Smart-Phone Obesity Prevention Trial for AdolescentBoys in Low-Income Communities: The ATLAS RCT

WHAT’S KNOWN ON THIS SUBJECT: Adolescent males from low-income communities are a group at increased risk of obesity andrelated health concerns. Obesity prevention interventionstargeting adolescents have so far had mixed success. Targetedinterventions, tailored for specific groups, may be more appealingand efficacious.

WHAT THIS STUDY ADDS: A multicomponent school-basedintervention using smartphone technology can improve muscularfitness, movement skills, and key weight-related behaviors amonglow-income adolescent boys.

abstractOBJECTIVE: The goal of this study was to evaluate the impact of theActive Teen Leaders Avoiding Screen-time (ATLAS) intervention for ado-lescent boys, an obesity prevention intervention using smartphonetechnology.

METHODS: ATLAS was a cluster randomized controlled trial conductedin 14 secondary schools in low-income communities in New SouthWales, Australia. Participants were 361 adolescent boys (aged 12–14 years) considered at risk of obesity. The 20-week interventionwas guided by self-determination theory and social cognitive theoryand involved: teacher professional development, provision of fitnessequipment to schools, face-to-face physical activity sessions,lunchtime student mentoring sessions, researcher-led seminars,a smartphone application and Web site, and parental strategies forreducing screen-time. Outcome measures included BMI and waistcircumference, percent body fat, physical activity (accelerometers),screen-time, sugar-sweetened beverage intake, muscular fitness, andresistance training skill competency.

RESULTS: Overall, there were no significant intervention effects for BMI,waist circumference, percent body fat, or physical activity. Significant in-tervention effects were found for screen-time (mean 6 SE: –30 6 10.08min/d; P = .03), sugar-sweetened beverage consumption (mean: –0.6 60.26 glass/d; P = .01), muscular fitness (mean: 0.96 0.49 repetition; P =.04), and resistance training skills (mean: 5.7 6 0.67 units; P , .001).

CONCLUSIONS: This school-based intervention targeting low-incomeadolescent boys did not result in significant effects on bodycomposition, perhaps due to an insufficient activity dose. However, theintervention was successful in improving muscular fitness, movementskills, and key weight-related behaviors. Pediatrics 2014;134:e723–e731

AUTHORS: Jordan J. Smith, BEd,a Philip J. Morgan, PhD,a

Ronald C. Plotnikoff, PhD,a Kerry A. Dally, PhD,a Jo Salmon,PhD,b Anthony D. Okely, PhD,c Tara L. Finn,a and David R.Lubans, PhDa

aPriority Research Centre in Physical Activity and Nutrition,School of Education, University of Newcastle, Callaghan, NewSouth Wales, Australia; bCentre for Physical Activity and NutritionResearch, Deakin University, Burwood, Victoria, Australia; andcInterdisciplinary Educational Research Institute, University ofWollongong, Wollongong, New South Wales, Australia

KEY WORDSadolescent, intervention studies, obesity, physical activity, physicalfitness, randomized controlled trial, schools, sedentary lifestyle

ABBREVIATIONSATLAS—Active Teen Leaders Avoiding Screen-timeRT—resistance trainingSEIFA—Socio-Economic Indexes For AreasSSB—sugar-sweetened beverage

Mr Smith was involved in the study design and acquisition, analysis,and interpretation of data; he also participated in the drafting andrevision of the manuscript. Drs Morgan, Dally, Plotnikoff, Salmon,and Okely obtained funding for the study and were involved in thestudy concept and design; and Ms Finn was involved in theacquisition of data. Dr Lubans obtained funding for the study; wasinvolved in the study concept and design; participated in theanalysis and interpretation of data; and was involved in thedrafting and revision of the manuscript. All authors revised andapproved the final version of the manuscript as submitted.

This trial has been registered with the Australian and NewZealand Clinical Trials Registry (ACTRN 12612000978864;registration date was October 8, 2012 [www.anzctr.org.au]).

www.pediatrics.org/cgi/doi/10.1542/peds.2014-1012

doi:10.1542/peds.2014-1012

Accepted for publication Jun 20, 2014

Address correspondence to David R. Lubans, PhD, Priority ResearchCentre in Physical Activity and Nutrition, School of Education,Faculty of Education and Arts, University of Newcastle, Callaghan,NSW, Australia 2308. E-mail: david.lubans@newcastle.edu.au

PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).

Copyright © 2014 by the American Academy of Pediatrics

FINANCIAL DISCLOSURE: The authors have indicated they haveno financial relationships relevant to this article to disclose.

FUNDING: This study was funded by an Australian ResearchCouncil Discovery Project grant (DP120100611). The sponsor hadno involvement in the design or implementation of the study, inanalyses of data, or in the drafting of the manuscript.

POTENTIAL CONFLICT OF INTEREST: The authors have indicatedthey have no potential conflicts of interest to disclose.

COMPANION PAPER: A companion to this article can be found onpage e846, online at www.pediatrics.org/cgi/doi/10.1542/peds.2014-1940.

PEDIATRICS Volume 134, Number 3, September 2014 e723

ARTICLE

http://www.anzctr.org.aumailto:david.lubans@newcastle.edu.auhttp://www.pediatrics.org/cgi/doi/10.1542/peds.2014-1940http://www.pediatrics.org/cgi/doi/10.1542/peds.2014-1940

Although the global prevalence of obesityseems tohaveplateaued in recent years,1

the overall proportion of young peoplewho are overweight or obese remainshigh, particularly among those of lowsocioeconomic status.2 Considering theserious consequences of pediatric obe-sity,3 and the high likelihood of weightstatus tracking into adulthood,4 there isa strong rationale for targeting thehealth behaviors of adolescents.5–7

It has been recommended that obesityprevention efforts should be directed to-ward those most susceptible, such asadolescents living in low-income commu-nities.8 Adolescent boys of low socioeco-nomic status are particularly predisposedto unhealthy weight gain, and the globalprevalence of obesity is higher amongmale adolescents compared with fe-male adolescents.1 In addition, althoughadolescent boys are typically more ac-tive than girls,9 they are more likely toengage in high levels of recreationalscreen-time and consume large amountsof sugar-sweetened beverages (SSBs).9,10

However, apart from our pilot study,11

no interventions have specifically tar-geted adolescent boys from low-incomecommunities.

The challenges of modifying the healthbehaviors of adolescents and designingculturally appropriate interventionshave prompted researchers to explorethe utility of novel behavior changetechniques. Such strategies include theuse of e-health (ie, Internet-based) andmHealth (ie,mobile phone) technologiesto encourage young people to devel-op physical activity behavioral skills(ie, self-monitoring, goal setting)12,13 andimprove lifestyle behaviors.14 Mobilephone (and smartphone) ownershipamong young people is accelerating ata rapid rate.15,16 Although evidence forthe efficacy of mHealth interventions toimprove health behaviors in youngpeople is starting to emerge in thepublished literature,17,18 it is unlikelythat such interventions will provide the

“silver bullet” to the global obesitypandemic. Alternatively, they may havemore utility as adjuncts to face-to-facebehavior change interventions. To theauthors’ knowledge, no previous studyhas used smartphone technology ina school-based obesity prevention pro-gram14 and few existing smartphone“apps” include evidence-based behaviorchange techniques.19 Therefore, theprimary aim of the present study was toevaluate the effects of the multicompo-nent, school-based obesity preventionintervention incorporating smartphonetechnology, known as ATLAS (Active TeenLeaders Avoiding Screen-time). This ar-ticle reports the 8-month (immediatepostprogram) intervention effects.

METHODS

Study Design and Participants

Ethics approval for this study wasobtained from the human researchethics committees of the University ofNewcastle, Australia (July 3, 2012), andthe New South Wales Department ofEducationandCommunities(September6, 2012). School principals, teachers,parents, and study participants allprovided informed written consent. Thedesign, conduct, and reporting of thistrial adhere to the CONSORT statement.20

The rationale and study protocols havebeen reported previously.21 Briefly,ATLAS was evaluated by using a clusterrandomized controlled trial conductedin state-funded coeducational second-ary schools within low-income areas ofNew South Wales, Australia. The Socio-Economic Indexes For Areas (SEIFA) ofrelative socioeconomic disadvantage(scale: 1 = lowest to 10 = highest) wasused to identify eligible schools. Publicsecondary schools located in the New-castle, Hunter, and Central Coast re-gions of New South Wales with a SEIFAvalue of #5 (lowest 50%) were consid-ered eligible. All male students in theirfirst year at the study schools com-pleted a short screening questionnaire

to assess their eligibility for inclusion.Students failing to meet internationalphysical activity or screen-time guide-lines22 were considered eligible andwere invited to participate.

Sample Size and Randomization

Power calculations were conducted todetermine the required sample size fordetecting changes in the primary out-comes (ie, BMI, waist circumference).Baseline posttest correlations and SDestimates for BMI (r = 0.97, SD = 1.1) andwaist circumference (r = 0.96, SD = 11.6)were taken from our pilot study, andcalculations assumed a school clusteringeffect with an intraclass correlation of0.03.11 Based on 80% power, an a level of0.05, and a potential dropout rate of 20%,it was calculated that 350 participants (ie,25 from each school) would be requiredto detect a between-group difference inBMI of 0.4 kg.m22. In addition, the pro-posed sample size would be powered todetect a between-group difference of 1.5cm in waist circumference. After baselineassessments, schools were paired on thebasis of their geographic location, size,and SEIFA value and were randomized toeither the control or intervention group.Randomization was performed by an in-dependent researcher with the use ofa computer-based random number–producing algorithm.

Intervention

ATLASwas informed by the PALs (PhysicalActivity Leaders) pilot study,11,23,24 anda detailed description of the interventionis reported elsewhere.21 In summary,ATLAS is a multicomponent interventiondesigned to prevent unhealthy weightgain by increasing physical activity, re-ducing screen-time, and lowering SSBconsumption among adolescent boysattending schools in low-income areas.Self-determination theory25 and socialcognitive theory26 formed the theoreticalbasis of the program. Briefly, the in-tervention aimed to increaseautonomous

e724 SMITH et al

motivation for physical activity throughenhancing basic psychological needssatisfaction (ie, autonomy, competence,relatedness) during scheduled schoolsports. In addition, the intervention fo-cused on improving resistance training(RT) self-efficacy and also aimed todevelop self-regulatory skills (ie, self-monitoring and goal setting) to increaseincidental physical activity. Similarly,the intervention was designed to in-crease participants’ autonomous motiva-tion to limit screen-time27 by providinginformation regarding the consequencesof screen-time and strategies for self-regulation. ATLAS was aligned with cur-rent guidelines recommending that youthregularly engage in vigorous aerobic ac-tivities andphysical activities to strengthenmuscle and bone.22

The intervention was delivered fromDecember 2012 to June 2013 and in-volved a number of components that aredescribed in Table 1. The smartphoneapp was designed to supplement thedelivery of the enhanced school sportand interactive sessions by providingparticipants with a medium to monitorand track their behaviors, set goals, andassess their RT skill competency. In ad-dition, the app provided tailored moti-vational and informational messages via“push prompts.” The parental news-letters were designed to engage parentsand encourage them to manage theirchildren’s recreational screen-time.

The control group participated in usualpractice (ie, regularly scheduled schoolsports and physical education lessons)for the duration of the intervention butwill receive an equipment pack anda condensed version of the programafter the 18-month assessments.

Assessments and Measures

Trained research assistants completedbaseline data collection at the studyschools during November through De-cember 2012, at the same time of daywhenever possible. Follow-up assess-

ments were conducted 8 months frombaseline (immediate postintervention)andwill be conducted again at 18monthsfrom baseline (long-term follow-up). As-sessors were blinded to treatment allo-cation at baseline but not at follow-up.

Primary Outcome Measures

Height was recorded by using a porta-ble stadiometer (model no. PE087,Mentone Educational Centre, Moor-abbin, Victoria, Australia), and weightwas measured with a portable digital

scale (model no. UC-321PC, A&D Com-pany Ltd, Tokyo, Japan). BMI was cal-culated by using the standard equation(weight in kilograms/height in meterssquared). Waist circumference wasmeasured to the nearest 0.1 cm againstthe skin in line with the umbilicusby using a nonextendible steel tape(KDSF10-02, KDS Corporation, Osaka,Japan). Weight status was establishedfrom BMI z scores calculated by usingthe LMS method (World Health Orga-nization growth reference centiles).28

TABLE 1 Description and Dose of Intervention Components in the ATLAS Intervention

Intervention Component Dose Description

TeachersTeacher professional

developmentTwo 6-h

workshopsTeachers attend 2 professional development workshopsduring the study period (preprogram and mid-program).The workshops provide a rationale for the program, outlinethe intervention strategies (ie, program components,behavioral messages), and explain the theory behind theintervention.

One fitnessinstructorsession

Each school receives 1 visit during their regularly scheduledsport session fromapracticingfitness instructor (ie, personaltrainer). Thefitness instructorwill deliver thesessionwhile theteacher observes and completes the session observationchecklist.

ParentsParent newsletters Four

newslettersParents of study participants receive 4 newsletters containinginformationon thepotential consequences of excessive screenuse among youth, strategies for reducing screen-basedrecreation in the family home, and tips for avoiding conflictwhen implementing rules. They are also provided with theirchild’s baseline fitness test results.

StudentsResearcher-led

seminarsThree 20-min

seminarsParticipants attend 3 interactive seminars delivered bymembersof the research team. Seminars provide key informationsurrounding the program’s components and behavioralmessages, including current recommendations regardingyouth physical activity, screen-time, and resistance training,and also outline the student leadership component of theintervention.

Enhanced schoolsport sessions

Twenty 90-minsessions

Sport sessions are delivered by teachers at the study schools.Activities include elastic tubing resistance training, aerobic-and strength-based activities, fitness challenges, andmodifiedball games. Behavioral messages are reinforced during thecool-down period.

Lunchtime physicalactivity–mentoringsessions

Six 20-minsessions

Students participate in 6 lunchtime physical activity mentoringsessions. These self-directed sessions involve recruiting andinstructing grade 7 boys in elastic tubing resistance training.

Smartphone appand Web site

15 wk The smartphone app and Web site are used for physical activitymonitoring, recording of fitness challenge results, tailoredmotivationalmessaging, peerassessment of RT skills, andgoalsetting for physical activity and screen-time.

Pedometers 17 wk Participants are provided with pedometers for self-monitoring.Students are encouraged to set goals to increase their dailystep counts and monitor their progress using the pedometer.Pedometer step counts can also be entered into the smartphoneapp for review.

ARTICLE

PEDIATRICS Volume 134, Number 3, September 2014 e725

Secondary Outcome Measures

Body fat percent was determined byusing the Imp SFB7 bioelectrical im-pedanceanalyzer(ImpediMed, Ltd., EightMile Plains, Queensland, Australia).29

Physical activity was assessed accord-ing to standardized protocols30 usingActigraph accelerometers (model GT3X+ActiGraph, LLC, Fort Walton Beach, FL).Analyses for weekday physical activitywere performed for participants whowore their monitor for $600 minuteson at least 3 weekdays (Monday–Friday);analyses for weekend physical activityincluded participants who wore theirmonitor for$600 minutes on at least 1weekend day (Saturday–Sunday). Non–wear timewas defined as 30minutes ofconsecutive zeroes. Mean counts perminute were calculated to providea measure of overall activity, and thecut points proposed by Evenson et al31

were used to categorize intensity (ie,time spent in moderate to vigorousphysical activity). Hand grip dynamom-etry (Smedley’s dynamometer; TTM,Tokyo, Japan) and the 90-degree anglepush-up test29,32 provided a measure ofupper body maximal strength and localmuscular endurance, respectively. Rec-reational screen-time was self-reportedby using a modified form of the Ado-lescent Sedentary Activity Question-naire.33 Two items were used to assessconsumption of SSBs.9 Finally, RT skillcompetency was assessed by usingvideo analysis of the Resistance Train-ing Skills Battery.34,35 Participants per-formed 6movement skills considered tobe the foundation for more complexmovements used in RT programs.

Process Evaluation

A number of process measures wereused to determine the reach, imple-mentation, and participant and teachersatisfaction of the ATLAS intervention.The process evaluation included: (1)intervention implementation (ie, thepercentage of intended school sportssessions and lunchtime mentoring

sessions conducted by teachers); (2)school sport session fidelity determinedby using the ATLAS session observationchecklist (ie, compliance with the pro-posed session structure and activities,recorded by a member of the researchteam); (3) attendance at sessions; (4)engagement with intervention com-ponents (eg, smartphone app, pedo-meters); and (5) program satisfaction(ie, responses to a postprogram evalu-ation questionnaire).

Statistical Analysis

All analyseswereconducted inDecember2013 by using SPSS for Windows version20.0 (IBM SPSS Statistics, IBM Corpora-tion, Armonk, NY; 2010) with a levels setat P, .05; data were assessed for nor-mality. Intervention effects for the pri-mary and secondary outcomes wereexamined by using linear mixed modelsadjusted for school clustering and par-ticipant socioeconomic status, and allanalyses followed the intention-to-treatprinciple.36 Prespecified subgroup anal-yses21 for all body composition outcomeswere conducted for those classified asoverweight/obese (combined as a singlegroup) at baseline. In addition, the pro-portional difference between treatmentgroups among those improving theirweight status (ie,moving from “obese” to“overweight” or from “overweight” to“healthy weight”) or regressing toa poorer weight status (ie, moving from“healthy weight” to “overweight” or from“overweight” to “obese”) was exploredby using Pearson’s x2 test.

RESULTS

The flow of participants through thestudy is reported in Fig 1. Fourteenschools were recruited, and 361 boys(mean age: 12.7 6 0.5 years) were as-sessed at baseline (Table 2). Follow-upassessments at 8 months were com-pleted for 154 (85.6%) control groupparticipants and 139 (76.8%) interven-tion group participants, representing

an overall retention rate of 81.2% frombaseline. Participants who did notcomplete follow-up assessments weremore active on weekdays (P = .03) andweekends (P = .01). There were no sig-nificant differences for body composi-tion outcomes.

Changes in Body Composition

Changesforalloutcomesarereported inTable 3. No intervention effects werefound for the primary outcomes of BMIand waist circumference or for percentbody fat. Changes in BMI (mean6 SE:–0.4 kg.m22 6 0.26; P = .15), waistcircumference (mean: –0.56 0.95 cm;P = .57), and percent body fat (mean:–0.9% 6 0.77%; P = .22) for thoseclassified as overweight/obese at base-line were all in favor of the interventiongroup. However, these effects were notstatistically significant. Of the 19 partic-ipants who improved their weight sta-tus, 13 (68%) were in the interventiongroup; of the 9 participants who re-gressed to a more unhealthy weightstatus, only 1 (11%) was in the inter-vention group. Pearson’s x2 test indi-cated a significant difference in favor ofintervention boys: x2 (2) = 8.08, P = .02.

Changes in Behavioral Outcomes

No significant differences were observedfor overall activity (mean counts perminute) ormoderate to vigorous physicalactivity. However, intervention boysreported less screen-time (mean:230610.08 min/d; P = .03) and SSB consump-tion (mean:20.66 0.26 glass/d; P = .01)than control boys at follow-up.

Changes in Fitness and SkillOutcomes

There was a significant intervention ef-fect for upperbodymuscularendurance(mean: 0.96 0.49 repetition; P = .04). Inaddition, a significant between-groupdifference was observed for RT skillcompetency in favor of interventionboys (mean: 5.76 0.67 units; P, .001).

e726 SMITH et al

Process Evaluation

No adverse events or injuries werereported during the school sports ses-sions, lunchtime leadership sessions, orassessments. On average, schools con-ducted 79% 6 15% of intended schoolsports sessions and 64% 6 40% ofintended lunchtime sessions. Four sportsession observations (2 per schoolterm) were conducted at each school.Adherence to the proposed sessionstructure at observations 1, 2, 3, and 4was 61%, 58%, 90%, and 96%, re-spectively. Students were expected toattend at least 70%of sport sessions andat least two-thirds of lunchtime sessions.Sixty-five percent of boys attended$70% of the sport sessions but only44% of boys attended at least two-thirdsof lunchtime sessions. Participant sat-isfactionwith the ATLAS interventionwashigh (mean: 4.5 6 0.7 [scale of 1 =strongly disagree to 5 = strongly agree]).Students enjoyed the sports sessions(mean: 4.5 6 0.7); however, satisfaction

with the lunchtime sessions was some-what lower (mean: 3.76 1.0).

A detailed evaluation of the smartphoneapp can be found elsewhere.37 Briefly,smartphone (or similar device) owner-ship was reported by 70% of boys, and63% reported using either the iPhone orAndroid version of the ATLAS app. Thosestudents who did not have access toa smartphone could access the samefeatures via the ATLAS Web site. Almostone-half of the group agreed or stronglyagreed that the “push prompt” mes-sages reminded them to be more ac-tive, reduce their screen-time, anddrink fewer sugary drinks, and 44% ofparticipants agreed or strongly agreedthat the ATLAS app was enjoyable touse. Self-reported pedometer use wasmoderate, with 44% of boys wearingtheir pedometer sometimes and 30%wearing their pedometer often. In ad-dition, all 4 newsletters were sent to86% of parents. Teacher satisfactionwith the intervention was high (mean:

4.4 6 0.5), and they reported enjoyingboth the preprogram (mean: 5.06 0.0)and mid-program (mean: 4.9 6 0.4)professional development workshops.

DISCUSSION

The goal of the present study was todeterminetheeffectivenessof theschool-based ATLAS intervention for adolescentboys. No significant intervention effectswere observed overall for body compo-sition. However, for those who wereoverweight/obese at baseline, there wasa trend in favor of intervention partic-ipants for all body composition out-comes. Significant intervention effectswere found for secondary outcomes,including upper body muscular endur-ance, RT skill competency, self-reportedscreen-time, and SSB consumption.

The intervention effects for body com-positionoutcomeswerenegligible,whichis similar to the findings of a trial in-volving Dutch teenagers.38 Our inclusioncriteria aimed to identify boys at in-creased risk of obesity based on theirphysical activity and screen behaviors.This approach was selected to reducethe potential for weight stigmatization,which may occur if inclusion is contin-gent on participants’ BMI. However, it ispossible that by using these broad in-clusion criteria, our ability to see signif-icant improvements in anthropomorphicmeasures was minimized, as a numberof “healthy weight” boys with little scopefor change were included in the study.Indeed, the majority of recruited boyswere classified as having a healthyweight at baseline and remained so forthe duration of the intervention. In-terestingly, it has been suggested thatwhile school-based interventions shouldcontinue to target all students, analysisof the primary outcome(s) should per-haps focus on overweight/obese youth.39

Thefindings of the present studywere incontrast to those of our pilot study inwhich significant interventioneffects formultiple measures of body composition

FIGURE 1Flow of participants through the study process.

ARTICLE

PEDIATRICS Volume 134, Number 3, September 2014 e727

were observed.11 This inconsistencycould be due to differences in thequantity and intensity of physical activ-ity during the enhanced school sportsessions. Process data indicated thattoward the end of the program, boys inthe PALs study became disengaged dueto the lack of variety in activities. Tomaintain engagement, the ATLAS sportsessions provided a greater variety ofactivities and also incorporated astronger focus on movement skill de-velopment. Although program satisfac-tion was higher in ATLAS compared withPALs, these modifications may haveresulted in lower overall activity and/orlower activity intensity during the ses-sions, and hence smaller effects onbody composition. Alternatively, be-cause PALs participants had a higherbaseline BMI, they may have hada greater propensity for change.

Although our study was not powered todetect subgroup differences, changes in

body composition outcomes favored in-tervention boys who were overweight orobese at baseline. The magnitude ofthese changes, although not statisticallysignificant, may nonetheless be clini-cally meaningful. For example, the ad-justed mean difference in body fat foroverweight/obese participants in theATLAS intervention was 0.9%. Accordingto Dai et al,40 an increase of 1% body fatis significantly associated with unfavor-able changes in total, high-density lipo-protein, and low-density lipoproteincholesterol, as well as triglycerides.Furthermore, in a study of children andadolescents, Weiss et al41 reported thateach 0.5-unit increase in BMI was asso-ciated with significantly increased riskof themetabolic syndrome. The adjustedmean difference in BMI for overweight/obese subjects in our study was –0.4units, which may have clinical signifi-cance. Finally, the proportional shift inweight status between study groups

provides additional support for the ef-ficacy of the intervention for overweight/obese participants.

Recent literaturehas identifiedmuscularfitness as an important indicator ofhealth status for young people.42,43 No-tably, we found significant interventioneffects for upper body muscular endur-ance and RT skill competency. The in-tervention activitieswere predominantlyresistance-based and as such focusedon developing muscular fitness. Fur-thermore, the workouts and fitnesschallenges performed throughout theintervention were designed to be highrepetition, targeting local muscular en-durance rather than maximal strengthspecifically. Therefore, the significantimprovement in muscular enduranceand nonsignificant findings for mus-cular strength are not surprising. Inaddition, the improvement in skill com-petency was expected because a corecomponent of the sport sessions wastime dedicated to RT skill developmentduring which teachers modeled correctexercise technique and provided cor-rective feedback on boys’ movementskill performance. Furthermore, ap-proximately two-thirds of boys reportedusing the app to assess and monitortheir RT technique.

Intervention boys in our study reportedspending30minuteslessperdayengagedin screen-based recreation at follow-upcompared with control subjects. Similarfindings were described in the PlanetHealth intervention,44 with the authorsreporting an adjusted difference of 24minutes in favor of intervention boys.The reduction in screen-time observed inATLAS is likely to be conservative com-paredwith other studies, as ourmeasureof screen-time was modified to accountfor screenmultitasking. Reducing screen-time was an explicit intervention target,and ATLAS used a number of strategiesto encourage boys to reduce theirscreen-time. The relative contributionof the individual intervention components

TABLE 2 Baseline Characteristics of Study Sample

Characteristics Control (n = 180) Intervention (n = 181) Total (N = 361)

Age, y 12.7 6 0.5 12.7 6 0.5 12.7 6 0.5Born in Australia 168 (93.3) 174 (96.1) 341 (94.7)English language spoken at homea 169 (94.4) 175 (96.7) 344 (95.6)Cultural backgroundb

Australian 132 (73.7) 145 (80.6) 277 (77.2)European 31 (17.3) 22 (12.2) 53 (14.8)African 6 (3.4) 1 (0.6) 7 (1.9)Asian 3 (1.7) 4 (2.2) 7 (1.9)Middle Eastern 2 (1.1) 0 2 (0.6)Other 5 (2.8) 8 (4.4) 13 (3.6)

Socioeconomic positionc

1–2 55 (30.9) 49 (27.1) 104 (29.0)3–4 81 (45.5) 120 (66.3) 201 (56.0)5–6 27 (15.2) 4 (2.2) 31 (8.6)7–8 8 (4.5) 8 (4.4) 16 (4.5)9–10 7 (3.9) 0 7 (1.9)

Weight, kg 53.1 6 13.4 54.0 6 15.0 53.5 6 14.2Height, cm 160.2 6 8.4 160.9 6 9.0 160.5 6 8.7BMI 20.5 6 4.1 20.5 6 4.1 20.5 6 4.1Weight statusUnderweight 5 (2.8) 2 (1.1) 7 (1.9)Healthy weight 115 (63.9) 110 (60.8) 225 (62.3)Overweight 38 (21.1) 39 (21.5) 77 (21.3)Obese 22 (12.2) 30 (16.6) 52 (14.4)

Waist circumference, cm 76.5 6 12.3 76.2 6 12.2 76.3 6 12.2

Data are presented as mean 6 SD or n (%).a One participant did not report language spoken at home.b Two participants did not report cultural background.c Socioeconomic position determined according to population decile by using SEIFA of relative socioeconomic disadvantagebased on residential postal code (1 = lowest, 10 = highest). Two participants did not report residential postal code.

e728 SMITH et al

to change in screen-time is difficult toascertain. However, the consequencesof excessive screen-time and currentscreen-time guidelines were made ex-plicit to boys during the researcher-ledseminars and were reinforced byteachers during the face-to-face sportsessions. In addition, the majority ofparents received and read the screen-time newsletters, as reported by theboys. Finally, 70% of boys reported us-ing the goal-setting function of the app,which allowed users to set goals forreducing screen-time.

In addition to these effects on screen-time, intervention boys also reportedsignificantly reducing their consump-tion of SSBs. The adjusted mean dif-ference was 0.6 glass per day (∼150mL). A reduction in the consumption ofSSBs has been recommended to pre-vent unhealthy weight gain and theonset of metabolic disorders.48 Al-though improvements in body composi-tion have accompanied reductions inSSB consumption in previous studies,45,46

these studies were of longer durationthan ATLAS and also focused solely on

this outcome. If the reduction in SSBconsumption observed in our study issustained, the corresponding decreasein daily energy intake may have a con-siderable impact on body compositionover the longer term.

Although it is difficult to determine therelative contribution of individual compo-nents inmulticomponent interventions,by conducting a comprehensive pro-cess evaluation we were able to gatherimportant informationon theefficacyofindividual strategies. Attendance at thesport sessions was reasonable, with

TABLE 3 Changes in Primary and Secondary Outcomes

Outcomea Baseline 8 Months Change P Value Adjusted Differencein Change

P Value

BMIIntervention 20.7 6 0.64 21.3 6 0.64 0.60 6 0.09 ,.001 0.0 6 0.12b .84Control 20.6 6 0.57 21.2 6 0.57 0.61 6 0.08 ,.001

Waist circumference, cmIntervention 77.1 6 1.89 77.1 6 1.89 0.0 6 0.33 .98 0.5 6 0.45b .16Control 77.0 6 1.69 76.5 6 1.69 20.5 6 0.31 .10

Body fat, %Intervention 20.3 6 1.27 21.6 6 1.28 1.3 6 0.35 ,.001 0.0 6 0.48 .99Control 22.5 6 1.14 23.8 6 1.14 1.3 6 0.33 ,.001

Grip strength, kgIntervention 22.5 6 0.97 28.5 6 0.98 6.0 6 0.32 ,.001 0.5 6 0.45 .30Control 20.4 6 0.87 25.9 6 0.88 5.5 6 0.31 ,.001

Push-ups (repetitions)Intervention 9.1 6 0.99 9.8 6 1.0 0.7 6 0.35 .04 0.9 6 0.49b .04Control 6.6 6 0.89 6.5 6 0.89 20.1 6 0.34 .73

Weekday PA, counts/minc

Intervention 538 6 30.81 515 6 33.51 223 6 18.08 .21 219 6 23.30 .41Control 477 6 27.18 473 6 28.58 23 6 14.69 .81

Weekend PA, counts/mind

Intervention 435 6 47.19 410 6 54.85 225 6 40.25 .53 28 6 53.94b .57Control 404 6 42.42 387 6 47.13 217 6 35.97 .64

Weekday MVPA, %c

Intervention 8.6 6 0.58 8.3 6 0.63 20.4 6 0.34 .28 20.7 6 0.44 .14Control 7.5 6 0.51 7.8 6 0.54 0.3 6 0.28 .30

Weekend MVPA, %d

Intervention 6.2 6 0.78 6.0 6 0.90 20.2 6 0.67 .73 20.16 0.90b .80Control 5.8 6 0.70 5.7 6 0.78 20.1 6 0.60 .82

Screen-time, min/dIntervention 109 6 14.18 112 6 14.52 3 6 7.25 .67 230 6 10.08b .03Control 132 6 12.78 165 6 12.94 33 6 7.0 ,.001

SSB intake, glasses/dIntervention 3.9 6 0.40 3.1 6 0.41 20.8 6 0.19 ,.001 20.6 6 0.26b .01Control 3.9 6 0.36 3.7 6 0.36 20.1 6 0.18 .44

RT skill competencye

Intervention 31.7 6 0.56 40.1 6 0.60 8.4 6 0.48 ,.001 5.7 6 0.67 ,.001Control 30.7 6 0.53 33.4 6 0.55 2.7 6 0.46 ,.001

Means 6 standard errors are reported for all outcomes. MVPA, moderate to vigorous physical activity; PA, physical activity.a All models were adjusted for school clustering and participant socioeconomic status.b Variable transformed for analysis.c A total of 240 and 120 participants wore accelerometers on weekdays at baseline and posttest, respectively.d A total of 120 and 83 participants wore accelerometers on weekend days at baseline and posttest, respectively.e Possible values range from 0 to 56.

ARTICLE

PEDIATRICS Volume 134, Number 3, September 2014 e729

approximately two-thirds of boys at-tending a satisfactory number of ses-sions, whereas attendance at thelunchtime sessions was poor. Boysreported lower satisfaction with thelunchtime sessions, which may be dueto a preference to use this period forsocializing. Compliance with the in-tended session structure wasmoderateat the first observation but improvedsubstantially over the course of theintervention. Usage of the ATLAS app forself-monitoring and goal setting wasmoderate. Therefore, additional strate-gies and features may be needed to en-hance engagement in adolescent boys. Itis important tonote that theproportionofdropouts who were overweight/obesewas lower than it was for completers,indicating that ATLAS was successful inretaining overweight/obese boys. Finally,all teachers agreed or strongly agreedthat their students benefited from in-volvement in ATLAS, thus providinga strong endorsement for the program.

Strengths of the present study includethe randomized controlled design, theidentification and targeting of ado-lescents at risk of obesity, objectiveassessment of physical activity, theextensive process evaluation, and thehigh retention at follow-up. However,there were also some limitations. Al-thoughBMI is considereda suitable andstablemeasureof change inadiposity,47

direct measures such as dual-energy

radiograph absorptiometry provide amore accurate assessment of bodyfat. Second, we cannot rule out socialdesirability bias in our assessment ofscreen-time and SSB consumption.Third, we were unable to collect ATLASapp usage data, which prevented amore thorough examination of the ef-ficacy of this novel component. Fourth,similar to previous studies with ado-lescents,12 poor compliance to accelero-meter protocols reduced the availablesample size, preventing more compre-hensive assessment of change in physi-cal activity. Finally, due to the targetednature of the intervention, the resultsmay not be generalizable to other groups(eg, female subjects, those from othersocioeconomic strata).

CONCLUSIONS

There is a clear need for innovativeobesity prevention programs that tar-get adolescents at risk of obesity.School-based interventions that usesmartphone technology have the po-tential for health behavior change, butstrategies for identifying and recruitingparticipants and increasing the in-tervention dose are needed. Althoughthe ATLAS program failed to achieveshort-term changes in body composi-tion in the overall study sample, therewas a trend in favor of overweight/obese boys. In addition, there were fa-

vorable outcomes for behaviors knownto be associated with adiposity andcardiometabolic disorders. This studydemonstrates that a school-based in-tervention targeting economically dis-advantaged adolescent boys can havea favorable impact onmuscular fitness,movement skills, and key weight-relatedbehaviors. We encourage practitionersand policy makers to advocate for tar-geted programs in schools for youngpeople who are disengaged in currentphysical education programs. Futureinterventions using smartphone tech-nology should capture objective data onapp/Web site usage throughout the in-tervention period, and analyses shouldbe conducted examining its associa-tion with changes in intervention out-comes. Furthermore, futuresmartphoneapps should integrate stimulatingfeatures such as social media linkageand “gamification” to support ongo-ing engagement with this interventioncomponent.

ACKNOWLEDGMENTSWethankSarahKennedy, EmmaPollock,and Mark Babic for their assistancewith data collection. In addition, wethank Geoff Skinner and AndrewHarvey for their assistance with theATLAS smartphone application. Finally,we thank the schools, teachers, pa-rents, and study participants for theirinvolvement.

REFERENCES

1. Ng M, Fleming T, Robinson M, et al. Global,regional, and national prevalence of over-weight and obesity in children and adultsduring 1980–2013: a systematic analysisfor the Global Burden of Disease Study2013. Lancet. 2014; doi: 10.1016/S0140-6736(14)60460-8

2. Stamatakis E, Wardle J, Cole TJ. Childhoodobesity and overweight prevalence trends inEngland: evidence for growing socioeconomicdisparities. Int J Obes (Lond). 2010;34(1):41–47

3. Tsiros MD, Coates AM, Howe PR, GrimshawPN, Buckley JD. Obesity: the new childhooddisability? Obes Rev. 2011;12(1):26–36

4. Singh AS, Mulder C, Twisk JW, van MechelenW, Chinapaw MJ. Tracking of childhood over-weight into adulthood: a systematic review ofthe literature. Obes Rev. 2008;9(5):474–488

5. Strong WB, Malina RM, Blimkie CJ, et al.Evidence based physical activity for school-age youth. J Pediatr. 2005;146(6):732–737

6. Boone JE, Gordon-Larsen P, Adair LS, PopkinBM. Screen time and physical activity duringadolescence: longitudinal effects on obesityin young adulthood. Int J Behav Nutr PhysAct. 2007;4(1):26

7. Maher C, Olds TS, Eisenmann JC, Dollman J.Screen time is more strongly associated

than physical activity with overweight andobesity in 9- to 16-year-old Australians.Acta Paediatr. 2012;101(11):1170–1174

8. Olsen NJ, Mortensen EL, Heitmann BL. Predis-position to obesity: should we target those mostsusceptible? Curr Obes Rep. 2012;1(1):35–41

9. Hardy LL, King L, Espinel P, Cosgrove C,Bauman A. NSW Schools Physical Activityand Nutrition Survey (SPANS) 2010: Full Re-port. Sydney: NSW Ministry of Health; 2011.Available at: www0.health.nsw.gov.au/pubs/2011/spans_full.html. Accessed June 1, 2014

10. Miller PE, McKinnon RA, Krebs-Smith SM, et al.Sugar-sweetened beverage consumption in

e730 SMITH et al

http://www0.health.nsw.gov.au/pubs/2011/spans_full.htmlhttp://www0.health.nsw.gov.au/pubs/2011/spans_full.html

the US: novel assessment methodology. AmJ Prev Med. 2013;45(4):416–421

11. Lubans DR, Morgan PJ, Aguiar EJ, Callister R.Randomized controlled trial of the PhysicalActivity Leaders (PALs) program for adoles-cent boys from disadvantaged secondaryschools. Prev Med. 2011;52(3-4):239–246

12. Dewar DL, Morgan PJ, Plotnikoff RC, et al.The nutrition and enjoyable activity for teengirls study: a cluster randomized controlledtrial. Am J Prev Med. 2013;45(3):313–317

13. Lubans DR, Morgan PJ, Okely AD, et al.Preventing obesity among adolescent girls:one-year outcomes of the nutrition andenjoyable activity for teen girls (NEAT Girls)cluster randomized controlled trial. ArchPediatr Adolesc Med. 2012;166(9):821–827

14. Bort-Roig J, Gilson ND, Puig-Ribera A, Con-treras RS, Trost SG. Measuring and influ-encing physical activity with smartphonetechnology: a systematic review. SportsMed. 2014;44(5):671–686

15. Lenhart A. Teens, Smartphones & Texting.Pew Internet & American Life ProjectWashington, DC: Pew Research Center; 2012

16. Weerakkody ND. Mobile phones and chil-dren: an Australian perspective. Paperpresented at: INSITE 2008: Informing Sci-ence + Information Technology Education;June 22–25, 2008; Varna, Bulgaria

17. Schiel R, Thomas A, Kaps A, Bieber G. An in-novative telemedical support system tomeasure physical activity in children andadolescents with type 1 diabetes mellitus. ExpClin Endocrinol Diabetes. 2011;119(9):565–568

18. Hingle M, Nichter M, Medeiros M, Grace S.Texting for health: the use of participatorymethods to develop healthy lifestyle messagesfor teens. J Nutr Educ Behav. 2013;45(1):12–19

19. Schoffman DE, Turner-McGrievy G, Jones SJ,Wilcox S. Mobile apps for pediatric obesityprevention and treatment, healthy eating,and physical activity promotion: just fun andgames? Transl Behav Med. 2013;3(3):320–325

20. Campbell MK, Elbourne DR, Altman DG;CONSORT group. CONSORT statement: ex-tension to cluster randomised trials. BMJ.2004;328(7441):702–708

21. Smith JJ, Morgan PJ, Plotnikoff RC, et al.Rationale and study protocol for the ‘ActiveTeen Leaders Avoiding Screen-time’ (ATLAS)group randomized controlled trial: an obe-sity prevention intervention for adolescentboys from schools in low-income communi-ties. Contemp Clin Trials. 2014;37(1):106–119

22. World Health Organization. Global Recom-mendations on Physical Activity for Health.Geneva, SwitzerlandWorld Health Organiza-tion; 2010

23. Lubans DR, Morgan PJ, Callister R. Potentialmoderators and mediators of intervention

effects in an obesity prevention programfor adolescent boys from disadvantagedschools. J Sci Med Sport. 2012;15(6):519–525

24. Morgan PJ, Saunders KL, Lubans DR. Im-proving physical self-perception in adoles-cent boys from disadvantaged schools:psychological outcomes from the PhysicalActivity Leaders randomized controlled tri-al. Pediatr Obes. 2012;7(3):e27–e32

25. Deci E, Ryan R. Intrinsic Motivation andSelf-Determination in Human Behaviour.New York, NY: Plenum; 1985

26. Bandura A. Social Foundations of Thoughtand Action: A Social Cognitive Theory. Engle-wood Cliffs, NJ: Prentice-Hall; 1986

27. Lubans DR, Lonsdale C, Plotnikoff RC, SmithJ, Dally K, Morgan PJ. Development andevaluation of the Motivation to LimitScreen-time Questionnaire (MLSQ) foradolescents. Prev Med. 2013;57(5):561–566

28. de Onis M, Onyango AW, Borghi E, Siyam A,Nishida C, Siekmann J. Development ofa WHO growth reference for school-agedchildren and adolescents. Bull WorldHealth Organ. 2007;85(9):660–667

29. Lubans DR, Morgan P, Callister R, et al. Test-retest reliability of a battery of field-basedhealth-related fitness measures for ado-lescents. J Sports Sci. 2011;29(7):685–693

30. Trost SG, McIver KL, Pate RR. Conductingaccelerometer-based activity assessmentsin field-based research. Med Sci SportsExerc. 2005;37 (suppl 11):S531–S543

31. Evenson KR, Catellier DJ, Gill K, Ondrak KS,McMurray RG. Calibration of two objectivemeasures of physical activity for children. JSports Sci. 2008;26(14):1557–1565

32. Cooper Institute for Aerobics Research.Fitnessgram: Test Administration Manual.Champaign, IL: Human Kinetics; 1999

33. Hardy LL, Booth ML, Okely AD. The reliabilityof the Adolescent Sedentary Activity Ques-tionnaire (ASAQ). Prev Med. 2007;45(1):71–74

34. Lubans DR, Smith JJ, Harries SK, BarnettLM, Faigenbaum AD. Development, test-retest reliability, and construct validity ofthe Resistance Training Skills Battery. JStrength Cond Res. 2014;28(5):1373–1380

35. Barnett L, Reynolds J, Faigenbaum AD,Smith JJ, Harries S, Lubans DR. Rateragreement of a test battery designed toassess adolescents’ resistance trainingskill competency [published online aheadof print December 17, 2013]. J Sci MedSport. doi: 10.1016/j.jsams.2013.11.012

36. White IR, Carpenter J, Horton NJ. Includingall individuals is not enough: lessons forintention-to-treat analysis. Clin Trials. 2012;9(4):396–407

37. Lubans DR, Smith JJ, Skinner G, Morgan PJ.Development and implementation ofa smartphone application to promote phys-ical activity and reduce screen-time in ado-lescent boys. Front Public Health. 2014;2:42

38. Singh AS, Chin A Paw MJ, Brug J, vanMechelen W, van Mechelen W. Dutch obesityintervention in teenagers: effectiveness ofa school-based program on body compo-sition and behavior. Arch Pediatr AdolescMed. 2009;163(4):309–317

39. Marcus MD, Hirst K, Kaufman F, Foster GD,Baranowski T. Lessons learned from theHEALTHY primary prevention trial of riskfactors for type 2 diabetes in middle schoolyouth. Curr Diab Rep. 2013;13(1):63–71

40. Dai S, Fulton JE, Harrist RB, Grunbaum JA,Steffen LM, Labarthe DR. Blood lipids inchildren: age-related patterns and as-sociation with body-fat indices: ProjectHeartBeat! Am J Prev Med. 2009;37(suppl1):S56–S64

41. Weiss R, Dziura J, Burgert TS, et al. Obesityand the metabolic syndrome in childrenand adolescents. N Engl J Med. 2004;350(23):2362–2374

42. Peterson MD, Saltarelli WA, Visich PS,Gordon PM. Strength capacity and car-diometabolic risk clustering in adolescents.Pediatrics. 2014;133(4). Available at: www.pediatrics.org/cgi/content/full/133/4/e896

43. Smith JJ, Eather N, Morgan PJ, PlotnikoffRC, Faigenbaum AD, Lubans DR. The healthbenefits of muscular fitness for childrenand adolescents: a systematic review andmeta-analysis [published online ahead ofprint May 1, 2014]. Sports Med.

44. Gortmaker SL, Peterson K, Wiecha J, et al.Reducing obesity via a school-based in-terdisciplinary intervention among youth:Planet Health. Arch Pediatr Adolesc Med.1999;153(4):409–418

45. James J, Thomas P, Kerr D. Preventingchildhood obesity: two year follow-upresults from the Christchurch obesity pre-vention programme in schools (CHOPPS).BMJ. 2007;335(7623):762

46. Ebbeling CB, Feldman HA, Chomitz VR, et al.A randomized trial of sugar-sweetenedbeverages and adolescent body weight. NEngl J Med. 2012;367(15):1407–1416

47. Cole TJ, Faith MS, Pietrobelli A, Heo M. Whatis the best measure of adiposity change ingrowing children: BMI, BMI %, BMI z-scoreor BMI centile? Eur J Clin Nutr. 2005;59(3):419–425

48. Hu, FB. Resolved: there is sufficient scientificevidence that decreasing sugar-sweetenedbeverage consumption will reduce the pre-valence of obesity and obesity-related dis-eases. Obesity Rev. 2013;14(8):606–619

ARTICLE

PEDIATRICS Volume 134, Number 3, September 2014 e731

http://www.pediatrics.org/cgi/content/full/133/4/e896http://www.pediatrics.org/cgi/content/full/133/4/e896