RE S E A R C H AR T I C L E

Individual, Family, and CommunityEnvironmental Correlates of Obesityin Latino Elementary School Children

JOHN P. ELDER, PhD, MPHa

ELVA M. ARREDONDO, PhDb

NADIA CAMPBELL, MPHc

BARBARA BAQUERO, MPHd

SUSAN DUERKSEN, MPHe

GUADALUPE AYALA, PhD, MPHf

NOC C. CRESPO, MPH, MSg

DONALD SLYMEN, PhDh

THOMAS MCKENZIE, PhDi

ABSTRACT

BACKGROUND: The prevalence of overweight children has reached epidemicproportions, and affects Latinos youth more than other subgroups in the United States.Given the prevalence of obesity and its economic consequences, community healthinitiatives have shifted toward primary prevention at younger ages.

METHODS: Data representing all levels of the ecological systems theory were collectedusing diverse methods. Participants were children enrolled in K-2nd grade and theirparents.

RESULTS: Overweight children were less active compared to normal weight children.The parents of overweight children provided less instrumental support to engage inactivity and set fewer limits on their childs activities. Similarly, parents of overweightchildren were less likely to control, but more likely to set limits on their childs dietcompared to parents of normal weight children. Parents who rated their health morepositively and were less acculturated were more likely to have children who wereoverweight. School and community level variables were not significantly correlated withchildrens weight. Adjusting for the aforementioned variables, parents weight status waspositively associated with childrens weight.

CONCLUSIONS: Social and structural environments in which Hispanic children arereared may play an important role in determining their risk for obesity and relatedbehaviors. Parents weight was among the strongest correlate of child weight; however,the extent to which this influence functions primarily through biological orsocial/structural influences is not entirely clear. The role of school and communityfactors on childs health practices and body mass index needs to be further examined.

Keywords: nutrition; diet; physical fitness; sport; chronic diseases; community health.

Citation: Elder J, Arredondo EM, Campbell N, Baquero B, Duerksen S, Ayala G, CrespoNC, Slymen D, McKenzie T. Individual, family, and community environmental correlatesof obesity in Latino elementary school children. J Sch Health. 2010; 80: 2030.

Received on November 1, 2009Accepted on August 24, 2009

aProfessor, (jelder@mail.sdsu.edu), Center for Behavioral and Community Health Studies, Graduate School of Public Health, San Diego State University, San Diego, CA 92123.bAssistant Professor, (earredondo@projects.sdsu.edu), Center for Behavioral and Community Health Studies, Graduate School of Public Health, San Diego State University, SanDiego, CA 92123.cProject Manager, (ncampbell@projects.sdsu.edu), Center for Behavioral and Community Health Studies, San Diego State University, San Diego, CA 92123.dEvaluation Coordinator, (bbaquero@projects.sdsu.edu), UCSD/SDSU Joint Doctoral Program, Center for Behavioral and Community Health Studies, San Diego State University, SanDiego, CA 92123.eIntervention Coordinator, (sduerksen@projects.sdsu.edu), Center for Behavioral and Community Health Studies, San Diego State University, San Diego, CA 92123.

20 Journal of School Health January 2010, Vol. 80, No. 1 2010, American School Health Association

Trends in US childrens overweight and obesityhave accelerated in recent years, including markedincreases in racial and ethnic disparities.1 Over the past3 decades, the childhood obesity rate has more thandoubled for children aged 25 years and adolescentsaged 1219 years, and it has more than tripledfor children aged 611 years. Recent national dataindicate that an estimated 14.5% of low incomepreschool children are overweight.2 Latino adults andchildren have been particularly affected by overweightand obesity.3 In 2008, 18.5% of Latino children wereoverweight.

According to the ecological systems theory (EST),childhood overweight is influenced by individual, fam-ily, and community factors.4,5 Children who are seden-tary and consume high amounts of sugar-sweetenedbeverages have a higher prevalence of overweight andobesity.68 Studies also show that family characteristicsand parenting styles shape childrens activity patterns,dietary intake, and obesity risk. Children of parentswho use controlling strategies (eg, demanding, rigid)are less likely to eat healthy.9 In contrast, monitoringof childrens eating and reinforcing healthy eatingencourage healthy eating.10,11 Although the relation-ship between parenting styles and childrens physicalactivity (PA) is not fully understood, research suggeststhat parents who encourage their children to engage inPA have more active children.12 Family characteristicssuch as parents income level are related to childrensweight status, healthy eating, and PA.13,14 AmongHispanics, studies have been inconsistent in show-ing an association between parental acculturation andchildrens body mass index (BMI).1517 Investigatorsfinding such a relationship note that the process ofparental acculturation likely influences the consump-tion of unhealthy foods, which places Hispanic familiesat risk for obesity.18 However, acculturation also hasbeen associated with the adoption of PA which com-plicates the relation between acculturation and weightstatus.19 Among the strongest predictors of childrensBMI is parents BMI.20

School-related factors also impact childrens risk forobesity. Studies have shown that verbal encourage-ment by school cafeteria staff is positively related tochildrens fruit and vegetable consumption.21 Students

fAssociate Professor, (ayala@mail.sdsu.edu), Center for Behavioral and Community Health Studies, Graduate School of Public Health, San Diego State University, San Diego, CA92123.gData Manager, (ncrespo@projects.sdsu.edu), Center for Behavioral and Community Health Studies, San Diego State University, San Diego, CA 92123.hProfessor, (dslymen@mail.sdsu.edu), Department of Biostatistics, San Diego State University, San Diego, CA 92123.iProfessor Emeritus, (tmckenzie@mail.sdsu.edu), Department of Exercise and Nutrition Sciences, San Diego State University, San Diego, CA 92123.

Address correspondence to: Elva M. Arredondo, Assistant Professor, (earredondo@projects.sdsu.edu), Center for Behavioral and Community Health Studies, Graduate School ofPublic Health, San Diego State University, 9245 Sky Park Court, Suite 221, San Diego, CA 92123.

This protocol was approved by the San Diego State University Institutional Review Board to ensure the protection of human participants.

*Indicates CHES and Nursing continuing education hours are available. Also available at: http://www.ashaweb.org/continuing education.html

in schools that facilitate PA by maintaining their activ-ity grounds and providing physical education classesare more active and less likely to be overweight.Veugelers and Fitzgerald22 found that schools offer-ing at least 2 physical education classes per weekhad fewer overweight children than schools offeringfewer classes. The neighborhood environment is alsohypothesized to influence childhood obesity risk, butthe evidence for this is mixed.2325 When consideringPA opportunities, Saelens and colleagues26 found thata walkable neighborhood environment (as assessed byresidential density, mixed land use, street connectivity,aesthetics, and safety) was associated with higher PAand lower overweight among adults.

Most studies examining correlates of childrensweight status have failed to simultaneously considerthe relative impact of multiple levels of EST on Latinochildrens weight status. As suggested by EST, theproposed study examines 4 levels of influence: childcharacteristics (eg, gender) and child risk factors (eg,dietary behavior); parenting and parents sociodemo-graphic characteristics; school and community relatedcharacteristics; and parents BMI. Findings from thisstudy can inform future interventions that target mod-ifiable factors with the greatest promise to change thecurrent upward trend in childhood obesity rates.

METHODS

Aventuras Para Ninos (Adventures for Kids) wasa randomized community trial designed to testan environmentally based intervention to preventchildhood obesity.27 Following baseline assessment,schools were randomly assigned to 1 of 3 interventionconditions or a control condition: micro (familyenvironment), macro (school and community), andmicro + macro (family environment, school andcommunity). Data for the present study werecollected at baseline using multiple methods: measureddata (anthropometric), self-reported, observed andgeographically coded data. These data were selectedto represent multiple levels of influence consistentwith EST,4 as presented in Figure 1.

SubjectsElementary schools within 3 school districts in

south San Diego County were invited to participate

Journal of School Health January 2010, Vol. 80, No. 1 2010, American School Health Association 21

Figure 1. Ecological Systems Theory as Applied to the Present Study

ECOLOGICAL SYSTEMS THEORY AS APPLIED TO THE PRESENT STUDY

COMMUNITY Density of restaurants to school, density of grocery stores to school, perceived neighborhood walkability

SCHOOL

SOPLAY, Cafeteria Nutrition Environment

FAMILY

Parenting styles, Parent health behaviors

INDIVIDUAL

Physical activity and eating habits

OUTCOME

Child BMI

in the study if they (1) had Hispanic enrollmentof at least 70%, (2) agreed to be randomized intointervention or control conditions, (3) did not haveany other obesity-related programs or special physicaleducation training28 within the past 4 years, and(4) had defined attendance boundaries (as opposedto charter or magnet schools that draw students froma broad region). A total of 13 public elementary schoolsfrom the 3 school districts that met these criteria weresuccessfully recruited.

Hispanic children entering kindergarten, first orsecond grade were eligible to participate in the study ifthey lived within the school attendance boundaries.Additional eligibility criteria included planning onliving in the target area for the next year, parentsability to read in English or Spanish, and willingnessto be randomized into one of the experimentalconditions. If the family had more than one eligiblechild, the youngest was selected. Children on amedically prescribed restricted diet or with a conditionthat limited their PA were excluded. Families wereineligible to participate if they anticipated movingout of the San Diego area during the study period.A total of 812 parent-child dyads were enrolledin the study. Following completion of the parentsurvey and anthropometric measurements, parent-child dyads were given a $20 incentive.

ProcedureData for the present study were collected at

baseline between August 2003 and January 2004and employed 4 modalities. Measured child andparent weight and height were obtained by trained

evaluation assistants (EAs) using portable scalesand stadiometers, respectively. Twenty percent ofthe anthropometric measurements were randomlyselected for reliability testing yielding inter-raterreliability estimates between 97% and 99%. Toobtain self-report data, a self-administered, 22-page questionnaire (available in English or Spanish)was given to the parents of each eligible child.Bilingual/bicultural EAs were available to assist withthe questionnaire. Observed data were collected bytrained research assistants. Geographically coded datawere collected by a trained GIS masters level student.Base GIS data was obtained from SanGIS (San DiegoGeographic Information Source, an agency of the cityand county of San Diego). Data licensed was theSouth County Landbase, updated quarterly during theperiod of July 2003July 2004. Layers included weremunicipal (city) boundaries, road centerlines (roads,major roads and minor highways, and freeways),parcels, and control monuments. The data dictionarywas uploaded to a GPS unit and field visits were madeto the areas previously classified as commercial toconfirm the accuracy of the list.

InstrumentsBody mass index (BMI) was the primary outcome

variable. Individual, family/home, school and commu-nity level influences on BMI were captured using thefollowing variables consistent with EST: (1) individual:childrens demographic characteristics, history of hav-ing been breastfed, and health behaviors (PA anddiet); (2) family: parents characteristics and parent-ing styles; (3) school: environmental influences that

22 Journal of School Health January 2010, Vol. 80, No. 1 2010, American School Health Association

prompt and support children to eat healthy and engagein PA; (4) community: environmental influences onPA and diet; and (5) parent BMI.

Child and Parent Body Mass IndexWeight was measured to the nearest pound using

a Health-o-Meter standard scale with the participantstanding without shoes and heavy objects removedfrom pockets. Height was measured to the nearest1/4 inch, using a standard portable stadiometer withshoes and head wear removed. Measures were alltaken in the clothing worn by the participant onthe day of measurement. Participants were askedto remove their jackets for weight measures. Thesemeasures were converted into metric equivalentsto calculate BMI. BMI was calculated using theQuetelet index (kg/m2). Child BMI percentiles andz-scores were calculated to standardize the datato national samples (adjusted for age and gender)using the Centers for Disease Control and Prevention(CDC) 2000 growth charts. Similar to the childanthropometric data collection procedures, adultweight (lbs) and height (in) were measured andcategorized using CDC guidelines.

Child Characteristics and Health BehaviorsChild Food Frequency Questionnaire (Parent

Report). Child diet was assessed using a food fre-quency questionnaire, with food items identified fromfocus groups involving the target population.29 Withinput from an anthropologist and our research team,items that were known to be obesogenic for Latinoswere added. Parents were asked to rate how oftentheir child consumed each food item using 4 responsecategories: (1) never, (2) at least once a month, (3)at least once a week, and (4) every day. Food itemswere categorized by 3 independent reviewers as totheir contribution to obesity risk and a summary scorewas computed to reflect daily consumption. The foodsincluded: (1) fruits and vegetables, (2) savory snacks,(3) sugar-sweetened beverages, and (4) plain water.

Child PA (Parent Report). Parents were asked, incomparison to other children, how active is yourchild? Response options ranged from 1 = muchless than others to 5 = much more than others.

Characteristics. Parents also reported on theirchilds gender, ethnicity, place of birth, familialrelationship to the child, and breastfeeding history.

Family Characteristics (Parents SociodemographicCharacteristics and Parenting Styles)

Encouragement of PA, Participation in PA, Instru-mental Support for PA (Parent Report). Parents wereasked how often during a typical week they encour-aged their child to participate in PA, participated in

PA with child, and provided transportation where thechild can be physically active. Response options rangedfrom 1 to 7 days a week.

Behavioral Strategies to Reduce Fat and IncreaseFiber (Parent Report). This 30-item scale wasdeveloped in a previous study30 and based, in part,on Kristals work31,32 in dietary behavioral strategiesto reduce dietary fat and increase fiber. This scalewas developed for Latinos and has demonstrated con-struct and predictive validity. The scale has been usedwith Latinos living in California and North Carolina.33

The Cronbach coefficient alpha for the fat subscale(19 items) was = .73 and for the fiber subscale(11 items) = .76.

Parenting Styles for Healthy Eating and PA (ParentReport). This scale assessed parenting styles associatedwith childrens dietary and activity behaviors. The26-item scale (14 for diet and 12 for activity)was developed for this project using qualitativeand quantitative methods.34 Concurrent validity wasestablished with the child feeding questionnaire35

and predictive validity was established with childBMI.36 Response options ranged from 1 = neverto 5 = always. A mean score for each subscalewas calculated. The alpha coefficient for each subscaleranged from .73 .87.

Frequency of Eating Away From Home (ParentReport). Parents noted how frequently their familiesate at fast food restaurants. Due to the distributionof the responses, responses specific to each settingwere dichotomized as less than versus once a weekor more.

Parent CharacteristicsSociodemographic Characteristics. Demographic

information such as country of formal education,household size, employment status, type of residentialdwelling, homeownership, age, history of breastfeed-ing, education, marital status, income, and dominantlanguage were collected from parents.

Self-Rated Health (Parent Report). Parents wereasked, compared to other people of your age, wouldyou say your health is. . .. Response options rangedfrom 5 = excellent to 1 = poor.

Parent Acculturation (Parent Report). Accultura-tion was measured using the acculturation ratingscale for Mexican Americans (ARSMA-II).37 Ele-ments included (1) language familiarity and usage,(2) ethnic pride and identity, (3) ethnic interaction,(4) cultural heritage, and (5) generational proxim-ity. Parents received an acculturation score reflectinggreater acculturation to the Anglo culture with ahigher/less negative score ( = .72).

Adult PA (Parent Report). The short version ofthe self-administered international PA questionnaire(IPAQ) assessed the frequency and duration of

Journal of School Health January 2010, Vol. 80, No. 1 2010, American School Health Association 23

walking, moderate intensity, and vigorous intensityPA for occupational purposes, house and yard work,transportation, and leisure during the last 7 days.38

The IPAQ has previously been used with Latinos.38

School CharacteristicsSchool Nutrition Environment (Observed). A scan

of the school environment was conducted on3 different days by 2 research staff to ascertain thenumber of visual and verbal prompts for healthy eat-ing. Any verbal prompt by the teacher, cafeteria aidworker, a volunteer or any adult in the lunch area toeat more fruit and/or vegetable was coded. Promptswere recorded during two 5-minute intervals. Theaverage number of verbal prompts from the 3 days ofobservation was used as an independent variable inthe model.

System for Observing Play and Leisure Activityin Youth (Observed). System for observing play andleisure activity in youth (SOPLAY) is an observationalmethod to determine level of PA among individualsobserved within a given target area. PA target areasat the schools were identified and assessed (N = 137;mean per school = 10.5; SD = 2.8). Trained assessorsconducted SOPLAY observations before school, duringrecess, and at lunch time in the 13 elementary schoolsover an 18-month period and on at least 5 daysat each school. During an observation in a targetarea, each child was coded as sedentary (ie, lyingdown, sitting, or standing), walking, or engaged invigorous activity using specialized counters. Activitycounts were summarized as the percentage of childrenengaging in different levels of activity. Assessors alsocoded area characteristics (eg, accessibility, usability;and presence of supervision, loose equipment, andorganized activities). Presence of PA equipment andpercent of children involved in sedentary activity wereincluded in the analyses.

Community CharacteristicsPerceived Neighborhood Safety/Aesthetics (Parent

Report). A modified version of the US Department ofTransportation, Partnership for a Walkable America,Pedestrian and Bicycle Information Center, and theUnited States Environmental Protection AgencysWalkability Checklist was used. Six questions assessedperceived neighborhood safety/aesthetics, includingthe presence of crime, lights, and vehicle exhaustin the neighborhood. The final model included asummarized score assessing perceived neighborhoodsafety/aesthetics (6 items; = .60).

Restaurant and Grocery Store Density Within aMile of Participants School (Geo-Coded). The numberof restaurants and grocery stores within a 1-mile radiusof the school was included in the analyses.

DATA ANALYSIS

Data were examined for normality and intraclasscorrelation (ICC) assessed for independence. The ICCfor BMI was zero and preliminary models adjustingfor school as a random effect showed no influenceon the regression coefficients or standard errors of thepredictor variables. This suggested no school clusteringand appropriate use of multiple linear regression asthe analytic approach. Percentages were calculatedfor categorical data and measures of central tendencyfor continuous data. Child BMI z scores were usedas the outcome variable for the final regressionmodels to adjust for differences by age and gender.Eight hundred and twelve families were recruitedand completed baseline assessment; however, only745 biological mothers were included in the currentstudy, given interest in examining the relationshipbetween biological parent weight and child weight.Because there were few fathers (N = 23), these caseswere also deleted. In addition, due to missing dataon acculturation (N = 91) and income (N = 48), 44additional cases were omitted from the analysis.Grocery store and restaurant density was calculatedusing GIS software (ESRI ArcMap 9.2). Densitywas defined as the total number of stores or restaurantswithin a 1-mile radius from each participants schooladdress.

A series of hierarchical regression analyses wereperformed to examine the effects of each theoreticallyderived correlate on child BMI. To examine the impactthat each EST level had on the amount of variance thatexplained child BMI, the R2 change was evaluated. Inthe first block, child characteristics were regressed onchild BMI. In the second block, child and parentingcharacteristics were regressed on child BMI. In thethird block, child, parenting, and family characteristicswere regressed on child BMI. In the fourth block, theaforementioned variables were entered in additionto school characteristics. The fifth block includedcommunity characteristics and the sixth and final blockincluded parent BMI.

RESULTS

Child and Parent Sociodemographic CharacteristicsTable 1 presents the general demographic, anthro-

pometric and other characteristics of the children andtheir parents. The mean age of the children was nearly6 years old and 86% of the children were born in theUnited States, contrasted to 28% of their parents. Par-ticipating families were generally very poor, with fully35.2% being at or below the 100% federal povertylevel of $1,720/month for a family of 4. Three-fourthsof the families rented their homes.

Nearly one half of the children were at risk forbeing overweight (ie, >85th percentile for their age

24 Journal of School Health January 2010, Vol. 80, No. 1 2010, American School Health Association

Table 1. Child and Parent Demographics andAnthropometrics (N = 745)Child

Mean age (SD) 5.97 (.943)% Girls 50.0% Children born in United States 86.2% Body mass index 85th percentile [N]

Girls 53.9 [152]Boys 53.1 [154]

% Body mass index 85th &

Tab

le3.

Cor

rela

tes

ofz

Scor

eC

hild

BM

I(N

=60

6)

Blo

ck1

Blo

ck2

Blo

ck3

Blo

ck4

Blo

ck5

Blo

ck6

(Std

bet

a)(S

tdb

eta)

(Std

bet

a)(S

tdb

eta)

(Std

bet

a)(S

tdb

eta)

Child

char

acte

ristic

sFr

uits

and

vege

tabl

espe

rday

.00

9.

006

.004

.013

.011

.006

Swee

tand

savo

rysn

acks

perd

ay.

007

.00

5.

016

.02

0.

022

.01

1Su

gar-s

wee

tene

dbe

vera

gesp

erda

y.0

27.0

50.0

43.0

41.0

41.0

29W

ater

drin

king

perd

ay.0

24.0

14.0

15.0

14.0

10.

006

Child

PAco

mpa

red

toot

herc

hild

ren

.18

7

.16

3

.14

2

.14

2

.13

9

.11

8

Age

ofch

ild.

017

.01

2.

020

.02

7.

024

.04

7Ch

ildge

nder

.

084

.07

6.

075

.07

9.

083

.07

3Ch

ildbr

east

fed

.01

4.

014

.02

6.

025

.02

8.

020

Hom

ech

arac

teris

tics

Pare

ntin

gEn

cour

agem

entf

orPA

.03

8.

036

.04

3.

040

.03

3Pa

rtici

pate

inPA

.031

.028

.039

.036

.039

Inst

rum

enta

lsup

port

forP

A.

093

.08

7.

082

.08

5.

091

Beha

vior

alst

rate

gies

forf

at.0

35.0

66.0

64.0

64.0

71Be

havi

oral

stra

tegi

esfo

rfibe

r.

041

.06

2.

063

.06

2.

056

PSm

onito

ring

ford

iet

.022

.012

.014

.012

.002

PSdi

scip

line

ford

iet

.046

.049

.044

.049

.056

PSco

ntro

lfor

diet

.25

9

.26

7

.26

2

.26

4

.23

5

PSlim

itfo

rdie

t.1

54

.158

.150

.1

47

.135

PS

rein

forc

emen

tfor

diet

.02

4.

022

.01

9.

016

.01

8PS

mon

itorin

gfo

ract

ivity

.055

.066

.072

.074

.065

PSdi

scip

line

fora

ctiv

ity.

044

.04

0.

035

.03

9.

041

PSco

ntro

lfor

activ

ity.0

20.0

11.0

17.0

18.0

16PS

limit

fora

ctiv

ity.

100

.09

5.

098

.10

1.

083

PSre

info

rcem

entf

orac

tivity

.061

.060

.059

.062

.063

Eatin

gou

tfas

tfoo

d.0

49.0

60.0

66.0

68.0

41Pa

rent

char

acte

ristic

sEm

ploy

men

t.

032

.04

2.

045

.04

6Ye

arso

fedu

catio

n.0

11.0

16.0

17.0

17M

arita

lsta

tus#

.01

5.

009

.01

3.

007

Mon

thly

inco

me

.

068

.07

6.

068

.06

3Ho

useh

old

size

.043

.050

.054

.026

Self-

rate

dhe

alth

.124

.1

29

.1

25

.086

Ad

ulta

ccul

tura

tion

.12

1.

077

.08

4.

097

Adul

ttot

alpe

rwee

kM

VPA

.056

.053

.052

.042

26 Journal of School Health January 2010, Vol. 80, No. 1 2010, American School Health Association

Tab

le3.

Con

tin

ued

Blo

ck1

Blo

ck2

Blo

ck3

Blo

ck4

Blo

ck5

Blo

ck6

(Std

bet

a)(S

tdb

eta)

(Std

bet

a)(S

tdb

eta)

(Std

bet

a)(S

tdb

eta)

Scho

olch

arac

teris

tics

Med

ian

num

bero

fvisu

alpr

ompt

spro

mot

ing

F&V

.05

6.

036

.07

5M

edia

nnu

mbe

rofv

erba

lpro

mpt

spro

mot

ing

F&V

.03

5.

037

.02

2PA

equi

pmen

t.

048

.000

.01

7Pe

rcen

tofs

eden

tary

child

ren

.00

5.0

15.

017

Com

mun

itych

arac

teris

tics

Wal

kabi

lity

.046

.026

Num

bero

fres

taur

ants

w/in

1m

ilefro

msc

hool

.099

.048

Num

bero

fgro

cery

stor

esw

/in1

mile

from

scho

ol.

067

.05

1Pa

rent

BMI

.227

p