiberintakec_{(g/day,median±SD) 17.1(30.1) 12.0(2.9) 11.6(3.0) <0.001}a

Yearsmaternalstudied(median±SD) 10.0(4.0) 8.6(3.8) 7.4(3.2) <0.001a

Incomepercapitad_{(median±SD) 422.5(422.2) 269.1(259.3) 207.2(173.2) <0.001}a

BMImaternal(kg/m2_{,median±SD) 26.2(11.5) 27.4(14.5) 27.1(13.7) 0.780}a

BMI,bodymassindex;GL,glycemicload;IG,glycemicindex;SD,standarddeviation. a_{Kruskal---Wallistest.}

b _ANOVA.

c _{Valueadjustedforenergyintake.}

d _{Valuerelatedtotheminimumwageinreals(R$510.00).}

andhadmotherswithnormalBMIs(69.7%)whohadlessthan nineyearsofeducation(50.4%).

Considering that meat, eggs and margarine have no carbohydratesand,therefore,donotcontributetothe cal-culation of the IG and CG, only 87.7% of the foods that constitute theFFQQ contributedto glycemicdiet profile. Theremainingfoodwascategorizedandconsumedbythe children in the following frequencies: cereals and beans in3---4×/week,milk/dairyproductsandsugar-addeddrinks in2---3×/week,bakeryproducts/biscuits,vegetables,fruits andsweetsin1---2×/week.

Table 1 shows the anthropometric, socioeconomic and dietary characteristics of the schoolchildren according to thetertilesofthedietaryglycemicindex.Itwasnotedthat BMI,TSF,andSSFchildrendonotdifferbythetertilesofGI. Basedonthe dietary variables,the average carbohydrate increasedprogressivelyamongthetertiles,unlikethe aver-agefiberandproteinintake,whichdecreasedastheaverage GIincreasedamongthetertiles.

In the dietary GI analysis, nodifference wasobserved betweentheanthropometricvariables (BMI)andadiposity (TSFandSSF)andthetertilesofdietaryGL.

Reviewing the behavior of the dietary variables dis-tributed according tothe GL tertiles, we discoveredthat while they resemble the GI regarding the carbohydrate and protein intake, they are not like that of fiber and energy.The energyconsumption is significantlyhigher for thelast GLtertile but withnodifferencefor fiber intake (Table2).

Bothanalysesrevealedthatchildrenfromfamilieswith higherper capita incomeand havingmothers withhigher educationlevelsingesteddietswithbetterglycemicprofiles (Tables1and2).

Theoverweightgrouphadahighermediancarbohydrate intake(adjustedforenergy)thanthenormalweightgroup, clearlyindicatingtheeffectofdietarycompositiononthe nutritionalstatus(Table3).

Theratiosofthecrudeandadjustedprevalenceof over-weightandadipositybydietaryglycemicloadandglycemic indexareshown inTable4.Note that,after adjustingfor the potentialconfounders(e.g.,maternal BMI, percapita income,exclusivebreastfeeding,mother’seducationlevel), the dietary GI and GL were not independent risk factors for overweight/obesity and high adiposity in the children evaluated.

Discussion

A reasonable number of studies have observed the influ-enceofthequalityofdietarycarbohydratesonpromoting weight gainand obesity.25---27 _{Lowglycemic index and low}

glycemic load diets have been used totreat and prevent overweight and obesity in several studies, mainly in the UnitedStates.28_{However,mostofthemwereconductedon}

adults andelderly patients withchronic diseases. Against such abackdrop of scientific literature, thepresent work exploresthishypothesisinchildrenwith5yearsofage.

Table2 Anthropometric,socioeconomicanddietarycharacteristicsofchildreninDiamantina,Brazil,accordingtothetertiles ofdietaryglycemicload.

Variables Dietaryglycemicload

1◦_Tertile

(n=77)

2◦_Tertile

(n=78)

3◦_Tertile

(n=77)

p-value

BMIchild(kg/m2_{,median±SD) 15.3(1.6) 15.2(1.6) 15.7(1.8) 0.163}a

Tricipitalskinfold(mm,median±SD) 8.5(2.9) 8.4(2.9) 8.5(2.8) 0.978a

Subscapularskinfold(mm,median±SD) 5.8(2.4) 5.5(1.9) 5.6(2.1) 0.888a

Daysofbreastfeedingperiod(median±SD) 282.3(11.7) 262.3(124.8) 248.6(129.8) 0.637a

GLadjustedenergy(%,mean±SD) 131.6(33.8) 165.2(6.4) 205.8(34.0) <0.001b

IGadjustedenergy(%,median±SD) 64.4(4.3) 67.3(4.1) 69.8(4.2) <0.001a

Energyintake(g/day,mean±SD) 1531.6(627) 1446(534) 1779(632) 0.002b

Carbohydrateintakec_{(g/day,mean±SD) 218.9(48.5) 234.3(14.2) 249.5(19.9) <0.001}a

Lipidintakec_{(g/day,mean±SD) 62.8(15.1) 61.8(7.8) 58.9(8.9) 0.092}b

Proteinintakec_{(g/day,median±SD) 51.3(13.6) 47.8(6.7) 42.1(8.7) <0.001}a

Fiberintakec_{(g/day,median±SD) 15.2(30.2) 12.7(2.7) 12.89(3.8) 0.677}a

Yearsmaternalstudied(median±SD) 9.7(4.0) 8.6(3.7) 7.7(3.6) 0.008a

Incomepercapitad_{(median±SD) 374.0(360.9) 280.1(337.3) 244.6(215.9) 0.013}a

BMImaternal(kg/m2_{,median±SD) 26.4(11.1) 27.1(14.7) 27.8(15.1) 0.835}a

BMI,bodymassindex;GL,glycemicload;IG,glycemicindex;SD,standarddeviation. a _{Kruskal---Wallistest.}

b _ANOVA.

c _{Valueadjustedforenergyintake.}

d _{Valuerelatedtotheminimumwageinreals(R$510.00).}

coulddecreasetheGIindietsduetotheirhighfibercontent wereconsumedinamuchlowerfrequency.Itisnoteworthy thatthecerealsconsumedwereallof therefinedvariety, becausetheFFQQuseddidnotincludewholefoods.Inthis questionnairedesign,wholefoodswerenotincludedasthey wererarelyconsumedbythepopulationstudied.19_Besides,

theregularconsumptionofwholefoodsbychildren(5years old) is very low, especially in this sample that included mostlythelowincomegroup.

TheupperGIandGLtertileshavesignificantlylower pro-teinlevels andhigher carbohydrate levelscomparedwith thelowertertiles.Consequently,higherGLdietswere com-posed of higher carbohydrate levels for the total caloric value,consideringthatthelipidsdonotdiffersignificantly between the tertiles. Apart from this, in the univariate

analysis,itwasonlythecarbohydrateintakethatdiffered between the groups, while overweight children consume dietswithhigher quantities ofthis nutrient.This wasthe reasonfor including the total carbohydrate intake in the statisticalanalysisasaconfoundingfactor,avoidingthe pos-sibleeffectofdietaryGIandGLfortheriskofoverweight andadiposityamongthestudiedchildren.

Inthiswork,theGIandGLwerenotassociatedwithbeing overweight,astheaverageGIandGLdidnotdifferbetween thenormalweightandoverweightgroups.Thisfindingmay berelatedtothelowprevalenceofobeseindividuals(0.2%) found in this sample, when compared with the national prevalence (14.3%).2 _{Perhaps this association is best}

per-ceivedinobesechildrenandadults,thegroupswhich gen-erallyconsumelargerfoodservings.Children5yearsofage

Table3 NutritionalcompositiondietsaccordingBodyMassIndex.

Variables Groups

Eutrophic (n=192)

Overweight (n=40)

p-value

GLadjustedenergy(%,median±SD) 164.8(38.9) 182.0(37.1) 0.090a

IGadjustedenergy(%,median±SD) 67.2(4.6) 66.9(5.3) 0.772b

Energyintake(g/day,median±SD) 1578(599) 1517(682) 0.742b

Carbohydratesintakec_{(g/day,median±SD) 232.5(35.2) 242.6(22.8) 0.041}a

Lipidsintakec_{(g/day,mean±SD) 61.5(11.5) 59.5(8.9) 0.226}b

Proteinintakec_{(g/day,median±SD) 47.4(11.2) 45.7(7.9) 0.552}a

Fiberintakec_{(g/day,media±SD) 13.9(19.3) 12.1(2.2) 0.269}a

GL,glycemicload;IG,glycemicindex;SD,standarddeviation. a _{Studentt-test.}

b _{Mann---Whitneytest.}

Table4 Crudeandadjustedprevalenceratios(PR)foroverweightandadiposityaccordingdietaryglycemicloadandglycemic index.

Variables BMI TSF SSF

RP p-value RP p-value RP p-value

Crudeanalyse

GLadjustedforenergyintake

1◦_{Tertile 1 1 1}

2◦_{Tertile 0.82 0.623 0.99 0.987 0.82 0.738}

3◦_{Tertile 1.50 0.228 0.75 0.700 1.00 1.000}

GIadjustedforenergyintake

1◦_{Tertile 1 1 1}

2◦_{Tertile 0.68 0.278 0.14 0.064 0.62 0.377}

3◦_{Tertile 0.81 0.538 0.43 0.027 0.50 0.241}

Adjustedanalysea

GLadjustedforenergyintake

1◦_{Tertile 1 1 1}

2◦_{Tertile 0.77 0.509 1.12 0.872 0.94 0.917}

3◦_{Tertile 1.20 0.599 0.98 0.981 1.31 0.634} GIadjustedforenergyintake

1◦_{Tertile 1 1 1}

2◦_{Tertile 0.71 0.331 0.16 0.082 0.71 0.536}

3◦_{Tertile 0.81 0.579 0.57 0.465 0.61 0.441}

GL,glycemicload;GI,glycemicindex;BMI,bodymassindex;TSF,tricepsskinfold;SS,subscapularskinfold.

a_{Adjustedanalysesforbodymassindexmaternal,incomepercapita,yearsmaternalstudied,carbohydrateintakeadjustedforenergy,} allcontinuousvariables.AnalysisbyPoissonregression.

normallyconsumesmallservings;therefore,eveniftheyeat hyperglucidicfoods,becauseof thesmallquantities, only lowerGIandGLlevelsareobserved.Themethodologyused toidentifytheGIandGLconsiderstheamountof carbohy-dratethatthechildconsumesperserving,21_{reaffirmingthe}

impactofservingsizeinthecalculationoftheseparameters. Theassociationbetweenglycemicdietprofileandriskfor beingoverweightinchildhoodisstillnotwellestablishedin theliteraturebecauseof thelimitednumberofstudiesin childrenandwithquitecontroversialresults.6---11

Scaglioneetal.,6_{evaluatedthedietof111Italian}

chil-dren using the Food-Frequency Questionnaire (FFQ) and didnotidentifyany associationbetween bodymass index and dietary GI and GL. A prospective study of 380 chil-dren,recruitedfromtheGermanicDortmundNutritionand Anthropometric Longitudinally Designed Association, also found nosuch relationship.7 _{This hypothesis was alsonot}

confirmedinthestudydoneamong316children,6---7years ofage,inHongKong,inwhichthedietaryGLwasobtained fromthree dietary recalls and overweight by BMI.8 _Davis

etal.,9 _{examinedtheBMIandhigh adiposityin 120young}

overweight Latin-Americans by Dual Energy X-ray Absorp-tiometry (DEXA); the authors found that only total sugar intake,obtainedbytwodietaryrecalls,wasassociatedwith theseparameters.

In contrast, other researchers have identified a pos-itive association between the glycemic diet profile and overweight.10,11 _{Nielsen et al.}10 _{studied a population}

of 485,364 Danish children and adolescents, using one dietary recall and the sum of skinfolds. They reported that the dietary GI and GL were positively associated

with the parameters analyzed, only in the male adoles-cents. Murakami et al.11 _{evaluated Japanese children}

and adolescentsin a longitudinal study inwhich the data wereobtainedfromasemi-quantitativefoodconsumption frequency questionnaire, self-reported by the students’ families;theauthorsalsoanalyzedweightandheight. Sim-ilartothestudyofNielsenetal.,10_{theseauthorsobserved}

that thedietary GLwasassociatedwithan increased risk ofoverweightonlyamongchildrenandadolescentmales.

Two studies found a positive association between overweight10 _{andhigh adiposity}11 _{andGIandGL. Inthese}

studies, theoverweight grouphadhigher GIandGL aver-ageswhencomparedwiththenormalweightgroup,afinding notobservedinourstudy.Inaddition,thesamplesizewas likelytoinfluencethestatisticalanalysis.Murakamietal.11

and Nielsenet al.10 _{workedwith muchlargerpopulations}

thantheotherstudies.

The first four studies reported6---9 _{a variable that was}

includedinthestatisticalanalysesasapotentialconfounder, maternaloverweight.Thisfindingwasincommonwiththe present study. This same variable was not considered in thetwostudies10,11 _{which hadidentifiedapositive}

associ-ationbetween GI andGLand overweight.The fact is the maternaloverweightisapotentialconfoundingfactorand warrantsfurtheranalysis.Additionally,itisknownthat chil-drenwithageneticpredispositionmaybemorevulnerable tothemetabolicconsequencesofconsumingfoodswithhigh GIandGLcomparedwiththosewithoutafamilyhistoryof overweight.29

suchasthedietaryhabitsandlifestylesofthepopulations studied,dietaryassessment methodemployed,nutritional statusmeasuredandthepotentialconfoundersincluded(or not)intheanalyses.

Inthiscontext,thepresentstudyincludes some limita-tions.Amongthem,thecross-sectionaldesignofthestudy maybeinappropriatetoinvestigatethefrequency offood intakeandanthropometriccharacteristics.Inthiscase,the phenomenon of reverse causality may have occurred, as mothersofoverweightchildrencouldhaveoffered health-ierfoodcomparedwiththefoodtheynormallyconsumed. Asthefoodfrequencyquestionnaire19_{addressedthedietary}

intakeofthepriormonthandtheintervalbetweenthe invi-tationanddatacollectioncouldexceedonemonth, these recentchangesmayhaveinfluencedtheassociationthatthe studysoughttoinvestigate.Thislimitedtimemayhavebeen sensitivetotheeffectofreversecausality.

Another limitation is relatedto the GIFood Tables: In Brazil, wedo nothave a FoodTable thatincludes a wide varietyoflocalfood.Thus,tostudytheGI,aninternational

Table4hadtobeused,whichcouldnotaccuratelyrepresent theintakeofthechildren.

Inconclusion,dietaryglycemicindexandglycemicload arenotindependentriskfactorsforoverweight/obesityand adiposity in 5 year-old Brazilian children. Despite these results,itisnecessarytoinvestigatethishypothesisinother studies, both prospective and observational, because the literature revealed that these indexesare strategic tools thatcanpreventandtreatoverweight.However,although that outcome is not established in childhood, teaching anddirectingchildrentoselecthealthier,low-carbohydrate meals,shouldbeauniversalresponsibility,especiallywithin thefamily.

Funding

ThisworkwassupportedbyFoundationoftheStateofMinas GeraisResearchCaseNo.APQ-00428-08.

Conflicts

of

interest

Theauthorsdeclarenoconflictsofinterest.

References

1.BrazilianInstituteofGeographyandStatistics.Nationalstudy offamily expenditure 1974---1975:food consumption. Riode Janeiro:IBGE;1977.

2.BrazilianInstituteofGeographyandStatistics.Household Bud-get Survey 2008---2009: anthropometry and analysis of the nutritionalstatusofchildren,adolescentsandadultsinBrazil. RiodeJaneiro:IBGE;2010.

3.Jenkins DA, Wolever TM, Taylor RH, Barker H, Fielden H, Baldwin JM,et al. Glycemic index offoods: a physiological basis for carbohydrate exchange. Am J Clin Nutr. 1981;34: 362---6.

4.Foster-PowellK,HoltSH,Brand-MillerJC.Internationaltableof glycemicindexandglycemicloadvalues:2002.AmJClinNutr. 2002;76:5---56.

5.Ludwing DS. Dietary glycemic index and obesity. J Nutr. 2000;130Suppl.2:S280---3.

6.ScaglioniS,StivalG,GiovanniniM.Dietaryglycemicload, over-allglycemicindex,andseruminsulinconcentrationsinhealthy schoolchildren.AmJClinNutr.2004;79:339---40.

7.BuykenAE,ChengG,GüntherAL,LieseAD,RemerT, Karaolis-DanckertN.Relationofdietaryglycemicindex,glycemicload, added sugar intake, or fiber intake to the development of body compositionbetween ages2 and 7 y.Am J Clin Nutr. 2008;88:755---62.

8.HuiLL,NelsonEA,YuLM,LiAM,FokTF.Riskfactorsfor child-hoodoverweightin6-to7-y-oldHongKongchildren.IntJObes. 2003;27:1411---8.

9.DavisJN,AlexanderKE,VenturaEE,KellyLA,LaneCJ, Byrd-WilliamsCE,etal.Associationsofdietarysugarandglycemic indexwithadiposityandinsulindynamicsinoverweightLatino youth.AmJClinNutr.2007;86:1331---8.

10.NielsenBM,BjørnsboKS,TetensI,HeitmannBL. Dietary gly-caemicindexandglycaemicloadinDanishchildreninrelation tobodyfatness.BrJNutr.2005;94:992---7.

11.MurakamiK,MiyakeY,SasakiS,TanakaK,ArakawaM.Dietary glycemicindexand glycemicloadinrelationtoriskof over-weightinJapanesechildrenandadolescents:theRyukyusChild HealthStudy.IntJObes(Lond).2011;35:925---36.

12.Nobre LN, Lamounier JA, Franceschini SC. Preschool chil-dren dietary patterns and associated factors. J Pediatr. 2012;88:129---36.

13.LessaAC[PhDthesis]Foodandgrowthinthefirstyearoflife: acohortstudy.Salvador,BA:UFBA;2010.p.122.

14.Jelliffe DB. Evaluación delState nutriciónde lacomunidad. Geneva:WHO;1968.

15.LohmanTG,RocheAF,MartorellR.Antrhopometric standard-izationreferencemanual.Champaign:HumanKinetics;1991. 16.World Health Organization. WHO child growth standards:

length/height-for-age, weight-for-age, weight-for-length, weight-for-heightand bodymassindex-for-age.Methods and development. WHO (nonserial publication). Geneva: WHO; 2006.

17.DeOnisM,OnyangoAW,BorghiE,SiyamA,NishidaC,Siekmann J.DevelopmentofaWHOgrowthreferenceforschool-aged chil-drenandadolescents.BullWorldHealthOrgan.2007;85:660---7. 18.World Health Organization. Obesity: preventing and manag-ing the global epidemic. World Health Organ Tech RepSer. 2000;894:1---253.

19.Sales RL, Silva MM, Costa NM, Euclydes MP, Eckhardt VF, RodriguesCM,etal.Developmentofaquestionnairetoassess foodintakeofpopulationgroups.RevNutr.2006;19:539---52. 20.WillettWC.Nutritionalepidemology.2nded.NewYork:Oxford

UniversityPress;1998.

21.WoleverTM,JenkinsDJ,JenkinsAL,Josse RG.Theglycemic index:methodologyandclinicalimplications.AmJClinNutr. 1991;54:846---54.

22.Faculty ofPharmaceuticalSciences [homepage on the Inter-net]. Brazilian Food Composition Table-USP. Available from: http://www.fcf.usp.br/tabela[cited02.04.02].

23.WillettWC, HoweGR, KushiLH. Adjustment for totalenergy intake in epidemiologic studies. Am J Clin Nutr. 1997;65 Suppl.:1220S---8S.

24.NationalResearchCouncil.Dietaryreferenceintakesforenergy, carbohydrate,fiber,fat,fattyacids,cholesterol,protein,and aminoacids.Washington:NationalAcademyPress;2002. 25.Lau C, Toft U, Tetens I, Richelsen B, Jørgensen T,

Borch-Johnsen K, Glümer C.Association between dietary glycemic index, glycemic load, and body mass index in the Inter99 study:isunderreporting aproblem?AmJClinNutr.2006;84: 641---5.

27.MaY,OlendzkiB,ChiribogaD,HebertJR,LiY,LiW,etal. Asso-ciationbetweendietarycarbohydratesandbodyweight.AmJ Epidemiol.2005;161:359---67.

28.Bouché C,Rizkalla SW, Jing L, Vidal H, Veronese A, Pacher N,et al.Five-week,low-glycemic indexdiet decreasestotal fat mass and improves plasma lipid profile in moderately

overweight nondiabetic men. Diabetes Care. 2002;25: 822---8.