The healthy body image (HBI) intervention: Effects of a school-based cluster-randomized controlled trial with 12-months follow-up

(1)

ContentslistsavailableatScienceDirect

Body Image

j o u r n al ho me p ag e :w w w . e l s e v i e r . c o m / l o c a t e / b o d y i m a g e

The healthy body image (HBI) intervention: Effects of a school-based cluster-randomized controlled trial with 12-months follow-up

Christine Sundgot-Borgen

^a,∗

, Oddgeir Friborg

^b

, Elin Kolle

^a

, Kethe M.E. Engen

^a

, Jorunn Sundgot-Borgen

^a

, Jan H. Rosenvinge

^b

, Gunn Pettersen

^c

,

Monica Klungland Torstveit

^d

, Niva Piran

^e

, Solfrid Bratland-Sanda

^f

aNorwegianSchoolofSportSciences,DepartmentofSportsMedicine,Sognsveien220,N-0806,Oslo,Norway

bUiT–TheArcticUniversityofNorway,FacultyofHealthSciencesDepartmentofPsychology,9037,Tromsø,Norway

cUiT-TheArcticUniversityofNorway,FacultyofHealthSciencesDepartmentofHealthandCaringSciences,N-9037,Tromsø,Norway

dUniversityofAgder,FacultyofHealthandSportSciences,Postbox422,4604,Kristiansand,Norway

eUniversityofToronto,DepartmentofAppliedPsychologyandHumanDevelopment,252BloorStreetWest,Toronto,Ontario,M5S1V6,Canada

fUniversityCollegeofSoutheastNorway,DepartmentofSports,PhysicalEducationandOutdoorStudies,P.O.Box235,N-3603,Kongsberg,Norway

a r t i c l e i n f o

Articlehistory:

Received17December2018

Receivedinrevisedform22March2019 Accepted22March2019

Availableonline28March2019 Keywords:

Healthpromotion Embodiment Bodyappreciation Adolescents Qualityoflife

a b s t r a c t

WeexaminedtheeffectsoftheHealthyBodyImage(HBI)interventiononpositiveembodimentand health-relatedqualityoflifeamongNorwegianhighschoolstudents.Theinterventioncomprisedthree interactiveworkshops,withbodyimage,medialiteracy,andlifestyleasmainthemes.Intotal,2,446 12^thgradeboys(43%)andgirls(meanage16.8years)from30highschoolsparticipatedinacluster- randomizedcontrolledstudywiththeHBIinterventionandacontrolconditionasthestudyarms.Data werecollectedatbaseline,post-intervention,3-and12-monthsfollow-up,andanalysedusinglinear mixedregressionmodels.TheHBIinterventioncausedafavourableimmediatechangeinpositiveembod- imentandhealth-relatedqualityoflifeamonginterventiongirls,whichwasmaintainedatfollow-up.

Amonginterventionboys,however,weakpost-interventioneffectsonembodimentandhealth-related qualityoflifevanishedatthefollow-ups.FuturestudiesshouldaddressstepstomaketheHBIinterven- tionmorerelevantforboysaswellasdeterminewhetherthenumberofworkshopsorthemesmaybe shortenedtoeaseimplementationandtoenhanceinterventioneffects.

1. Introduction

Positive embodiment and body appreciation are important aspects of health and quality of life (Avalos, Tylka, & Wood- Barcalow,2005;Piran,2019;Tiggemann,2011).Inpreviousstudies, positiveembodimentandbodyappreciationhavebeenassociated withpositiveself-andbodyesteem,healthyeating,andperforming regularphysicalactivityinboysandgirls(Cash&Fleming,2002;

Neumark-Sztainer,Paxton,Hannan,Haines,&Story,2006;Santos, Tassitano,doNascimento,Petribú,&Cabral,2011;Tylka&Homan,

∗ Correspondingauthorat:NorwegianSchoolofSportSciences,Departmentof SportsMedicine,Sognsvegen220,N-0806,Oslo,Norway.

E-mailaddresses:c.s.borgen@nih.no(C.Sundgot-Borgen), oddgeir.friborg@uit.no(O.Friborg),elin.kolle@nih.no (E.Kolle),k.m.e.engen@nih.no(K.M.E.Engen),

jorunn.sundgot-borgen@nih.no(J.Sundgot-Borgen),jan.rosenvinge@uit.no (J.H.Rosenvinge),gunn.pettersen@uit.no(G.Pettersen),monica.k.torstveit@uia.no (M.KlunglandTorstveit),niva.piran@utoronto.ca(N.Piran),

Solfrid.Bratland-Sanda@usn.no(S.Bratland-Sanda).

2015).Further,bodyimagehasbeenfoundtopredicthealth-related qualityoflifeinboysandgirls(Grifﬁthsetal.,2017;Haraldstad, Christophersen,Eide,Natvig,&Helseth,2011).

Thereishoweverawell-knowngenderdifference,asfewerado- lescentboysstrugglewithbodyimageissues(13–45%)compared toadolescentgirls(45–71%)(Martinsen,Bratland-Sanda,Eriksson,

&Sundgot-Borgen, 2010; Torstveit,Aagedal-Mortensen, &Stea, 2015).In the samevein, adolescent boysreport more satisfac- tionwiththeirbodiesandhigherlevelsofembodimentcompared toadolescentgirls(Franko,Cousineau,Rodgers,&Roehrig,2013;

Holmqvist,Frisén,&Piran,2018;Neumark-Sztaineretal.,2006;

Santosetal.,2011).Fromadevelopmentalperspective,changesin theexperienceofthebodyduringthecriticalphaseofadolescence canhavealong-termimpactonbodyimage(Wertheim,Paxton,&

Blaney,2009).Promotingpositiveembodimentinadolescenceis thereforevitaltoestablishagoodbasisforhealth-relatedquality oflife,assuchqualityoflifehasprovedstableduringthelifecourse (Bisegger,Cloetta,vonRueden,Abel,&Ravens-Sieberer,2005),and canbeviewedasacoreissueforpublichealth.

https://doi.org/10.1016/j.bodyim.2019.03.007

(2)

Systematicreviewsshowthatuniversalinterventionprograms thataresuccessfuladdressthereductionofriskfactors,asforexam- plebodydissatisfaction,inordertopreventeatingdisordersamong adolescents (Le, Barendregt, Hay, & Mihalopoulos, 2017; Stice, Shaw, &Marti, 2007; Yager, Diedrichs, Ricciardelli, &Halliwell, 2013).Withinahealthpromotionperspective,promotingpositive embodimentrepresentsatheoreticalandmethodologicalparadig- maticshiftfromthedisease-preventingfocus,e.g.,bypreventing bodydissatisfaction,toahealth-promotionfocus(Leetal.,2017;

Stice,Becker,&Yokum,2013).Thisshiftopensnewpossibilitiesto assesshealth-promotioninterventions(Piran,2015;Tylka&Wood- Barcalow,2015; forexamples,seeAllevaetal.,2018; Halliwell, Jarman,Tylka,&Slater,2018;McCabe,Connaughton,Tatangelo, Mellor,&Busija,2017).

Theresearch-basedpositiveembodimentconstructisdeﬁned as“positivebodyconnectionandcomfort,embodiedagencyand passion,andattunedself-care”(Piran,2016,p.47).Positiveembod- imentrelatesconceptually tobodyappreciation (Tylka&Piran, 2019),themostcommonly usedconstructinassessing positive body image(Tylka,2019).Bothpositive embodimentand body appreciationemphasizepositiveconnectionto,andappreciation of,thebody,aswellasattunedcareofthebody(Tylka&Piran, 2019).Thepositiveembodimentconstruct,however,includesin addition,experiences ofagencytoactintheworldandcomfort withbodilydesires(Piran,2019).

Researchers have called for intervention studies that aim to enhance embodiment and health-related quality of life (Alleva, Sheeran, Webb, Martijn, &Miles, 2015; Tylka &Piran, 2019).Yet, mostexisting intervention studies lack inclusion of multidimensional instruments of positive embodiment (Webb, Wood-Barcalow, & Tylka, 2015). In particular, no randomized, controlledoutcomeevaluationstudieshavebeenconductedasa universalpromotingprogramaimedatenhancingpositiveembod- imentinbothboysandgirlsinlateadolescence(Allevaetal.,2015).

1.1. DevelopmentandimplementationoftheHBIintervention

We have developed the universal,multi-component healthpromotion intervention “Healthy Body Image” (HBI; Sundgot- Borgen et al., 2018). The HBI intervention focuses on positive embodiment and health-related quality of life among Norwe- gianhighschoolstudents,andemploysaninteractiveeducational approach,whichhasbeenfoundsuitableinschoolsettings(Yager etal.,2013).

The HBI intervention comprised three overarching themes relatedtobodyimage,medialiteracy,andlifestyle,asthesehave beenfoundtoimprovephysicalself-perception,bodysatisfaction andappreciation,physicalcompetence,andbodyesteem,some- timeswithlargeeffectsizes(Allevaetal.,2015;Espinoza,Penelo,

& Raich, 2013; Franko et al., 2013; Tomyn, Fuller-Tyszkiewicz, Richardson, & Colla, 2016).A more detailed description of the programanditsrationalehasbeenpublishedelsewhere(Sundgot- Borgenetal.,2018).

Theprogramwasconstructedtoincludebothboysandgirlsin lateadolescence.Thiswasimportantbecausethepeerenvironment isshapedbysocioculturalidealsofbothgenders.Bothboys’and girls’attitudesmustchangeifthesocialenvironmentofthewhole schoolcanbechanged(Yageretal.,2013).Duetothemixed-gender sample,theinterventioncontainedgenderneutralizedandgender speciﬁccontents(e.g.,pictures,videos,communicationexamples), to make it relevant for both genders. Despite some debate on whatageismostappropriateforinitiationofbodyimageinter- ventions,evidencesuggeststhatinpreventionstudies,itmightbe beneﬁcialtotargetyoungadolescentsprior totheonsetofeat- ingdisorders(Espinozaetal.,2018;Rohde,Stice,&Marti,2015).

However,lateadolescenceinvolvespubertal,cognitive,andinter-

personalchanges, whichincrease adolescents’ability toreacha moreabstractcharacterizationofthemselves,theinﬂuenceoftheir peersincreases(Rohdeetal.,2015),andtheymaybecomemore awareofandvulnerabletopressurestoattainsocioculturalbeauty ideals.Theyareatanagewheretheriskforeatingdisorderspeaks (Espinozaetal.,2018;Rohdeetal.,2015;Sticeetal.,2007),and promotionofpositiveembodimentisespeciallycrucial,astheyare movingtowardstheindependenceofyoungadulthood.Also,their improvedabilityforabstractreasoningmakesthemmorelikely tocomprehendtheinterventioncontent,relateskillstotheirown lives,andtakeadvantageofsuchtaughtskills.

Theschoolcontextalsoensuresarelativelycomparablepar- ticipationratebetweengenders,whichisanobviousassetsince fewexistingstudieshavemanagedtoincludeabalancedgender sample.Moreover,amixed-genderapproachmayofferamorereal- lifesettinginuniversallyimplementedhealthpromotioninitiatives (Yageretal.,2013).

1.2. Hypothesis

We hypothesized that the HBIintervention would be effective,resultinginmorefavourablescoresonpositiveembodiment (higher)andhealth-relatedqualityoflife(higher)inintervention studentscomparedtocontrolstudents.

2. Method

2.1. Designandrandomization

Acluster-randomizedcontrolleddesignwasusedwithschools astheclusteringfactorataratioof1:1.Schoolswererandomlyallo- catedtoeithertheHBIinterventionorthecontrolgrouptoequalize samplesize,andtheeffectofsocioeconomicanddemographicvari- ables,notablyrelatedtoethnicityandtheurban-ruraldimension.

Thesamplewouldbeconsideredrepresentativeoftheadolescent populationofOsloandAkershusCounty.Therandomizationwas conductedbyaprofessionalnotafﬁliatedwiththestudytomini- mizecontaminationbiaseswithinschools.Duringtheintervention period,studentsatthecontrolschoolsfollowedtheirregularschool curriculum.Fig.1presentsadiagramoftheinclusionandrandom- izationprocessofschoolsandstudents,respectively.

2.2. Samplecharacteristics

Thirty schools were randomized and 2,446, 1,254, 1,278, and 1,080 students consented to participate at pre-test, post- intervention,and3-and12-monthsfollow-up,respectively(Fig.1).

Themean(range)numberofstudentsconsentingateachschool was82(22–184),42(5–97),43(4–125),and36(3–103)atpre-test, post-intervention,and3-and12-monthsfollow-up,respectively.

Thenumberofstudentsincludedintheprimaryoutcomesanalyses were1,742,1,190,1,172,and955fortheExperienceofEmbodi- mentScale,and1,688,1,173,1,158,and925fortheKIDSCREEN-10 andGeneralhealthacrossthefourmeasurementoccasions.The participantswere16.8(SD=0.76)yearsold,and11%,and1%were categorizedasoverweightandobese,respectively.Amongthepar- ticipants,13%werecategorized asimmigrants,39%hadparents withatotalincomeof≥1millionNOK,and82%reportedoneor bothparentshavingahighereducation.

2.3. Ethicsapprovalandconsenttoparticipate

The study met the intent and requirements of the Health ResearchActandtheHelsinkideclaration,andwasapprovedby theRegionalCommitteeforMedical andHealthResearchEthics (P-REK 2016/142).It wasenrolledintheinternationaldatabase

(3)

Fig.1.Recruitment,clusterrandomizationofschools(c)andstudents(N),andresponserateofparticipatingstudents.

ofcontrolledtrialswww.clinicaltrials.gov(ID:PRSNCT02901457).

Students at consenting schools had the prerogative to decline participationafterconsent.Insuchcases,studentswereallowed tofollowtheHBIworkshops,butwithoutcompletingtheques- tionnaires.Aftertheﬁnal12-monthsfollow-up,controlschools wereofferedonelecturewheretheprogramhighlightswerecom- pressed.Themethodsandresultsaredescribedaccordingtothe ConsortStatement(Moher,Schulz,&Altman,2001).

2.4. Procedureanddatacollection

AsaresultofasubsequentpilotstudyduringMarchandApril 2016among12012^thgradehighschoolers,afewquestionnaire

itemsaboutbodyperceptionandnutritionweredeletedtoreduce theriskoferrorvarianceduetoacquiescencebias.Inaddition,the amountofworkshopassignmentswasreducedtoallowformore timeallocatedtodiscussmoodandbodysatisfactionissues.

TheHBIinterventionincludedall12^thgradehighschoolclasses followingageneralstudyprogram,excludingstudentsfollowing avocationalstudyprogram.Nofurtherexclusioncriteriawereset.

DuringSpring2016,principalsofallpublicandprivatehighschools inOsloandAkershusCountyinNorwaywerecontactedbye-mail.

Oralandwrittenstudyinformationwasprovidedtostudentsand staffattheconsentingschools.TheNorwegianHealthResearchAct statesthatadolescents,16yearsorolder,cangivetheirinformed consentwithnoparentalconsentneeded.Studentsweresentane-

(4)

mailwithstudyinformationandaletterofinformedconsent.Ifthey pressed ¨yes¨tothequestionofconsent,theyweregivenaccesstoa linkthatmadethequestionnairepackageavailable,andtheycom- pletedthequestionnairepackagethroughtheonlinesurveysystem SurveyXact8.2.Ethicalapprovalofthestudyrequiredthatthestu- dentscompletedthequestionnairesoutsideregularschoolhours.

Studentswereinformedabouttheirallocationintotheintervention orcontrolgroupaftertherandomization.

2.5. Measures

As described in the study protocol (Sundgot-Borgen et al., 2018),participantscompletedstandardizedquestionnairesrelated todemographics,positiveembodiment,andhealth-relatedqual- ityoflifeatbaseline,post-intervention,andat3-and12-months follow-up,respectively.Allbaselineassessmentswereconducted priortotherandomization.Post-interventionassessmentwasnot availablethesamedayasthelastworkshop,butwithinoneweek (Sundgot-Borgenetal.,2018).

2.5.1. Demographicvariables

Thedemographicvariableswerecollectedatallmeasurement occasions,includingage,gender,and self-reportedbodyweight (kg) and height (cm).BMI was calculated as body weight (kg) dividedbytheheightsquared(m²).Categorizationofweightsta- tuswasbasedoninternationalage-andgender-adjustedcut-off scores(Cole,Bellizzi,Flegal,&Dietz,2000).Totalparentalincome wasmeasuredbyaskingthestudentswhattheybelievedtobetheir parents’totalincome,selectingoneofﬁveoptions(lessthanNOK 200.000,NOK200.000-400.000,NOK500.000-800.000,NOK900.000 -1million,morethanNOK1million,respectively).Studentsalso tickedoffiftheirparentshadcompleted1.Primaryschool,2.High school,3.College/University,orwhetherthey4.Didnotknow.Immi- grationstatuswasmeasuredbyaskingwhetherthestudentorboth parentshadimmigrated(YesIhave,Yesbothmyparents,No).

2.5.2. Positiveembodiment

PositiveembodimentwasmeasuredusingtheExperience of Embodiment Scale (EES) (Teall & Piran, 2012). The Cronbach’s alphafor thecurrent study was .93 for girls and .92 for boys, similartootherstudieswiththerangeof.91–.94(Chmielewski, Bowman,&Tolman, 2019; Holmqvistetal., 2018; Piran,2019;

Teall,2006,2014).Test-retestreliabilityovera3-weekperiodof theEESwasalsopreviouslyfoundtobeacceptable(r=.93)(Piran, 2019).The34itemscoveredpositiveconnectionwiththebody, agencyandfunctionality,experienceandexpressionofdesire,body attunement,self-carevs.harm/neglect,andsubjectivelensvs.self- objectification(e.g., Ïamproudofwhatmybodycando änd Ïcare moreabouthowmybodyfeelsthanabouthowitlooks¨).Theitems hadaLikert-formatrangingfrom1(stronglydisagree)to5(strongly agree),andthe17negativelyframeditems(e.g., Ïignorethesigns mybodysendsme änd ¨Mydissatisfactionwithmybody/appearance hasanegativeeffectonmysociallife¨)werereversedsothatthesum scorereflectedhigherlevelsofpositiveembodiment.

AdequateconstructvalidityoftheEEShasbeenfoundinpre- viousstudiesonyoungadultsasreﬂectedbypositivecorrelations withmeasuresofbodyesteeminwomen(rs=.76–.79)andmen (r=.69), body responsiveness (r=.73), body connection(r=.60), well-being(rs=.55–.80),andlifesatisfactioninmen(r=.68)and women(r=.66).Further,theEEScorrelatednegativelywithmea- suresofobjectiﬁedbody consciousness(rs=-0.55, -.73),eating problems (rs = -0.43, -.70), alexithymia (rs = -0.51, -.54), and depression(r=-0.63) (Chmielewski,Tolman,&Bowman,2018;

Holmqvistetal.,2018;Piran,2019;Teall,2006,2014).Youngmen havereportedhigherEESscorescomparedtowomen(Holmqvist etal.,2018).Sincethepresentinvestigationincludedlateadoles-

cents,ages16–17,thestudyusedtheadultversionoftheEES.To date,mostvalidationstudiesoftheEESwereconductedinyoung adultsamples,suchasChmielewskietal.(2018)thatincluded340 womenbetweentheagesof18–26withanaverageageof19.81.

Based onaseries ofconﬁrmatoryfactor analyses, theglobal EESscorewasusedasanoutcomemeasure.Whileitsoriginal6- factormodelshowedanadequateﬁtwhenmodelingthemethod variance relatedtothepositively andnegatively worded items,

␹²(507)=3311,p<.001,RMSEA=0.056,CFI/TLI=.890/.867,SRMR

=.066,weusedaglobalscoresinceageneralsecond-orderfactor,

␹²(516)=3431,p<.001,RMSEA=.057,CFI/TLI=.875/.864,SRMR= .076,accountedadequatelyforthe6-factormodel.

2.5.3. Health-relatedqualityoflife

Health-relatedqualityoflifewasmeasuredbytheKIDSCREEN- 10, which is a widely used and validated self-report tool (Ravens-Sieberer,2006),andhasbeenvalidatedinNorwegianado- lescents(Haraldstad&Richter,2014).Thescaleconsistsof10-items (e.g., ¨Haveyoufeltfitandwell? änd ¨Haveyoufeltsad?¨).Thesum scoreofthe1–10providesageneralhealth-relatedqualityoflife index.AseparateitemincludedintheKIDSCREEN-10measured perceivedGeneral Health(Ïn general,how would you sayyour healthis?¨),whichhasbeenfoundtocorrelatewellwithmeasuresof physicalwell-being(r=.63)andpsychologicalwell-being(r=.51) (Bartheletal.,2017).Allitems,1–11,hada5-pointLikert-typefor- matfrom1(notatall/never)to5(extremely/always)for10items,and from1(excellent)to5(poor)fortheGeneralHealthitem.Negatively wordedquestionswerereversed,andhenceahigherscoreindi- catedhigherlevelsofhealth-relatedqualityoflife.Standardized T-scoreswerepresentedatbaselinetoenablecomparisonofmeans acrossstudysamplesandcomparedatatohealth-relatedqualityof lifenormdata.Ascoreof50representsthemean.AT-score<38 ontheKIDSCREEN-10indicateslowerhealth-relatedqualityoflife, whilescores≥38indicatepreferablereportedhealth-relatedqualityoflife(Ravens-Sieberer,2006).Theinternalconsistencyforthis samplewas␣=.81,andhasbeenfoundtobesatisfactoryinother samplesofadolescentboysandgirls(Haraldstadetal.,2011).

2.6. TheHBIintervention

There is no consensus as to which theoretical orientation may provide the most effective approach when developing a healthpromotioninterventionaimingtopromoteembodimentand health-relatedqualityoflife(Allevaetal.,2015).However,asocio- culturalperspective(Thompson,Heinberg,Altabe,&Tantleff-Dunn, 1999)wasnaturaltoconsiderwhenaimingtochangeattitudes, beliefs, and knowledge related toidealized lifestyles(involving e.g., extreme exerciseand diet regimes) and bodies, to further strengthentheresiliencetowardsunhealthyinternalization,and strengthen life-managingskills ina mixed-genderschool-based setting. Also,anetiological modelof riskand protective factors (Piran,2015;Smolak&Piran,2012)aswellasthedevelopmental theoryofembodiment(Piran,2017;Teall&Piran,2012)withinthe realmofpositivepsychology(Seligman&Csikszentmihalyi,2000), wereimportantinitsdevelopment.

AlthoughthoroughlydescribedintheAppendix,someimpor- tant aspects of the intervention speciﬁcallyaiming topromote positiveembodimentarepresented.

Through the body image and medialiteracy workshops,we aimedtoimprovecriticalawarenessofunhealthybodyandlifestyle idealization,criticalandconstructiveuseofsocialmedia,includ- ingconsequencesofcurrentbodyidealsforboysandgirls.Bythis, weintendedtoreducetheriskofinternalizationofunhealthyide- als,self-harm,andneglect,aswellaspromoteasubjectivelens whilereducingself-objectiﬁcation.Toimproveapositiveconnec- tionwiththebody,weaimedtostrengthenattitudestowards,and

(5)

Table1

EstimatedBaselineMean(SD)DifferencesinDemographics,PositiveEmbodiment(EES),andHealth-RelatedQualityofLife(KIDSCREEN)betweenInterventionandControl students.

Boys(N=1044) Girls(N=1402)

Intervention(n=632) Control(n=412) p-value(d/␸) Intervention(n=867) Control(n=535) p-value(d/␸)

Ageinyears 16.84(0.57) 16.78(0.64) .117 16.80(0.54) 16.78(0.53) .426

BMI,kg/m² 21.85(3.45) 21.77(3.26) .741 21.41(2.82) 21.43(3.65) .946

Immigrationstatus^a 62(9.81%) 71(17.20%) .001(0.11^g) 109(12.50%) 87(16.20%) .057

Parents’income≥1NOKmillion^b 319(49.9%) 186(44.6%) .101 324(36.8%) 143(26.3%) <.001(-.11^g) Parents’educationallevel^c 544(86.5%) 314(76.6%) <.001(-.13^g) 745(85.1%) 416(77.5%) <.001(-.10^g)

EES^d 130.15(20.91) 126.73(22.18) .054 117.31(22.70) 114.03(24.31) .023(0.15)

KIDSCREEN-10^e 38.29(6.10) 38.00(6.43) .580 35.78(6.01) 34.53(5.92) .001(0.21)

KIDSCREEN-10T-score 53.10(9.76) 52.55(10.30) .580 48.99(9.64) 46.98(9.48) .001(0.21)

GeneralHealth^f 3.70(1.07) 3.59(1.17) .254 3.30(1.05) 3.05(1.10) <.001(0.23)

Note.BMI=Bodymassindex.EES=TheExperienceofEmbodimentScale.^aImmigrationstatus:bothparentsareimmigrants,^bParents’income:parentswithtotalincome≥ 1millionNOK,^cParents’educationallevel:oneortwoparentswithcollegeoruniversityeducation,presentedastotalnumberandpercentage(%)oftotalnineachgroup andforeachgender.^dEESglobalscorerangesfrom34to170;^eKIDSCREEN-10rangesfrom10-50.^fGeneralHealthscorerangesfrom1-5.p-value<.05.d=Cohen’sdand^gϕ

=phi-coefﬁcientarepresentedforsigniﬁcantdifferences.

knowledgeabout,howtopromoteself-careandexperienceofbody functionalitywhendiscussinglifestylefactors,suchasnutrition, exercise,andsleep.

Theinterventionwasdevelopedtosuitthecognitivedevelop- mentamongadolescents16yearsofageintermsoftheirabilityfor abstractreasoning.Theworkshopdeliverywasbasedontheelab- orationlikelihoodmodel(Petty&Bri ˜no,2012;Petty&Cacioppo, 1986).Accordingtothismodel,aswellaspreviousﬁndings(Alleva et al., 2015; Stice et al., 2013, 2007), the program contained three90-mininteractiveworkshopstofacilitateextensivestudent discussions.Allworkshopswerearrangedinclassroomsduringreg- ularschoolhours.About60boysandgirls(i.e.,twoschoolclasses) participatedperworkshop.Studentattendancewasregisteredat eachworkshoptocalculateprogramadherence.A3-weekinterval betweeneachworkshopresultedina3-monthinterventionperiod.

Theﬁrstandfourthauthorfacilitatedtheintervention.Bothare specializedinphysicalactivityandhealth,sportsnutrition,motiva- tionalinterviewing,andbodyimageamongadolescents.Detailed information about theintervention contentand targetscan be foundinthestudyprotocol(Sundgot-Borgenetal.,2018).

2.7. Samplesizeandpoweranalyses

Thestatisticalpowerestimationwasbasedontwocomparison groups(␣=.05andb=.20)withanaveragewithin-clustersam- plesizeof70students.Theexpectedeffectsizewas.28according toameta-analysis(Hausenblas&Fallon,2006)thatincluded35 studiesexamininginterventioneffectsonbodyimagesvariables.

Moreover,weassumedthatthewithin-clusterdependencyrelated toschoolsaccountedforapproximately3%(ICC=.03).Thisisfair forvariablesrelatedtopsychologicalormentalhealthoutcomes, asselectionfactorslikesocioeconomicstatusaffectthesevariables lessthanforexampleacademicperformance.Theseconsiderations requiredaminimumof10clusterswithineachgroup,requiring atotalsamplesizeof10schools×2groups×70students ˜1400 students.

2.8. Statisticalanalysis

Thesoftwareprogram Mplus,version8.0, wasused tocarry outfactoranalyses,whileremainingstatisticswereanalysedusing IBMSPSS24 for Windows.The adequacy ofthe randomization procedurewasexaminedbycomparinggroupdifferencesatbase- linewithindependentt-tests,chi-squaretests,orKruskal-Wallis tests (Table 2). A case was recorded as dropout if all post- intervention and follow-up data were missing. Due to several layersofdependencyin theoutcomedata,linearmixedregres- sionmodelswereﬁt,assuggestedincomparablestudies(Wilksch

etal.,2017).Dependencywithintheschoolclusterswasaccounted for by adding school as a random factor, whereas dependency betweentherepeatedmeasureswasaccountedforbyﬁttingacom- poundsymmetrymatrixtotheresidualmatrices(thusassuming equal-sizedcorrelations between measurement occasions).Stu- dentswerenestedwithinschools,whichalsowasaccountedfor.

Thebaselinescorewasusedasacovariatetoadjustforimperfec- tionsintherandomizationprocedureandtoincreasethestatistical power.Thefixedfactorsweregroup(onecoefficientforthedif- ference between the intervention and the control group), time (acoefficientforeach timepointexceptthelast,thusdetecting a non-linear change), and group × time(to detect if interven- tioneffectswereparticularlypronouncedatcertaintimepoints).

In order toexamine ifthelevel of participations atworkshops inﬂuenced the outcomes, workshop attendance (WA-number of workshops) was added as linear covariate, as well as interac- tiontermsexaminingifWAinﬂuencedtheoutcomeparticularly at certain time points (WA × time) or additionally within just one of the groups (WA × time × group). The restricted maxi- mumlikelihood procedure and Type IIIF-tests werepreferred.

The analyseswere stratified for gender. Statistically significant effects setto p< .05, werefollowed-up withplanned compar- isontests (LSD)examininggroup differencesat each follow-up assessment. Resultsare expressedas absolutenumbers(n)and percentage(%) forcategoricaldata andmodelestimated means including95%confidenceintervalsandstandarddeviation(SD)for continuousdata.EffectsizesarepresentedasCohen’sdandphi- coefficients.

3. Results

3.1. Participantdemographics

Participant demographics for each group are presented in Table1.Atbaseline,allparticipantswere16–17yearsofage,with ameanBMIwithinthenormalweightrangeforyouths(Coleetal., 2000).ThebaselinecorrelationbetweenEESandKIDSCREEN-10 wasr=.60(p<.001)amongbothboysandgirls.Girlsintheinter- ventionhadhigherscoresonpositiveembodiment,health-related qualityoflife,andthegeneralhealthitemcomparedtogirlsinthe controlgroup.Nosigniﬁcantdifferencebetweengroupswasfound inboysfortheseoutcomemeasures.Basedonparents’totalincome andeducationlevel,girlsintheinterventiongroupweremorelikely tobedeﬁnedwithahighersocialeconomicstatuscomparedto girlsinthecontrolgroup.Boysintheinterventiongrouphadpar- entswithahigherlevelofeducation,andfewerwerecategorized asimmigrantscomparedtoboysinthecontrolgroup(Table1).The

(6)

Table2

ImmediateandFollow-upInterventionEffectsinPositiveEmbodimentSeparatelyforBoysandGirls.

Intervention Control

n TotalEESscore[CI95%] n TotalEESscore[CI95%] Meandifference[CI95%] p-value Cohen’sd Boys

Baseline 428 130.15[129.39,131.77] 220 126.73[124.72,128.17] 3.42[-0.05,6.90] .054

Post-intervention 268 136.93[135.23,138.63] 132 133.96[131.55,136.36] 2.98[0.03,5.93] .048 0.21 Follow-up(3months) 245 136.76[134.69,138.82] 132 135.58[132.56,138.59] 1.18[-2.48,4.84] .526

Follow-up(12months) 192 137.54[135.00,140.07] 94 133.99[130.17,137.81] 3.55[-1.04,8.13] .129 Girls

Baseline 696 117.31[116.26,117.00] 377 114.03[112.58,114.44] 3.45[0.53,6.37] .023 0.15

Post-intervention 534 123.80[122.74,124.87] 256 119.78[118.28,121.29] 4.02[2.17,5.87] <.001 0.35 Follow-up(3months) 536 124.89[123.59,126.18] 259 120.49[118.62,122.36] 4.40[2.12,6.68] <.001 0.31 Follow-up(12months) 459 125.54[124.06,127.03] 210 119.08[116.87,121.30] 6.46[3.79,9.13] <.001 0.42 Note.Allestimationswereadjustedforschoolasarandomfactor,andBMI,age,immigrationstatus,parents’income,andparents’educationasfixedcovariates(ifstatistically significant).p-value<.05.EES=TheExperienceofEmbodimentScale.ThebaselineEESscorewasincludedasacovariate.EESscorerange:34-170.CI95%=95%confidence interval.d=Cohen’sd,andarepresentedforsignificantdifferences.

Table3

ImmediateandFollow-upInterventionEffectsinHealth-RelatedQualityofLifeseparatelyforBoysandGirls.

Intervention Control

n Totalscore[CI95%] n Totalscore[CI95%] Meandifference[CI95%] p-value Cohen’sd Boys

KIDSCREEN-10^a

Baseline 418 38.29[38.05,38.73] 213 38.00[37.51,38.50] 0.29[-0.74,1.32] .580

Post-intervention 263 38.26[37.69,38.82] 127 37.48[36.68,38.28] 0.78[-0.20,1.76] .119 Follow-up(3months) 243 38.62[37.98,39.26] 128 37.65[36.72,38.58] 0.97[-0.16,2.10] .093 Follow-up(12months) 188 37.98[37.22,38.74] 89 36.84[35.71,37.97] 1.14[-0.22,2.50] .100 GeneralHealth^b

Baseline 418 3.70[3.64,3.77] 213 3.59[3.50,3.68] 0.10[-0.07,0.29] .580

Post-intervention 263 3.84[3.73,3.95] 127 3.61[3.45,3.77] 0.23[0.03,0.42] .021 0.25

Follow-up(3months) 243 3.72[3.59,3.84] 128 3.72[3.54,3.89] 0.00[-0.21,0.22] .989

Follow-up(12months) 188 3.78[3.63,3.92] 89 3.63[3.40,3.85] 0.15[-0.11,0.42] .256

Girls

KIDSCREEN-10^a

Baseline 692 35.78[35.61,36.13] 365 34.53[34.18,34.88] 1.25[0.49,2.01] .001 0.21

Post-intervention 530 34.82[34.41,35.22] 253 34.82[34.24,35.39] −0.00[-0.71,0.70] .999 Follow-up(3months) 532 34.87[34.42,35.32] 255 34.88[34.23,35.54] −0.01[-0.81,0.79] .980

Follow-up(12months) 446 34.88[34.38,35.38] 202 33.62[32.87,34.37] 1.26[0.36,2.16] .006 0.23 GeneralHealth^b

Baseline 692 3.30[3.24,3.33] 365 3.05[2.99,3.12] 0.24[0.11,0.38] <.001 0.23

Post-intervention 530 3.35[3.27,3.43] 253 3.18[3.07,3.29] 0.17[0.04,0.30] .013 0.19

Follow-up(3months) 532 3.29[3.21,3.37] 255 3.15[3.04,3.27] 0.14[-0.01,0.28] .059 0.15

Follow-up(12months) 446 3.39[3.30,3.48] 202 3.16[3.02,3.29] 0.23[0.07,0.40] .006 0.24

Note.Allestimationswereadjustedforschoolasarandomfactor,andBMI,age,immigrationstatus,parents’income,andparents’educationasfixedcovariates(ifstatistically significant).p-value<.05.KIDSCREEN-10=Health-relatedqualityoflife.ThebaselineKIDSCREEN-10scorewasincludedasacovariate.âKIDSCREEN-10scorerangesfrom 10-50.^bGeneralHealthscorerangesfrom1-5.CI95%=95%confidenceinterval.d=Cohen’sd,andarepresentedforsignificantdifferences.

linearmixedregressionmodelswereadjustedforgroupdifferences atbaseline.

3.2. Dropoutanalysis

Nodifferenceswereobservedintheoutcomevariablesbetween dropoutsandcompletersineitherboysorgirls.Morestudentsin thecontrolgroup(p=.001,␸=10.61),andmoreboys(p<.001,

␸=52.48)droppedout.Boyswhodroppedouthadslightlyhigher BMI(p=.044,d=0.15)andbodyweight(p=.010,d=0.20),while girlswhodroppedoutwereslightlyolder(p=.014,d=0.17).Effect analyseswerethereforeadjustedforthesevariables.

3.3. Positiveembodimentinterventioneffects

Forboys,thelinearmixedregressionmodelshowedthatthe maineffectofgroup(p=.072),time(p=.756)andtheinterac- tion effectof group × time (p =.543) were nonsigniﬁcant.The plannedcomparisonanalysesshowedthatboysintheintervention groupreportedhigherpositiveembodimentatpost-intervention comparedtoboysinthecontrolgroup,suggestingashort-term

favorablesmalleffect.However,thiseffectwaslostatthe3-and 12-monthfollow-ups(Table2).

For girls, the main effect of group was signiﬁcant, F(1, 777)=33.11,p<.001,whiletime(p=.267)andgroup× time(p

=.133)effectswerenonsigniﬁcant.Theplannedcomparisonanal- ysesshowedasigniﬁcantandfavorableeffectoftheintervention onpositiveembodimentforgirlsintheinterventiongroup.This effectwasmaintainedatthe3-and12-monthsfollow-up,respectively.Theeffectsizeincreasedslightlyovertime,andwithapeak atthelastfollow-upassessment(seeTable2).

3.4. Health-relatedqualityoflifeinterventioneffects

Forboys,thelinearmixedregressionmodelshowedasignif- icant maineffect ofgroupfor health-relatedquality oflife, F(1, 360)=4.78,p=.029,whilethetime(p=.148)andgroup×time(p= .871)effectswerenonsigniﬁcant.Althoughthemeandifferences betweenboysin theintervention andcontrol groups increased acrosstheassessmenttime-points,noplannedcomparisonanal- ysesshowedstatisticalsigniﬁcance(seeTable3).

(7)

Table4

Dose-ResponseAnalyseswithDegreeofAttendanceasaModeratorofPositiveEmbodiment(EES)InterventionEffects.

Degreeofattendance 0workshops MdiffCI95%

p(d)

1workshop MdiffCI95%

p(d)

2workshops MdiffCI95%

p(d)

3workshops MdiffCI95%

p(d)

Boys n=508 n=491 n=621 n=829

Post-intervention 5.54[-2.27,13.35]

.164

4.15[-4.11,12.40]

.324

0.08[-4.04,4.19]

.972

3.86[0.66,7.06]

.018(0.16) Follow-up(3months) −0.57[-9.26,8.13]

.898

2.34[-9.65,14.33]

.701

−0.21[-5.29,4.88]

.936

1.83[-2.16,5.82]

.369 Follow-up(12months) 2.51[-8.24,13.27]

.646

−3.83[-18.28,10.63]

.602

2.75[-3.13,8.64]

.358

4.26[-0.83,9.35]

.101

Girls n=635 n=630 n=838 n=1053

Post-intervention −0.32[-5.52-4.88]

.906

1.36[-2.61-5.33]

.501

3.86[1.43-6.30]

.002(0.23)

5.48[3.40-7.56]

<.001(0.32) Follow-up(3months) −1.37[-7.61-4.89]

.668

4.94[0.13-9.76]

.044(0.24)

4.13[1.15–7.11]

.007(0.20)

5.57[3.02-8.12]

<.001(0.26) Follow-up(12months) −2.16[-9.35-5.03]

.555

4.49[-0.79-9.78]

.095

7.28[3.84–10.71]

<.001(0.30)

7.75[4.77-10.72]

<.001(0.31)

Notes.EES=TheExperienceofEmbodimentScale.Mdiff=Meangroupdifference(apositivescorefavorstheintervention).CI95%=95%conﬁdenceinterval,p-value<.05.d= Cohen’sd,andarepresentedforsigniﬁcantdifferences.

For thegeneralhealth outcomeitem, themodel showedno effectofgroup(p=.120),time(p=.953),orgroup×time(p=.191) forboys.Theplannedcomparisonanalysesdidshowafavorable andsigniﬁcantpost-interventioneffectforboysintheinterven- tiongroupcomparedtoboysinthecontrolgroup,whichwasnot maintainedatfollow-up(seeTable3).

Forgirls,themaineffectofgroupforhealth-relatedqualityof lifewasnotsignificant(p=.186),whereas significanttime,F(2, 860)=3.99,p=.019,andgroup×time,F(2,860)=4.47,p=.012, effectswereobserved.Theplannedcomparisonanalysesshowed nosignificantdifferenceinhealth-relatedqualityoflifebetween girlsintheinterventionandcontrolgroupsatpost-intervention and3-monthsfollow-up.However,a“sleepingeffect”wasevident, asgirlsintheinterventiongrouphadasignificantlyhigherhealth- relatedqualityoflife(smalleffectsize)atthe12-monthsfollow-up comparedtogirlsinthecontrolgroup(seeTable3).

Themodelwiththegeneralhealthvariableasoutcomeshowed asignificantgroup effect,F(1,807) =10.54,p=.001, whilethe effectoftime(p=.466)and group×time(p=.598)werenon- significant.Theplannedcomparisonanalysesshowedthatgirlsin theintervention grouphadsignificantly morefavorablegeneral healthatpost-interventioncomparedtogirlsinthecontrolgroup (smalleffectsize),whichwasmaintainedatfollow-up,aswell(see Table3).

3.5. Dose-responseeffectrelatedtothenumberofattended workshops

Sincethedegreeofattendancewasirrelevantforthecontrol group,thegroupvariablewasrecodedas0(controlgroup),and 1–4(1=0workshopsininterventionstudent,2=1workshop,3=2 workshops,4=3workshops).Neither group(p= .290),time(p= .715),nortime×group(p=.750)weresignificantamongboysin theinterventiongroup.However,ingirls,themaineffectofgroup wassignificant,F(4,756)=10.96,p<.001.Thetime(p= .284)and theinteractioneffects(time×group)(p= .335)werenonsignif- icant.Thefollow-uptests,aspresentedinTable4,indicatethat anincreasingattendanceyieldedastrongerinterventioneffect.A noteworthyfindingwasthatboysandgirlsneededtoattendat leastthree andtwoworkshops,respectively,inordertobenefit fromtheHBIintervention.Thismoderationeffectwaslostamong boysatfollow-up,butnotamonggirls.Alleffectsizeswereinthe smallrange(seeTable4).Comparableanalysesonhealth-related qualityoflifeandthegeneralhealthvariablerevealednosignificant moderationeffects.

4. Discussion

TheHBIinterventionpromotedapost-intervention effecton positive embodiment and perceived general health for boys, althoughnosustainedeffectswereobserved.However,forgirls, theHBIinterventionpromotedimmediateandsustainedpositive embodiment.Additionally,forgirls,therewasaconsistentpattern ofimprovementinperceivedgeneralhealthatpost-intervention and12-monthsfollow-up,whereastheeffectsonhealth-related qualityoflifewereonlydemonstrated at12-monthsfollow-up.

Theseﬁndingsseemtoconvergewithotherbodyimageprograms thatincludefollow-upmeasures(Espinozaetal.,2013;Neumark- Sztaineretal.,2010).Theeffectsizesingirlswerealsostrongestat the12-monthsfollow-up,whichisnoteworthy.Thecurrentstudy increasestheknowledgebaseofthelong-termanddelayedeffect ofbodyimageinterventions,whichcurrentlyisscarce.Ourstudy emphasisestheimportanceoflong-termfollow-upsassomeinter- ventioneffectsmaymatureinaslowermanner.

Theinterventionwasintendedtofacilitateawarenessofhow attitudestowardsthebody andlifestylechoicesaretransmitted throughdifferentlearnedsocialchannels,and,throughthat,shape students’attitudes,feelings,andlifestyles.Accordingtoasocio- culturalperspective(Thompsonetal.,1999),anincreaseincritical awarenesscouldhaveimprovedtheabilitytowithstandunhealthy idealization, reducing the risk of internalization of such ideals (Teall&Piran,2012).Studentswerealsotaughttobecomeaware of,and use,factorsin everydaylifethat enhancetheirembodi- ment.Further,bodyfunctionalityandwell-beingwereemphasized, rather than appearance, when discussing lifestyle factors. This couldhavepromotedhealthyperspectivesonhowtoengagein lifestylebehaviours,similartopositiveembodimentcharacteristics (Tylka&Wood-Barcalow,2015).

TheHBIinterventionistoourknowledge,thefirstoneamong bodyimage interventionstoreportoneffects onhealth-related qualityof life. Ingirls, thediffusionof thehealth-relatedqual- ityoflifeeffectfromimprovingtheirembodimentwasexpected becausethesevariableshavebeenfoundtobehighlycorrelated (Griffiths etal.,2017; Haraldstadetal.,2011).Bystrengthening theabilitytofiltermediainformation,reduceunhealthycompar- isons,andpromotepositiveself-talk,itmightbeeasiertoimprove bodyacceptancewhichmaytransformintobetterpsychological well-being.Moreover,improvingself-careandahealthyconscious lifestyle,mayultimatelyimprovephysiologicalhealth,whichmay explaintheobservedimprovementsinhealth-relatedqualityoflife.

(8)

Theeffectsizeswereingeneralsmallandcomparablewithpre- viousstudies(Frankoetal.,2013;Halliwell,Jarman,McNamara, Risdon,&Jankowski,2015;Lindwall&Lindgren,2005; Morgan, Saunders,&Lubans,2012;Sharpe,Schober,Treasure,&Schmidt, 2013).In contrasttoclinicalstudies,theinterpretationofsmall effectsizesmaybemorefavourable.Thus,suchsmalleffectsizes arecommoninuniversalinterventions,andmaybeexpecteddueto lowbaseratesforclinicalsymptoms,andahighprobabilityofceil- ingeffectforpositivehealthindices.Similarly,bydeﬁnition,study variablesinhealthpromotionstudiesdonotpre-selectparticipants havingscoreswithinaclinicalrange(Wilksch,2014).

Attentionhasbeengivenintheliterature(Piran,2001)tohow studentsperceivethecredibilityofthosewhodeliverintervention programs.Inthepresentstudy,studentswereinformedaboutthe facilitators’educationandacademicposition.Inaddition,thefacil- itatorswereattentivetothequalityoftheirverbalandnon-verbal communicationwiththestudents.Nevertheless,thestudents’per- ceivedcredibilityoftheworkshopfacilitatorswasnotassessed.

AnexplicitrationalefortheHBIinterventionwastopromote theinteractionbetweenboysandgirls,andtomirrortheacross- gendersocioculturalinﬂuencesonbodyexperiencesthatoccurs inarealisticreal-lifesetting.Strategiestoaccomplishthisratio- naleincludedtheuseof differentinteractivecomponents,thus, toenhancethechanceofeffectinbothgenders.Our studyonly foundlong-termeffectsingirls.Thismaysupportprevioussug- gestionsthatgirlsaremorereceptivetobodyimageinterventions (Sticeetal.,2007)evenwheneffortshavebeenmadetomakethe interventiongenderneutral.Importantly,ourresultsdonotdoc- umentthatasingle-genderinterventionispreferred.Further,the HBIinterventionisahealthpromotionintervention,wheretheaim isnotonlytoreduceriskfactors,buttopromotehealth-relatedfac- tors.Basedonourﬁndings,amixed-genderapproachmighthave beenimportanttogirlsdespitethelackofeffectinboys.Tofur- therinvestigatewhethersingle-ormixed-genderapproaches is mosteffective,futurestudiesneedtoincludemorearms(control, mixed-gender,single-gender-group)intothestudydesign.

SimilartotheeffectsoftheHBIintervention,weakandtran- sienteffectsfroma bodyimageintervention hasbeenfoundin otherstudiesonyoungadultmen(Jankowskietal.,2017).Impor- tantly,althoughundocumented,thepresentersobservedthatthe boysfoundthetopicsof"comparison,""self-talk,"and"commu- nication"notasrelevantasthegirls,whichcouldhavemadeit moredifﬁculttobeengagedandreceptivetotheworkshopcontent.

Previousstudieshaveshownthatenhancingpeercomraderyand connection,andincludingmasculinepointsofreference,helped engageboys and men in an intervention (Seaton et al., 2017).

Perhapsthe female implementers in the HBIintervention may havehadchallengeswithpotentiallyimportantfactorstoengage boysas well as may have under-communicated the masculine aspects.

Virtually no effects among boys may also be explained by scoresabovenormdataforhealth-relatedqualityoflifeatbase- line(Ravens-Sieberer,2006).AlthoughnonormdatafortheEES existsforlateadolescentboys,onestudyonyoungmenshowed thatboysscoredsigniﬁcantlyhigherontheEEScomparedtogirls (Holmqvistetal.,2018).Thiscouldreﬂectthatboysatbaseline aremore acceptingof theirbodies,and therefore havea lower improvementpotentialcomparedtogirls.Atpresent,itremains unsettledwhethertheinterventionmayworkbetteramongboys withlowerbaselinehealth-relatedqualityoflifeandembodiment, andwhetheritmayworkequallywellinagirls-onlygroup.

Ourfindingscontradictthesuggestion(Wilksch,2017)thata single-session (workshop) intervention may suffice. Although a one-sessionmaybemorefeasibleinschoolsettings,ourresultsare inlinewiththeelaborationlikelihoodmodel(Petty&Bri ño,2012;

Petty&Cacioppo,1986)andpreviousmeta-analyses(Stice&Shaw,

2004;Sticeetal.,2007),thatatleasttwoworkshopsessionswere neededforgirlstomaintaintheinterventioneffectsatfollow-up.

4.1. Strengths,limitations,andfuturedirections

Assetsofthepresentstudyarethetheoreticalframework,the userinvolvementthroughapilotstudy,therandomizedcontrolled designandtheadequatestatisticalpower.However,alossofpower atthefollow-upsmayhaveincreasedtheprobabilityofTypeII errors, especially in boys. The fact that boyswho dropped out hadaslightlyhigherBMIisconsistentwithpreviousobservations inhealth-andbody image-relatedinterventionsandclassroom- basedactivities(Finn,Faith,&Seo,2018),thatthosewithhigher BMIfeelself-consciouswhenexposedtotheinterventioncontent.

However,boyswhodroppedoutdidnotdifferinpositiveembodi- mentorhealth-relatedqualityoflife,whichreducesthereasonsto believethatmanyofthosewhomightespeciallybeneﬁtfromour interventiondroppedout.

Drop-outsseemalmostinevitable,yetsomestepsmaybemen- tionedtocounteractthem.Althoughweusedmeasuresofpositive aspectsofbodyimageandnotmeasuresofbodydissatisfaction, careshouldbetakenwhenconsideringthecomprehensivenessof thequestionnaire,andtodecreasethenumberofincludedques- tions,notablythoseofasensitivenature.Tofacilitateimprovement potential,onechallengeistoselectoutcomemeasureswhereboth gendershaveroomforimprovement.

BeforeabroaderdisseminationoftheHBIintervention,modiﬁ- cationstotheworkshopsshouldbetested,withmalefacilitators,to furtherinvestigatewhetheritmightbepossibletoachievegenuine andsustainableeffectsforboys.Also,althoughthecredibilityof theworkshopholderswasplannedandfacilitatedfor,thestudents’

perceptionsofthiscredibilitywerenotassessed.Thisisalimitation, and futurestudiesshouldinclude suchassessment. In addition, thereisa needtostudythedismantlingpotentials.Thepresent ﬁndingsclearlyindicatethatamonggirls,twointeractiveandmul- ticomponentworkshopsmaysufﬁce.However,futurestudiesneed toaddresstheissueofwhichofthethreeworkshopsthatmaybe deletedfromtheprogram.Thiswouldinformwhichofthethemes (i.e.bodyimage,medialiteracy,andlifestylefactors)thatshould beretained.

4.2. Conclusion

TheHBIinterventionpromoteda post-interventioneffecton positive embodimentand perceivedgeneralhealth inboys.The interventionpromotedasustainedeffectonpositiveembodiment and health-relatedquality oflife ingirls.Futurestudies should examinetheeffectofonlytwoworkshopsforgirlsandmodiﬁca- tionsoftheworkshopsfor boystoseeifitispossibletoobtain sustainedeffectsinboysaswell.

Competinginterests

Theauthorsdeclarethattheyhavenocompetinginterests.

Funding

Thiswork wassupported byThe Norwegian Woman‘s Pub- licHealthAssociation(H1/2016),theNorwegianExtraFoundation forHealthandRehabilitation(2016/FO76521),andTINESA.The sponsorscameinafterthestudyprotocolwasdevelopedanddid nothaveanyroleindevelopmentofstudydesign,datacollection, analysisorinterpretationofdata,ormanuscriptwritingandsub- mission.