Systematic Review of Community-Based ChildhoodObesity Prevention Studies
abstractOBJECTIVE: This study systematically reviewed community-basedchildhood obesity prevention programs in the United States andhigh-income countries.
METHODS:We searched Medline, Embase, PsychInfo, CINAHL, clinicaltrials.gov, and the Cochrane Library for relevant English-language studies.Studies were eligible if the intervention was primarily implemented inthe community setting; had at least 1 year of follow-up after baseline;and compared results from an intervention to a comparison group.Two independent reviewers conducted title scans and abstract reviewsand reviewed the full articles to assess eligibility. Each article receiveda double review for data abstraction. The second reviewer confirmedthe first reviewer’s data abstraction for completeness and accuracy.
RESULTS: Nine community-based studies were included; 5 randomizedcontrolled trials and 4 non–randomized controlled trials. One studywas conducted only in the community setting, 3 were conducted in thecommunity and school setting, and 5 were conducted in thecommunity setting in combination with at least 1 other setting suchas the home. Desirable changes in BMI or BMI z-score were found in 4of the 9 studies. Two studies reported significant improvements inbehavioral outcomes (1 in physical activity and 1 in vegetable intake).
CONCLUSIONS: The strength of evidence is moderate that a combineddiet and physical activity intervention conducted in the community witha school component is more effective at preventing obesity or over-weight. More research and consistent methods are needed to under-stand the comparative effectiveness of childhood obesity preventionprograms in the community setting. Pediatrics 2013;132:e201–e210
AUTHORS: Sara N. Bleich, PhD,a Jodi Segal, MD, MPH,a,b
Yang Wu,c Renee Wilson,a and Youfa Wang, MD, PhDc
aDepartment of Health Policy and Management, cJohns HopkinsGlobal Center on Childhood Obesity—Where Systems ScienceMeets Public Health, Department of International Health, andCenter for Human Nutrition, Johns Hopkins Bloomberg School ofPublic Health, Baltimore, Maryland; and bDepartment ofMedicine, Division of General Internal Medicine, Johns HopkinsUniversity School of Medicine, Baltimore, Maryland
Dr Bleich conceptualized and designed the study, drafted theinitial manuscript, and approved the final manuscript assubmitted; Dr Segal conceptualized and designed the study andcritically reviewed and revised the manuscript; Ms Wu reviewedand revised the manuscript; Ms Wilson coordinated andsupervised data the systematic review; she also reviewed andrevised the manuscript; and Dr Wang conceptualized anddesigned the study and critically reviewed and revised themanuscript.
www.pediatrics.org/cgi/doi/10.1542/peds.2013-0886
doi:10.1542/peds.2013-0886
Accepted for publication Apr 22, 2013
Address Correspondence to Sara N. Bleich, PhD, Department ofHealth Policy and Management, Bloomberg School of PublicHealth, Johns Hopkins University, 624 N. Broadway, Room 451,Baltimore, MD 21205. E-mail: [email protected]
FINANCIAL DISCLOSURE: The authors have indicated they haveno financial relationships relevant to this article to disclose.
FUNDING: This project was funded under contract 290-2007-10061-I from the Agency for Healthcare Research and Quality, USDepartment of Health and Human Services. The authors of thisreport are responsible for its content. Statements in the reportshould not be construed as endorsement by the Agency forHealthcare Research and Quality or the US Department of Healthand Human Services. This work was also supported by a grantfrom the National Heart, Lung, and Blood Institute at the NationalInstitutes of Health (1K01HL096409). Funded by the NationalInstitutes of Health.
COMPANION PAPER: A companion to this article can be found onpage e193, online at www.pediatrics.org/cgi/doi/10.1542/peds.2013-0786.
PEDIATRICS Volume 132, Number 1, July 2013 e201
REVIEW ARTICLE
by guest on October 30, 2020www.aappublications.org/newsDownloaded from
Childhood obesity represents one of themost important and challenging publichealth problems in developed coun-tries. Although there appears to be anoverall leveling off of the obesity epi-demic among the pediatric populationin theUnitedStatesandotherdevelopedcountries, low socioeconomic groupsremain disproportionately affected.1,2
Despite evidence of body weight stabi-lization in the general population, obe-sity prevention should remain a prioritybecause current prevalence is high,3
and previous stable phases have his-torically been followed by additionalincreases in obesity prevalence.4,5
The drivers of the positive energy bal-ance (driven by either too much energyintake or too little energy expenditure)are increasingly being identified in theenvironment.6–9 As a result, recenttrends in research, in the United Statesand other developed countries, em-phasize the use of community-basedsettings as an important strategy forpreventing obesity. This shift towarda stronger community focus is echoedby the recent Institute of Medicine re-port Accelerating Progress in ObesityPrevention, which recommends a com-prehensive approach to childhoodobesity prevention that includes thecommunity.10 This community-basedfocus is especially important for chil-dren who generally have little or nocontrol over the social and environ-mental factors where they live.
Given that most previous childhoodobesity preventionprogramshave beenconducted primarily in schools, themajority of systematic reviews havefocused on that setting11–13. An excep-tion is a recent Cochrane review,14 butlittle attentionwasdevoted to community-based studies. Rather, the majority ofincluded studies in that review wereinterventions conducted in school set-tings and interventions implementedfor ,12 months, an important limita-tion because of the weakness and bias
of data derived from short-term be-havior change studies.
In recent years, the body of literaturefocusing on childhood obesity pre-vention in the community setting hasincreased considerably, making thesynthesis of this evidence base impor-tant. This study contributes to the lit-erature by focusing specifically oncommunity-based childhood obesityprevention programs in high-incomecountries. To our knowledge, no studyhassystematicallyreviewedthisevidencebase. We defined community-based in-terventions as those interventions re-sulting from policy, legislative, builtenvironment, andeconomic/pricing/foodsubsidy changes that aimed to reducepopulation risk of obesity. We selectedhigh-income countries, rather than allcountries, given the paucity of inter-vention studies in other parts of theworld. We compared diet, physical ac-tivity, or combined diet and physicalactivity interventions. This study is partof a larger, 2-year project of a system-atic review funded by the Agency forHealthcare Research and Quality thatexamined the effectiveness of child-hood obesity prevention studies in de-veloped countries.15
METHODS
We used the methods recommendedby the Agency for Healthcare Researchand Quality Methods Guide for Effec-tiveness and Comparative EffectivenessReviews.16
Search Strategy
WesearchedMedline,Embase,PsychInfo,CINAHL, and the Cochrane Librarythrough August 11, 2012, for relevantstudies (the search was not limited to astart date, so all literature was cata-loged). We developed a search strategyfor Medline based on medical subjectheadings terms and text words of keyarticles that we identified a priori.We reviewed the reference lists of all
included articles and relevant reviewarticles to identify articles that thedatabase searches might have missed.We uploaded the articles into Distill-erSR, a Web-based software packagedeveloped for systematic review anddatamanagement. Studies published inother languages or unpublished stud-ies were also examined, but none metour inclusion criteria.
Selection of Studies
We identified studies conducted in high-income countries that described theeffectsof interventionstopreventobesity(or “excessive weight gain”) in childrenand adolescents aged 2 to 18 years old.We included only randomized controlledtrials (RCTs), quasi-experimental stud-ies, and natural experiments. The stud-ies needed to follow children for $1year after the intervention.
The interventions of interest involveda modification of diet, a modification ofphysicalactivityorsedentaryactivity, ora combination of these. We requiredthat the study reported on the attaineddifferences between the interventionand control groups in weight-relatedoutcomes (discussed subsequently).We excluded studies that targeted onlyoverweight or obese subjects or thosewith a medical condition such as di-abetes or heart disease. We includedonly articles published in English butreviewed the abstracts of non–Englishlanguage articles to assess agreementwith the results published in English
In particular, studies were eligible forinclusion if they (1) were primarily lo-cated in the community setting; (2)targeted at the pediatric population(ages 2–18); (3) had at $1 year offollow-up after baseline; (4) comparedresults from an intervention to a com-parison group (eg, usual care, anotherdifferent intervention, or no inter-vention); (5) reported differences inweight between the intervention andcontrol groups (eg, BMI, BMI z-score
e202 BLEICH et al by guest on October 30, 2020www.aappublications.org/newsDownloaded from
and percentile, waist circumference,percent body fat, skinfold thickness,prevalence of obesity and overweight);(6) described results from RCTs, quasi-experimental studies, and natural ex-periments, such as those that describedoutcomes from a community that hada food policy change; or (7) were pub-lished in English but reviewed the ab-stracts of non–English language articlesto assess agreement with the resultspublished in English. We focused onprevention but not treatment of over-weight and obese children because thosestudies have been reviewed elsewhere17
and because it is typically harder tolose weight than to prevent an initialweight gain.
Studies were excluded if they were (1)observational (eg, cross-sectional); (2)targeted only at overweight or obesechildren or adolescents; (3) targetedonly at children or adolescents withchronicmedical condition (eg, diabetesor heart disease); (4) expressly tar-geted at weight loss; (5) collected onlyqualitative results (eg, interviews orfocus groups); (6) or published only inabstract form.Wedifferentiatednaturalexperiments from other observationalstudy designs by specifying that a nat-ural experiment was the implementa-tion of a policy or similar interventionata population level.
We also reviewed the reference lists ofall included articles, relevant reviewarticles, andrelatedsystematicreviewsto identify additional articles.
Definitions of Outcomes
The primary outcomes were adiposityand obesity-related outcomes, which in-cluded BMI z-score, BMI, prevalence ofobesity and overweight, percent body fat,waist circumference, and skinfold thick-ness (listed in hierarchical order basedon their association with obesity).18–21
For studies that reported multiple bodyweight outcomes, we only reported theone closest to the top of the list.
The intermediate outcomes were di-etary intake (energy intake, fruit andvegetable intake, fatty food intake, andsugar-sweetened beverage intake), phy-sical activity, and sedentary behavior. Foreach of these outcomes, we also de-veloped a hierarchy of measures andreported only 1 for each category.
Data Extraction
Two independent reviewers conductedtitle scans and abstract reviews andreviewed the full articles to assess el-igibility for inclusion for each study. Wecreated standardized forms for dataextraction. Each article received a dou-ble review for data abstraction. Thesecond reviewer confirmed the firstreviewer’s data abstraction for com-pleteness and accuracy. Reviewersextracted information on study char-acteristics, study participants, eligibil-ity criteria, interventions, outcomemeasures, the method of ascertain-ment, and the outcomes.
Data Synthesis
Included studies were categorized asbeing implemented in the communityonly or in the community with othersettings (eg, home, school, primarycare,child care). For each setting, we orga-nized the data by intervention target(diet only, physical activity only, andcombined diet and physical activity) andoutcomes (primary weight outcomesand intermediate outcomes). Becauseofthe limited number of studies for eachintervention, we did not quantitativelypool the results. We extracted clinicaloutcomes that are reported in the text ofthe evidence report (available at www.ahrq.gov).
Risk of Bias Assessment
We used the Downs and Black in-strument toassess theriskofbias in theincluded studies.22 We categorized thestudies as having low, moderate, orhigh risk of bias. We rated a study as
having low risk of bias only when it haddone all of the following: stated theobjective clearly, described the mainoutcomes, described the character-istics of the enrolled subjects, describedthe intervention clearly, described themain findings, randomized the subjectsto the intervention group, and con-cealed the intervention assignmentuntil recruitment was complete. Addi-tionally, the study had to have at leastpartially described the distributions of(potential) principal confounders ineach treatment group. If a study did notcomplete 1 of the above items or if thiswas difficult to verify, it was classified ashaving a moderate risk of bias; fora study that definitively did not do$2 ofthe items, it was classified as havinga high risk of bias.
Strength of Evidence Assessment
After synthesizing the evidence, wegraded the quantity, quality, and con-sistencyof thebest availableevidencebyadapting an evidence grading schemerecommended in theMethods Guide forConducting Comparative EffectivenessReviews.16 We assigned grades for allweight-related outcomes by setting upa hierarchy of outcomes. Within this hi-erarchy, each study contributed only 1weight-related measure to the grade.The hierarchy was as follows: BMIz-score, BMI, prevalence of obesity andoverweight, percent body fat, waist cir-cumference, and skinfold thickness.For example, if a study measured BMIz-score and body fat, we only gradedBMIz-score. We chose 6 categories of inter-mediate outcomes: energy intake, fruitand vegetable intake, fatty food intake,sugar-sweetened beverage intake, phy-sical activity, and sedentary activity.
We considered the 4 recommendeddomains including risk of bias, di-rectness of the evidence, consistencyacross studies, and precision of thepooled estimate or the individual studyestimates.
REVIEW ARTICLE
PEDIATRICS Volume 132, Number 1, July 2013 e203 by guest on October 30, 2020www.aappublications.org/newsDownloaded from
We classified evidence into 4 catego-ries: (1) “high” grade (indicating highconfidence that the evidence reflectsthe true effect, and further research isunlikely to change our confidence inthe estimate of the effect); (2) “mod-erate” grade (indicating moderateconfidence that the evidence reflectsthe true effect, and further researchmay change our confidence in the es-timate of the effect and may changethe estimate); (3) “low” grade (indi-cating low confidence that the evi-dence reflects the true effect, andfurther research is likely to changeour confidence in the estimate of theeffect and is likely to change the esti-mate); and (4) “insufficient” grade(indicating evidence is unavailable,there was only 1 study, and it hadmoderate to high risk of bias, ora conclusion could not be drawn basedon the data).
We considered the body of evidenceconsistent in direction if$70% of thestudies had an effect in the same di-rection (ie, showed desirable effectverse not). We considered a studyprecise if the results for the givenoutcome were significant at a P, .05or had narrow confidence intervalsthat excluded the null. If$70% of thestudies reported statistically signifi-cant results, we considered the bodyof evidence precise.
RESULTS
Literature Search
The literature search outlined in themethods identified 40 356 potentialarticles, including 5791 duplicates(Fig 1). We identified 601 articles thatpotentially met all criteria as a resultof title and abstract screening. Fourhundred and seventy articles did notmeet the inclusion criteria, and anadditional 122 articles were notcommunity-based interventions. Thus,9 articles were finally included in thisreview.
Study Settings and InterventionCharacteristics
The study settings varied considerably;1 intervention was exclusively in acommunity setting23 (Table 1). Theother studies were implemented insettings that included the communityand additional venue(s) (eg, school,home, primary care, child care). Theprograms delivered a variety of inter-ventions that were administered overa range of 12 months for an exerciseresistance program23 to 48 months forcomplex interventions involving multi-ple community stakeholders.24 Onestudy reported on a physical activityintervention23; all others reported oncombination interventions including bothdiet and physical activity. Five of thestudies were RCTs,23,25–28 and 4 of thestudies were quasi-experimental.24,29–31
Six of the studies were conducted inthe United States,24,25,27–30 1 in Australia,31
1 in Switzerland,23 and 1 in theNetherlands.26
Common characteristics found acrossmost of the studies included the use ofmultiple intervention components (eg,health education and family outreach),the inclusion of settings other than justthe community (eg, school, home, pri-mary care, child care), and a focus onchildren at middle school age oryounger. Each of the studies is nowdescribed in detail.
Chang et al30 examined a statewidestrategy to reduce childhood obesity inthe state of Delaware. The study in-cluded 4595 children from birth to age17 years. This community-based pop-ulation health intervention includedpolicy and practice changes includingstatewide regulations to reduce sed-entary behavior and promote healthyeating/physical activity, implementa-tion of wellness policies in schools,student fitness assessment, training ofchild care providers about healthybehaviors, and provision of primary carephysicians with tools to meet expert
committee recommendations aboutobesity care.
Chomitz et al,24 in a study called HealthyLiving Cambridge Kids, assessed theimpact of a community-based healthyweight intervention in the United Stateson child weight and fitness targetingcommunity, school, and home (familyand individuals). It included 1858 chil-dren in elementary school (gradeskindergarten to 5) of whom 37% wereAfrican American, 14% were Hispanic,37% were white, and 10% were Asian;the follow-up period was 36 months.The intervention included communityawareness of healthy eating, improve-ments to food service in schoolsincluding new recipes and menu de-velopment and cafeteria taste tests,and improving access to appealingphysical education programs in theschools.
De Silva-Sanigorski et al,31 in the Rompand Chomp intervention in Australia,examined the effectiveness of an in-tervention that aimed to reduce obesityand promote healthy eating and activeplay in children aged 0 to 5 years old.The sample included 16 869 2-year-oldsand 12 354 3-year-olds with a follow-upperiod of 48 months. The interventionincluded community capacity buildingand environmental (political, sociocul-tural, and physical) changes to in-crease healthy eating and active play inearly-childhood care and educationalsettings. It had 8 specific objectivesfocused broadly on capacity building,increased awareness, improved diet,and increased physical activity.
Economosetal,29 inShapeUpSommerville,examined whether a community-basedenvironmental intervention could pre-vent weight gain in young children. Itincluded 1178 children in grades 1 to 3attending public elementary schoolswith a follow-up period of 12 months.The intervention focused on increasingphysical activity options and availabilityof healthful foods within the before-,
e204 BLEICH et al by guest on October 30, 2020www.aappublications.org/newsDownloaded from
during-, after-school setting and in thehome and community.
Eiholzer et al23 examined whether high-intensity training increases spontane-ous physical activity in children inSwitzerland. It included 2 junior icehockey teams consisting of 46 boys withamean age of 13 years and had a follow-up period of 12months. The interventionincluded high-intensity training exer-cises including supervised 1-hour exer-cise sessions twice weekly.
Klesges et al,28 in the the Memphis GirlsHealth Enrichment Multi-Site Study,examined the impact of a community-and family-based obesity preventionprogram for low-income AfricanAmerican girls. The study included 303African American girls aged 8 to 10years and their parents or guardianand a follow-up period of 24 months.The intervention randomized familiesto 1 of 2 interventions: (1) group be-havioral counseling to promote healthy
eating and increased physical activityor (2) self-esteem and social efficacyintervention.
Robinson et al’s27 Stanford Girls HealthEnrichment Multi-Site Study examinedthe impact of a community- and family-based obesity prevention program forlow-income African American girls. Thestudy included 261 African Americangirls aged 8 to 10 years and theirparents or guardian and a follow-upperiod of 24 months. The intervention
FIGURE 1Literature search for community-based childhood obesity prevention studies in high-income countries. aSum of excluded abstracts exceeds 5600 becausereviewers were not required to agree on reasons for exclusion. bSum of excluded articles exceeds 592 because reviewers were not required to agree onreasons for exclusion.
REVIEW ARTICLE
PEDIATRICS Volume 132, Number 1, July 2013 e205 by guest on October 30, 2020www.aappublications.org/newsDownloaded from
randomized families to 1 of 2 inter-ventions: (1) afterschool hip-hop, Afri-can, and step-dance classes and ahome/family-based intervention toreduce screen media use or (2)information-based health education.
Sallis et al25 evaluated the effects ofenvironmental, policy, and social mar-keting interventions on physical activ-ity and fat intake of middle schoolstudents in the United States. It in-cluded 24 middle schools with meanenrollments of 1109 children with 44%nonwhite students and a follow-up pe-riod of 24 months. The intervention in-cluded physical activity (increasedphysical activity in physical educationclasses and throughout the schoolday), nutrition (provided low-fat foodsat all school food sources), and en-gagement of school staff and studentsin policy change efforts.
Singh et al’s26 Dutch Obesity Inter-vention in Teenagers (DoiT) examinedthe impact of a multicomponent healthpromotion intervention for Dutch ado-lescents (aged 12–14) on body com-position and dietary and physicalactivity behavior. It included 1108 ado-lescents (mean age 12.7 years) witha follow-up period of 20 months. Theintervention consisted of an individualcomponent (ie, an educational pro-gram) and an environmental compo-nent (ie, encouraging that schools offeradditional physical education classesand changes in and around schoolcafeterias).
With the exception of 2 studies, whichreceived active interventions,27,28 thecontrol groups for these studies re-ceived no intervention.
Primary Adiposity- and Obesity-Related Outcomes
The primary weight outcomes are de-scribed in Table 2. Overall, desirableand significant changes in BMI or BMIz-score were found in 4 of 9 studies,indicated by a checkmark in the farTA
e206 BLEICH et al by guest on October 30, 2020www.aappublications.org/newsDownloaded from
right column of the table. In general,the studies that found significantchanges were characterized by a rela-tively longer follow-up time, a focus onyounger children (middle school oryounger), a quasi-experimental designthat allowed for the inclusion of mul-tiple intervention components, and theinclusion of settings other than just thecommunity. As indicated in the footnoteto Table 2, the comparison group wasan active intervention in 2 studies.27,28
Chang et al’s study,30 conducted in acommunity in Delaware with compo-nents in the school, primary care, andchild-care settings, showed no signifi-cant change in the prevalence of obe-sity. The Chomitz et al study (HealthyLiving Cambridge Kids),24 conducted inCambridge, Massachusetts, in thecommunity setting with school in-volvement, demonstrated significantreductions in BMI z-score from 0.67(sd 5 1.06) to 0.63 (sd 5 1.03), P ,.001). De Silva-Sanigorski et al’s study,31
conducted in the community setting
with participation from family, primarycare physician, and child care, ob-served a significant reduction in BMIfor children aged 3.5 years (b = –0.06kg/m2; 95% CI: –0.10 to –0.01 kg/m2).Economos et al (Shape Up Sommer-ville),29 conducted in the communitysetting with school and home compo-nents, observed a significant reductionin BMI z-score (b = –0.1005; 95% CI: –0.1151 to –0.0859). Eiholzer et al,23
conducted in the community among 2boys’ ice hockey teams, showed nosignificant change in body fat. Kresgeset al’s study,28 conducted in the com-munity with a home component, did notobserve significant differences in BMIbetween the 2 groups. Robinson et al,27
conducted in the community with ahome component, did not observe sig-nificant differences in BMI between the2 groups. The study was conductedamong a small sample of AfricanAmerican girls. The health educationand physical activity interventions wereconducted in different study groups;
therefore, the participants did not re-ceive both the diet and physical activityintervention. Sallis et al’s study,25 con-ducted in the community with a schoolcomponent, demonstrated a significantreduction in BMI for boys (from 20.1 to19.8 kg/m2, P = .044) but not girls. Singhet al’s study,26 conducted in the com-munity with a school component, didnot observe significant differences inBMI.
There is moderate strength of evidencethat community-based interventionsthat include a school component ef-fectively prevent obesity or overweightin children (Table 3). Two of the 3studies24,25 implemented in the com-munity with school involvement, whichtested combined diet and physical ac-tivity interventions, detected a statisti-cally significant beneficial effect of theintervention. The evidence is insufficientto support the benefit of other inter-ventions conducted in the communityalone or with involvement from othersettings (eg, home, primary care, child
Boys: 0.2; girls: 0.2 Boys: CI (–0.1 to 0.4); girls:CI (–0.1 to 0.5)
Klesges, 201028b BMI change (kg/m2) NR NR 20.06 CI (-0.87 to 0.75)Robinson, 201027b,c BMI change (kg/m2) NR NR 0.04 CI (-0.18 to 0.27)
CG, control group; NA, not applicable, NR, not reported.a Change from baseline.b The comparison group is an active intervention. For the Kelsges et al study, group behavioral counseling (obesity prevention program) is compared with self-esteem and social efficacy(alternative intervention). For the Robinson et al study, dance classes and reduced screen use are compared with health education.c This intervention had no control; the health education intervention is reported in the control column. and the dance and screen time reduction intervention is reported in the interventioncolumn.
REVIEW ARTICLE
PEDIATRICS Volume 132, Number 1, July 2013 e207 by guest on October 30, 2020www.aappublications.org/newsDownloaded from
care) that use physical activity orcombination approaches for obesityprevention in children.
Intermediate Outcomes—BehavioralOutcomes
The reported types of intermediateoutcomes varied considerably acrossstudies with only 1 study finding a sig-nificant effect (Table 4). As indicated inthe footnote to Table 4, for 2 studies, thecomparison group was an active in-tervention.27,28 The de Silva-Sanigorskiet al study31 observed a statisticallymarginally significant (P , .010) in-crease in servings of vegetables in theintervention group compared with thecontrol (by 0.10, P = .07). The Eiholzerstudy in Switzerland detected a statisti-cally significant benefit of high-intensitytraining on spontaneous physical activ-ity in the intervention group comparedwith the control (–0.01 kcal/min vs 0.23kcal/min, P = .02).23
DISCUSSION
This systematic review identified 9relevant studies. Four of the studies,which used combined diet and physicalactivity approaches, reported signifi-cant reduction in adiposity and weight-related outcomes as a result of theintervention.24,25,29,31 One of the studiesreported significant improvements inintermediate weight-related outcome(physical activity) as a result of the
intervention.23 There are currently notenough studies with consistent meth-ods and outcomes to determine theimpact of community-based childhoodobesity prevention programs on pri-mary or secondary weight outcomes.However, the evidence suggests thatcombination interventions implementedin multiple settings may be more ef-fective at preventing weight gain inchildren than single-component inter-ventions located in the community only.In particular, we found moderate evi-dence that community-based interven-tions that include a school componentand use interventions focused on bothdiet and physical activity effectivelyprevent obesity or overweight in chil-dren, regardless of the design of thestudy (ie, RCT or non-RCT). The com-pleteness with which interventions areimplemented has been associated withprogram impact.32 There is consider-able heterogeneity across the studydesigns, which may contribute to theinconsistent findings. The studies useddifferent units of analysis (eg, child,school), different methodologies (eg,RCTs vs quasi-experimental studies),were located in nonuniform settings,and used different interventions (eg,physical activity or combined diet andphysical activity).
The studies that observed a significantdecline in adiposity outcomes generallyenrolled more participants and had
longer follow-up periods, a focus onchildren middle school age or younger,the inclusion of settings other than justthe community, and less rigorous studydesigns; 3 of the 4 studies that dem-onstratedsignificant reductions ischildweight were quasi-experimental.
This review has some limitations. Manystudies published in this area havesuboptimal study designs, which maylead to biased results. Unsuccessfulprograms may not have been includedin the analysis because of a lack ofpublisheddata.Asaresult, theremaybesome publication bias, which may haveaffected the magnitude of the results.Our study was restricted to articlespublished in English but examinedabstracts of studies in other languagesaswell. Our studywas also restricted tointerventions located primarily in thecommunity setting, which excludeda number of studies that included thecommunity as a secondary component.
This study has a number of strengths. Arange of community-based childhoodobesity prevention interventions fromdifferent countries were included. Ro-bust review methods were used in-cluding the use ofmultiple databases toidentify articles and paired reviewers.The findings from this review can helpguide future researchstrategiesaswellas decision-making for researchers,clinicians, public health practitioners,and policy makers about the most
TABLE 3 Summary of the Strength of Evidence for Interventions in the Community: Primary Weight Outcomes
Setting, N Intervention Year(s) of Study Enrolled Participants Studies With Low/Moderate/High Risk
of Bias (n)
% With Favorablea
OutcomeEvidenceStatement
Community only, 1 PA 2010 46 0/1/0 W InsufficientCommunity, school, 3 D and PA 1997–2010 2966 and 24 schools (mean
enrollment of 11090/3/0 2 Moderate
Community, school, home, 2 D and PA 2007–2008 1326 0/1/1 1 InsufficientCommunity, home, 2 PA 2008–2010 564 0/0/2 0 InsufficientCommunity, home, primarycare and child care, 1
D and PA 2010 43 811 0/1/0 1 Insufficient
Community, school, primarycare, and child carecomponents, 1
e208 BLEICH et al by guest on October 30, 2020www.aappublications.org/newsDownloaded from
appropriate setting and types of inter-ventions to focus on tomaximize effortsto prevent childhood obesity in de-veloped countries.
More generally, understanding the ef-fectiveness of community-based child-hood obesity prevention programs isan important area of study. Even if
interventions have a modest effect onindividual body weight, the cumulativeimpact across the population has thepotential to yield significant publichealth benefits.
In conclusion, the strength of evi-dence is moderate that community-based childhood obesity prevention
programs with a school componentfocusing on both diet and physical ac-tivity is more effective at preventingobesity or overweight. More researchand more consistent methods areneeded to understand the comparativeeffectiveness of these interventionprograms.
REFERENCES
1. Olds TS, Tomkinson GR, Ferrar KE, Maher CA.Trends in the prevalence of childhood over-weight and obesity in Australia between 1985and 2008. Int J Obes (Lond). 2010;34(1):57–66
2. Rokholm B, Baker JL, Sørensen TIA. Thelevelling off of the obesity epidemic sincethe year 1999—a review of evidence andperspectives. Obes Rev. 2010;11(12):835–846
3. Ogden CL, Carroll MD, Curtin LR, LambMM, Flegal KM. Prevalence of high body
mass index in US children and adoles-cents, 2007–2008. JAMA. 2010;303(3):242–249
4. Olds TS, Harten NR. One hundred years ofgrowth: the evolution of height, mass,and body composition in Australian chil-dren, 1899–1999. Hum Biol. 2001;73(5):727–738
5. Keith SW, Redden DT, Katzmarzyk PT, et al.Putative contributors to the secular
increase in obesity: exploring the roadsless traveled. Int J Obes (Lond). 2006;30(11):1585–1594
6. Hill JO, Peters JC. Environmental con-tributions to the obesity epidemic. Science.1998;280(5368):1371–1374
7. Nestle M, Jacobson MF. Halting the obe-sity epidemic: a public health policy ap-proach. Public Health Rep. 2000;115(1):12–24
TABLE 4 Secondary Weight Outcomes for Community-Based Childhood Obesity Prevention Studies in High-Income Countries (n = 9)
Study, Year Secondary Outcome Measurement of SecondaryOutcome at Follow-up
Magnitude of ChangeFrom Baselinea
Significance ofChange From Baselinea
Control Intervention
Chang, 201030 NRChomitz, 201024 NRde Silva-Sanigorski, 201031 Servings of vegetables NR NR 0.10 P = .07
Servings of fruit NR NR 0.07 P = .14Economos, 200729 NREiholzer, 201023 Change in physical activity
Boys: 104 Boys: 115 Boys: 0.09 Boys: P = .84Girls: 91 Girls: 93 Girls: 0.25 Girls: P = .55
Sedentary h/d/student Boys:3.87 Boys: 4.42 Boys: 0.17 Boys: P = .69Girls:4.61 Girls: 4.64 Girls: 0.11 Girls: P = .71
Singh, 200926 Change in SSB consumption (mL/d) 714 689 288 CI (–203 to 28)Active commuting to school, min/d 42 46 22 CI (–10 to 5)Screen-viewing behavior
(television viewing andcomputer use), min/d
248 258 22 CI (–9 to 5)
Klesges, 201028b Mean total energy intake, kcal NR NR 278 CI (–186.3 to 31.1)Servings of vegetables NR NR 0.2 CI (–0.0 to 0.3)Servings of fruit NR NR 20.0 CI (–0.2 to 0.2)Moderate to vigorous activity, min NR NR 0.6 CI (–1.3 to 2.4)
Robinson, 201027b Mean total energy intake, kcal NR NR 227.3 CI (–69.9 to 15.0)Weekday accelerometer counts,
counts/minNR NR 3.18 CI (–11.6 to 17.9)
Weekly total screen time, h NR NR 22.65 CI (–5.42 to 0.13)
NA, not applicable; NR, not reported; SpAEE, spontaneous physical activity energy expenditure; SSB, sugar sweetened beverages; TrAEE, training activity energy expenditure.a Control vs intervention.b The comparison group is an active intervention. For the Kelsges et al study, group behavioral counseling (obesity prevention program) is compared with self-esteem and social efficacy(alternative intervention). For the Robinson et al study, dance classes and reduced screen use are compared with health education.
REVIEW ARTICLE
PEDIATRICS Volume 132, Number 1, July 2013 e209 by guest on October 30, 2020www.aappublications.org/newsDownloaded from
9. Bleich SN, Thorpe RJ Jr, Sharif-Harris H,Fesahazion R, Laveist TA. Social contextexplains race disparities in obesity amongwomen. J Epidemiol Community Health.2010;64(5):465–469
10. Institute of Medicine. Accelerating Prog-ress in Obesity Prevention: Solving theWeight of the Nation. Washington, DC: In-stitute of Medicine; 2012
11. Katz DL, O’Connell M, Njike VY, Yeh MC,Nawaz H. Strategies for the prevention andcontrol of obesity in the school setting:systematic review and meta-analysis. Int JObes (Lond). 2008;32(12):1780–1789
12. Khambalia AZ, Dickinson S, Hardy LL, Gill T,Baur LA. A synthesis of existing systematicreviews and meta-analyses of school-basedbehavioural interventions for controllingand preventing obesity. Obes Rev. 2012;13(3):214–233
13. Nixon CA, Moore HJ, Douthwaite W, et al;ToyBox-study group. Identifying effectivebehavioural models and behaviour changestrategies underpinning preschool- andschool-based obesity prevention inter-ventions aimed at 4–6-year-olds: a system-atic review. Obes Rev. 2012;13(suppl 1):106–117
14. Waters E, de Silva-Sanigorski A, Hall BJ,et al. Interventions for preventing obesityin children. Cochrane Database Syst Rev.2011; (12):CD001871
15. Wang Y, Wu Y, Wilson R, et al. ChildhoodObesity Prevention Programs: A Compara-tive Effectiveness Review and Meta-analy-sis. Rockville, MD: Agency for HealthcareResearch and Quality; 2012
16. Agency for Healthcare Quality and Re-search. Methods Guide for Effectivenessand Comparative Effectiveness Reviews.Rockville, MD: Agency for Healthcare Re-search and Quality; 2012
17. Whitlock EP, O’Connor EA, Williams SB, BeilTL, Lutz KW. Effectiveness of weight man-agement interventions in children: a targetedsystematic review for the USPSTF. Pediat-rics. 2010;125(2):e396–e418
18. Mei Z, Grummer-Strawn LM, Pietrobelli A,Goulding A, Goran MI, Dietz WH. Validity ofbody mass index compared with otherbody-composition screening indexes forthe assessment of body fatness in childrenand adolescents. Am J Clin Nutr. 2002;75(6):978–985
19. Noppa H, Andersson M, Bengtsson C, BruceA, Isaksson B. Body composition in middle-aged women with special reference to thecorrelation between body fat mass andanthropometric data. Am J Clin Nutr. 1979;32(7):1388–1395
20. Sarría A, García-Llop LA, Moreno LA, Fleta J,Morellón MP, Bueno M. Skinfold thicknessmeasurements are better predictors ofbody fat percentage than body mass indexin male Spanish children and adolescents.Eur J Clin Nutr. 1998;52(8):573–576
21. Garrido-Chamorro RP, Sirvent-Belando JE,Gonzalez-Lorenzo M, Martin-Carratala ML,Roche E. Correlation between body massindex and body composition in elite ath-letes. J Sports Med Phys Fitness. 200;49(3):278–284
22. Downs SH, Black N. The feasibility of cre-ating a checklist for the assessment of themethodological quality both of randomisedand non-randomised studies of health careinterventions. J Epidemiol CommunityHealth. 1998;52(6):377–384
23. Eiholzer U, Meinhardt U, Petrò R, WitassekF, Gutzwiller F, Gasser T. High-intensitytraining increases spontaneous physicalactivity in children: a randomized con-trolled study. J Pediatr. 2010;156(2):242–246
24. Chomitz VR, McGowan RJ, Wendel JM, et al.Healthy Living Cambridge Kids: a commu-nity-based participatory effort to promote
healthy weight and fitness. Obesity (SilverSpring). 2010;18(suppl 1):S45–S53
25. Sallis JF, McKenzie TL, Conway TL, et al.Environmental interventions for eating andphysical activity: a randomized controlledtrial in middle schools. Am J Prev Med.2003;24(3):209–217
26. Singh AS, Chin A Paw MJ, Brug J, vanMechelen W. Dutch obesity intervention inteenagers: effectiveness of a school-basedprogram on body composition and behav-ior. Arch Pediatr Adolesc Med. 2009;163(4):309–317
27. Robinson TN, Matheson DM, Kraemer HC,et al. A randomized controlled trial of cul-turally tailored dance and reducing screentime to prevent weight gain in low-incomeAfrican American girls: Stanford GEMS.Arch Pediatr Adolesc Med. 2010;164(11):995–1004
28. Klesges RC, Obarzanek E, Kumanyika S,et al. The Memphis Girls’ health EnrichmentMulti-site Studies (GEMS): an evaluation ofthe efficacy of a 2-year obesity preventionprogram in African American girls. ArchPediatr Adolesc Med. 2010;164(11):1007–1014
29. Economos CD, Hyatt RR, Goldberg JP, et al. Acommunity intervention reduces BMIz-score in children: Shape Up Somervillefirst year results. Obesity (Silver Spring).2007;15(5):1325–1336
30. Chang DI, Gertel-Rosenberg A, Drayton VL,Schmidt S, Angalet GB. A statewide strategyto battle child obesity in Delaware. HealthAff (Millwood). 2010;29(3):481–490
31. de Silva-Sanigorski AM, Bell AC, Kremer P,et al. Reducing obesity in early childhood:results from Romp & Chomp, an Australiancommunity-wide intervention program. AmJ Clin Nutr. 2010;91(4):831–840
32. Te Velde SJ, Brug J, Wind M, et al. Effects ofa comprehensive fruit- and vegetable-promoting school-based intervention inthree European countries: the Pro ChildrenStudy. Br J Nutr. 2008;99(4):893–903
e210 BLEICH et al by guest on October 30, 2020www.aappublications.org/newsDownloaded from
DOI: 10.1542/peds.2013-0886 originally published online June 10, 2013; 2013;132;e201Pediatrics
Sara N. Bleich, Jodi Segal, Yang Wu, Renee Wilson and Youfa WangSystematic Review of Community-Based Childhood Obesity Prevention Studies
ServicesUpdated Information &
http://pediatrics.aappublications.org/content/132/1/e201including high resolution figures, can be found at:
Referenceshttp://pediatrics.aappublications.org/content/132/1/e201#BIBLThis article cites 28 articles, 8 of which you can access for free at:
Subspecialty Collections
http://www.aappublications.org/cgi/collection/obesity_new_subObesityrvices_subhttp://www.aappublications.org/cgi/collection/community_health_seCommunity Health Services_subhttp://www.aappublications.org/cgi/collection/community_pediatricsCommunity Pediatricsfollowing collection(s): This article, along with others on similar topics, appears in the
Permissions & Licensing
http://www.aappublications.org/site/misc/Permissions.xhtmlin its entirety can be found online at: Information about reproducing this article in parts (figures, tables) or
Reprintshttp://www.aappublications.org/site/misc/reprints.xhtmlInformation about ordering reprints can be found online:
by guest on October 30, 2020www.aappublications.org/newsDownloaded from
