psychosocial correlates of fruit, vegetable, and dietary fat intake among adolescent boys and girls
TRANSCRIPT
C
PDGMK
AOcegcaDaifdS5ySssRelch
MpaJle
MUDp
A
8
RESEARCH
urrent Research
Continuing Education Questionnaire, page 829Meets Learning Need Codes 4000, 4160, 6010, and 6040
sychosocial Correlates of Fruit, Vegetable, andietary Fat Intake among Adolescent Boys andirls
ARION F. ZABINSKI, PhD, MPH; TRACY DALY, MS, RD; GREGORY J. NORMAN, PhD; JOAN W. RUPP, MS, RD;
AREN J. CALFAS, PhD; JAMES F. SALLIS, PhD; KEVIN PATRICK, MD, MSvoCodtccscJ
AirftYgfitcfmaA3lgf
qatmufi
BSTRACTbjective This study examined whether hypothesized psy-hosocial correlates of behavior change (family/peer influ-nce, pros, cons, self-efficacy, parent/child change strate-ies, and household eating rules) are associated withonsumption of fruits, vegetables, and dietary fat amongdolescent boys and girls.esign This cross-sectional study used questionnaires tossess psychosocial variables and multiple 24-hour recallnterviews to assess dietary intake (daily servings ofruits and vegetables and percentage energy intake fromietary fat).ubjects In this study, 878 adolescents (53.6% female,7.9% white, mean age 12.8 years, age range 11 to 15ears) completed questionnaires.tatistical analyses performed Hierarchical linear regres-ions were conducted on the entire sample as well as onubgroups based on sex and age (young/old).esults Results indicated that child behavior change strat-gies, decisional balance, and household rules were re-ated to percentage energy intake from total fat, whereashild behavior change strategies, family influence, andousehold rules were related to daily servings of fruit and
. F. Zabinski and G. J. Norman are assistant adjunctrofessors, T. Daly is a dietitian, and K. Patrick is andjunct professor, University of California, San Diego.. W. Rupp is a lecturer and K. J. Calfas and J. F. Sal-is are professors, San Diego State University, San Di-go, CA.
Address correspondence to: Marion F. Zabinski, PhD,PH, Department of Family & Preventive Medicine,niversity of California, San Diego, 9500 Gilman Drive,ept 0811, La Jolla, CA 92093-0811. E-mail: [email protected] © 2006 by the American Dietetic
ssociation.0002-8223/06/10606-0006$32.00/0
odoi: 10.1016/j.jada.2006.03.014
14 Journal of the AMERICAN DIETETIC ASSOCIATION
egetables. More psychosocial correlates were found forlder than for younger adolescents.onclusions Both psychological and social correlates of ad-lescent eating behaviors were identified, and correlatesiffered somewhat by adolescent subgroup. Based onhese findings, promising intervention strategies that in-lude the following should be evaluated: helping adoles-ents alter decisional balance, teaching behavior-changetrategies, and helping parents support children’s dietaryhanges and institute supportive household rules.Am Diet Assoc. 2006;106:814-821.
diet rich in fruits and vegetables and low in fat isrecommended and has been shown to be related toimproved health (1,2). Fruits and vegetables are
mportant sources of fiber and are low in total fat, satu-ated fat, and sodium, but adolescent intake of theseoods do not meet recommended guidelines. Data fromhe Centers for Disease Control and Prevention’s 2003outh Risk Behavior Survey indicated that only 20.3% ofirls and 23.6% of boys in grades 9 to 12 reported eatingve or more servings of fruits/vegetables per day duringhe week before the survey (3). Not only do adolescentsonsume fewer than the recommended servings of ofruits and vegetables, but the proportion of adolescentseeting guidelines for the consumption of dietary fat is
lso less than the objectives for Healthy People 2010 (4).mong adolescents between the ages of 12 and 19, only6% of girls and 30% of boys met the objective of eatingess than 30% of energy from total fat, and only 34% ofirls and 27% of boys ate less than 10% of total energyrom saturated fat (4).
A major goal of health interventions is to improve dietuality; however, evidence about how to do this withdolescents is sparse. To create better health interven-ion programs, researchers need to examine mediator andoderator variables in relation to dietary outcomes to
nderstand what factors lead to behavior change (5). Therst step is to systematically identify potential correlates
f outcome behaviors that may serve as mediators, basing© 2006 by the American Dietetic Association
ttbrmie
iPtsaceFifawcptWsws
asChcasr
MPAtppptanw[e
arEco3aeptt1AA2Up
PP
he selection of mediators on validated behavior-changeheories. The current study examines potential correlatesased on Social Cognitive Theory (6) and the Transtheo-etical Model of Behavior Change (7). Together, theseodels account for personal, social, and environmental
nfluences as well as behavior change processes and strat-gies.Because dietary choice is an extremely complex behav-
or, multiple factors likely influence behavior change.revious research has been limited, however, to studieshat focused on a few correlates (8,9) and have demon-trated that psychosocial influences, such as self-efficacynd family/peer influences, can have an impact on in-reasing fruit and vegetable intake and decreasing di-tary fat intake among children and adolescents (10-14).or example, Cullen and colleagues (10) found positive
nfluences from family and friends increased self-efficacyor choosing and eating vegetables among 259 Girl Scoutsnd Granner and colleagues (13) found that self-efficacyas the strongest correlate of fruit and vegetable snack
hoice. Another study found that parental modeling wasositively correlated with consumption of fruit as well asotal fruit, 100% juice, and vegetables combined (11) and
oodward and colleagues (12) demonstrated that con-umption of a particular food among 12- to 15-year-oldsas significantly related to parents’ and friends’ con-
umption of that food.The present study aims to determine the correlates of
dolescent dietary behavior by examining a comprehen-ive set of behavior change constructs based on Socialognitive Theory and the Transtheoretical Model of Be-avior Change. Hypothesized psychosocial influences in-lude family/peer influence, decisional balance (eg, prosnd cons), self-efficacy, child strategies for change, parenttrategies for a healthful lifestyle, and household eating
Table 1. Demographic characteristics of adolescent boys and girls wcorrelates of behavior change are associated with consumption of fr
Demographic Total (n�839) Girls (n�
4™™™™™™™™™™™™™™™™™™™™™Age 12.8�1.3 12.8�1Height (in) 62.9�4.0 62.5�3Weight (lb) 129.6�40.3 130.0�3Body fat (%) 21.8�11.4 23.7�1BMIb 23.6�6.3 24.1�6Servings of fruits and vegetables 3.1�1.9 3.0�1% kcal from dietary fat 32.7�6.0 32.8�6Ethnicity
4™™™™™™™™™™™™™™™™™™™™™Asian/Pacific Islander 3.5 2.4African American 6.7 5.7Native American 0.7 0.9Hispanic 12.6 13.6White 58.0 58.3Multiethnic/Other 18.5 19.1
aSD�standard deviation.bBMI�body mass index; calculated as kg/m2.
ules. c
ETHODSarticipantsdolescents between the ages of 11 and 15 were recruited
hrough their primary care providers as part of a healthromotion intervention trial. A total of 45 primary careroviders from six clinic sites in San Diego County agreed toarticipate in the study. Study recruitment occurred be-ween May 2001 and June 2002. Trained study recruitersttempted to contact 3,366 households (including wrongumbers [13%], those not eligible [9%], individuals forhom recruitment was not completed [19%], and refusals
33%]), of which 878 adolescents and their parents (64% ofligible contacts) signed assent and consent forms.Reasons for ineligibility included: having a physical dis-
bility that precluded engaging in regular physical activity,elocating out of the region, inability to speak and readnglish, and no longer a patient of a participating primaryare provider. Reasons for refusal included: lack of interestr lack of time to participate in the study. Of participants,9 (4%) did not have complete data and were excluded fromnalyses. There were no significant differences between thexcluded group and those with complete data. Table 1 dis-lays demographic and anthropometric characteristics ofhe study sample. This sample was closely representative ofhe overall San Diego community, in which those ages 10 to7 are 45% non-Hispanic white, 35% Hispanic, 7% Africanmerican, and 12% Asian or other ethnicity (San Diegossociation of Governments, www.sandag.org, March 7,003). The Institutional Review Boards of San Diego Stateniversity and the University of California, San Diego ap-roved the study.
roceduresarticipants answered study questionnaires on a desktop
rticipated in a study to examine whether hypothesized psychosocialvegetables, and dietary fat
Boys (n�383)Younger (11-12 y)(n�385)
Older (13-15 y)(n�454)
™™™™™™™™ mean�SD a ™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™312.7�1.4 11.5�0.6 13.8�0.863.4�4.6 60.7�3.0 64.8�3.7
129.1�41.6 115.3�34.1 141.7�41.219.4�10.9 21.8�11.0 21.6�11.823.0�6.0 22.6�5.7 24.4�6.6
3.2�2.1 2.9�1.5 3.3�2.132.6�6.0 33.2�5.7 32.2�6.3
™™™™™™™™™™™™™ % ™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™34.7 4.9 2.27.8 6.5 6.80.5 0.5 0.9
11.5 11.9 13.257.7 56.6 59.317.7 19.5 17.7
ho pauits,
456)
™™™™.3.29.21.4.5.6.0
™™™™
omputer using a secure Web-based interface. The parent
June 2006 ● Journal of the AMERICAN DIETETIC ASSOCIATION 815
8
Table 2. Description of theoretically derived psychosocial variables for increasing servings per day of fruits and vegetables and reducing percentenergy from total fat
CorrelateNo. items(range) Question stem/sample items (response anchors) Cronbach �
Test-retest(ICCa)
Parent HealthfulLifestyleStrategies
13 (1-5) In the past month, how often have you:Put reminders around the home to help make and stick to healthy lifestyle
changes? Helped your child set short- and long-term goals to makehealthy lifestyle changes?
1�Never; 5�Very often
.94 —
Household eating rulesHealthful foods 3 (1-4) Do you have healthy snacks (eg, fruit or pretzels) around the house?
How often does dinner include vegetables?How often does breakfast include fruit and/or 100% fruit juice?1�Never; 4�Always
.50 —
Foods to limit 3 (1-4) Do you limit the amount of sweet snacks (eg, cookies and candy)?Do you limit the amount of dessert your child is allowed to eat?Do you limit the amount of soda your child is allowed to drink?1�Never; 4�Always
.82 —
Fruit/vegetableDecisional Balance
Pros 5 (1-5) How important is each statement to you:I would have more energy if I ate fruits and vegetables? I would feel
healthier if I ate fruits and vegetables? I would be doing somethinggood for my body if I ate fruits and vegetables?
1�Not important; 5�Extremely important
.75 .85
Cons 5 (1-5) How important is each statement to you:Fruits and vegetables are too difficult to prepare? I would feel
embarrassed if other kids saw me eating fruits and vegetables? I wouldrather eat sweets or high-fat snacks than fruits and vegetables?
1�Not important; 5�Extremely important
.85 .84
Self-efficacy 7 (1-5) How sure are you that you can do the following:Eat 5 servings of fruits and vegetables everyday? Eat fruits and vegetables
when eating out at a restaurant?1�I’m sure I can’t; 5�I’m sure I can
.81 .90
Family Influence 4 (1-5) During a typical week, how many days has a family member:Encouraged you to eat fruits and vegetables? Provided fruits or vegetables
as a snack or part of a meal?1�Never; 5�Every day
.75 .81
Peer Influence 3 (1-5) During a typical week, how often:Do your friends eat fruits or vegetables with you? Do other kids tease you
for eating fruits or vegetables?1�Never; 5�Every day
.43 .87
Child ChangeStrategies
15 (1-5) How often do you:Set goals to eat at least five servings of fruits and vegetables a day? Have
a friend or family member who encourages me to eat more fruits andvegetables? Reward myself for eating at least five servings of fruits andvegetables a day?
1�Never; 5�Many times
.91 .76
Dietary fatDecisional Balance
Pros 5 (1-5) How important is each statement to you:I feel good when I’m eating the high fat foods I enjoy? Foods high in fat
taste better than low-fat foods?1�Not important; 5�Extremely important
.75 .48
(continued)
16 June 2006 Volume 106 Number 6
apaiTmv
MDDr
DTTwMstspwavb
THfpspites(ic
wpieaspogco0r
ccompanying the adolescent to the assessment com-leted survey instruments about the adolescent’s homend neighborhood and their support for their child’s eat-ng a healthful diet and engaging in physical activity.hese surveys were completed with paper-and-pencil for-at, and trained research assistants entered response
alues into a database.
EASURESemographicsemographic variables included sex, age, and parental
eport of child’s ethnicity.
ietary Intakehree 24-hour food recalls assessed dietary intake.rained data collectors conducted dietary recalls for 2eekdays and 1 weekend day using the University ofinnesota Nutrition Data System for Research (NDS-R)
oftware version 4.04, 1998-2001 (15). Participants wereaught how to measure food portions with three-dimen-ional food models. The first interview was conducted inerson and the second and third by phone (participantsere given two-dimensional food models to use for phonessessments). Nutrient variables (servings of fruit andegetables, percent energy from total fat) were calculated
Table 2. Description of theoretically derived psychosocial variables foenergy from total fat (continued)
CorrelateNo. items(range) Question stem/sample items
Cons 5 (1-5) How important is each statemeIt bothers other people when I
foods now can mean health1�Not important; 5�Extremely
Self-efficacy 7 (1-5) How sure are you that you canChoose low-fat foods when oth
Ask someone in your family1�I’m sure I can’t; 5�I’m sur
Family Influence 4 (1-5) During a typical week, how maEaten low-fat foods with you?
eating low-fat foods?1�Never; 5�Every day
Peer Influence 3 (1-5) During a typical week, how oftDo your friends encourage you
low-fat foods with you?1�Never; 5�Every day
Child ChangeStrategies
15 (1-5) How often do you:Say positive things to myself a
make low-fat foods more enfood I eat?
1�Never; 5�Many times
aICC�intraclass correlation coefficient.
y averaging values from the three intake records. f
heoretically Derived Psychosocial Variablesypothesized correlates of fruit, vegetable, and dietary
at intake were developed for this study and based onrevious measures (16). A brief description of the con-tructs are provided in the following paragraphs andresented in more detail in Table 2 (eg, representativetems, response scale, and psychometric properties). Theable presents two reliability coefficients, Cronbach � co-fficients of internal consistency calculated for the overallample, and test-retest intraclass correlation coefficientsICCs) reported by Hagler and colleagues (16) from anndependent sample. ICCs were not available for scalesompleted by parents.ICCs were generally high, ranging from 0.75 to 0.93,ith the exception of household eating rules (��.50) andeer influence (��.43). These scales were comprised of annventory of items for which it was not expected thatndorsing one item in the scale would be associated withhigher likelihood of endorsing any other item in the
cale. However, test-retest reliability was high for theeer influence scales, indicating consistent responsesver time by adolescents. Likewise, test-retest ICCs wereenerally good for scales completed by adolescents, withoefficients ranging from 0.48 to 0.90. The pros and consf dietary fat had the lowest test-retest ICCs, 0.48 and.54, respectively. Although the scales exhibiting lowereliability coefficients can likely be improved, reliability
easing servings per day of fruits and vegetables and reducing percent
onse anchors) Cronbach �Test-retest(ICCa)
you:lot of high-fat foods? Eating high-fat
ems for me in the future?ortant
.85 .54
e following:round me are eating high-fat foods?y low-fat foods at the grocery store?
.81 .72
n
ys has a family member:ou that you are doing a good job
.75 .85
t low-fat foods? Do your friends eat.43 .73
eating low-fat foods? Do things tole? Keep track of how much high-fat
.93 .74
r incr
(resp
nt toeat aproblimp
do thers ato bue I ca
ny daTold y
en:to ea
boutjoyab
or all of the scales was considered acceptable.
June 2006 ● Journal of the AMERICAN DIETETIC ASSOCIATION 817
PDpvtfc
SStts(a
FTAfl
PIpt
CFtbctc
PApl
HPrsytcaco(
HAbmid
Mptl
DBgaatttcc�advms1yspf
RMdatudms
tshT
TF6c(hm(e(
SAvwhs
8
ros and Cons of Changeecisional balance consists of two constructs, labeled theros and cons of change that address cognitive and moti-ational aspects of human decision-making (17,18). Forhe current study, five pros and five cons for increasingruit and vegetable consumption and three pros and fiveons for reducing dietary fat were included.
elf-efficacyituational self-efficacy represents a person’s confidencehat he or she can meet a behavioral criterion in situa-ions that may present barriers to the behavior. Theelf-efficacy scales were adapted from previous measures19) and consisted of seven items for fruit and vegetablesnd eight items for dietary fat.
amily Influencewo scales with four items each were adapted from themherst Health and Activity Study to assess family in-uence on dietary fat and fruit and vegetable intake (20).
eer Influencetems similar to the family influence measure assessedeer influence related to dietary fat and fruit and vege-able intake.
hild Change Strategiesifteen items were created that reflect thoughts, activi-ies, and feelings people may experience when making aehavior change. These change strategies were similar toonstructs described as processes of change in the Trans-heoretical Model (7) and intervention constructs in So-ial Cognitive Theory (6).
arent Healthy Lifestyle Strategies13-item scale was created for this study to assess what
arents have done to help their children lead a healthfulifestyle.
ousehold Eating Rulesarents responded to six items that related to householdules related to eating behaviors using a 4-point Likertcale ranging from 1 (never) to 4 (always). A factor anal-sis with varimax rotation indicated a two-factor solu-ion. The first factor related to “healthful foods” and wasomprised of three items (having healthful snacks avail-ble at home, including vegetables with dinner, and in-luding fruit with breakfast). The second factor focusedn “foods to limit” and was comprised of three itemslimiting sweet snacks, desserts, and soda).
eight and Weightwall stadiometer (Seca Accu-Hite Wall mount with
uilt-in level bubble, Creative Health Products Inc, Ply-outh, MI) measured standing height to the nearest 0.25
nch. Weight was measured in pounds with a calibrated
igital scale (Body Comp Scale, American Weights and a18 June 2006 Volume 106 Number 6
easures, Rancho Santa Fe, CA) to the nearest 0.01ound. Each measure was taken twice and the mean ofhe two readings calculated. Body mass index was calcu-ated as kg/m2.
ata Analysisivariate correlations were calculated between demo-raphic variables (age, ethnicity, and body mass index)nd the dependent variables (mean daily servings of fruitnd vegetables, percentage energy from total fat) to de-ermine statistically significant covariates to include inhe multivariate models. Three sets of hierarchical mul-ivariate linear regression analyses were conducted withovariates entered in the first block and psychosocialorrelates entered in the second block. The standardizedcoefficients were evaluated to determine the magnitude
nd direction of the associations between each indepen-ent variable and the dietary variables. Overall model R2
alues indicated the total variation explained in eachodel. In addition to the entire sample, stratified analy-
es were conducted by sex and age group, in which 11- to2-year-olds were grouped as “younger” and 13- to 15-ear-olds were grouped as “older.” The stratified regres-ion analyses were conducted to determine if differentatterns of significant psychosocial correlates emergedor these different groups of adolescents.
ESULTSeans and standard deviations of the descriptive and
ependent variables for the total study sample, by sex,nd by age grouping, are presented in Table 1. For sta-istical analyses the fruit and vegetable servings/day val-es were log transformed to approximate a more normalistribution of the data. Table 1 displays the group geo-etric means for fruits and vegetables in the original
ervings/day metric.Bivariate correlations of demographic variables and
he dependent variables are presented in Table 3. Allignificant relations were entered into the first step of theierarchical regressions of the multivariate models (seeable 4).
otal Sampleor fruits and vegetables, the overall model accounted for.4% of the variance (P�0.001) with several significant psy-hosocial predictors including family support (P�0.01), prosP�0.05), child behavior change strategies (P�0.02), andealthful household eating rules (P�0.01). The overallodel for dietary fat accounted for 4.1% of the variance
P�0.001), with cons (P�0.01), child behavior change strat-gies (P�0.02), and healthful household eating rulesP�0.01) as significant predictors.
ex Subgroup Analysesmong the boys, the multivariate model for fruits andegetables accounted for 8.1% of the variance (P�0.001)ith child behavior change strategies (P�0.05) andealthful household eating rules (P�0.02) emerging asignificant predictors. For dietary fat, the overall model
ccounted for 4.9% of the variance (P�0.01), with pros
(e
ttt(s
AAfawet
f(ses(7(P
DTdaat
P�0.
P�0.05) and healthful household eating rules (P�0.001)merging as significant psychosocial predictors.For the girls, the overall model accounted for 4.6% of
he variance (P�0.01) with family influence emerging ashe only significant predictor (P�0.05). For dietary fat,he overall model accounted for 4.0% of the varianceP�0.01), and cons (P�0.01) and child behavior changetrategies (P�0.05) emerged as significant predictors.
ge Subgroup Analysesmong the younger participants, the multivariate model
or fruits and vegetables accounted for 5.8% of the vari-nce (P�0.001) and food-limiting household eating rulesas the only significant predictor (P�0.02). Peer influ-nce approached significance (P�0.07). For dietary fat,he overall model was not significant.
Table 3. Correlations between demographic variables (age, ethnicityfat) in a study of adolescent boys and girls
Total (n�839) Girls (n�45
Fruits and vegetablesAge 0.09** �0.01Ethnicity 0.02 0.03BMI �0.09* �0.10*Dietary fatAge �0.08* �0.12*Ethnicity 0.06 0.04BMI 0.04 0.04
aBMI�body mass index; calculated as kg/m2.*P�0.05.**P�0.01.
Table 4. Standardized � coefficients for significant predictors enter
Fruits and Vegetables
Predictor � wei
Total sample Age 4.0 (Family support 3.0 (Pros �2.2 (Child strategies 2.5 (Rules-healthful 2.7 (
Boys Child strategies 2.2 (Rules-healthful 2.5 (
Girls Family support 2.0 (
Young Ethnicity 2.1 (Rules-limit 2.5 (
Old Body mass index �2.2 (Self-efficacy �2.0 (Family support 3.0 (Pros �2.7 (Child strategies 2.2 (Rules-healthful 3.3 (
Among the older participants, the overall model for o
ruits and vegetables accounted for 8.3% of the varianceP�0.001) and several psychosocial variables emerged asignificant predictors: self-efficacy (P�0.05), family influ-nce (P�0.01), pros (P�0.01), child behavior changetrategies (P�0.05), and healthful household eating rulesP�0.01). For dietary fat, the overall model accounted for.2% of the variance. Significant predictors included consP�0.001) and healthful household eating rules (P�0.05).ros approached significance (P�0.055).
ISCUSSIONhis study examined theoretically based correlates ofietary behaviors, specifically mean daily servings of fruitnd vegetables and percentage of energy from total fatmong adolescent boys and girls. The findings indicatedhat household rules and strategies for change were the
a) and dependent variables (servings of fruit and vegetables, dietary
Boys (n�383) Young (n�385) Old (n�454)
0.19** 0.003 0.01�0.001 0.12* �0.05�0.07 �0.03 �0.16**
�0.03 0.10 �0.11*0.09 0.07 0.050.04 0.02 0.08
o the hierarchical regressions presented by model
Dietary Fat
Predictor � weight
001) Age �2.6 (P�0.01)01) Cons 3.1 (P�0.01)05) Child strategies �2.4 (P�0.02)02) Rules-healthful �2.9 (P�0.01)01)05) Pros �2.2 (P�0.05)02) Rules-healthful �3.7 (P�0.001)05) Age �2.5 (P�0.02)
Cons 2.7 (P�0.01)Child strategies �2.3 (P�0.05)
05)02)05) Cons 3.6 (P�0.001)05) Rules-healthful �2.2 (P�0.05)01)01)05)01)
, BMI
6)
ed int
ght
P�0.P�0.P�0.P�0.P�0.P�0.P�0.P�0.
P�0.P�0.P�0.P�0.P�0.P�0.P�0.
nly variables associated with food choices for both fruit
June 2006 ● Journal of the AMERICAN DIETETIC ASSOCIATION 819
aiaht
eadfaatccbs
hvccAoffiscdbFa
eyhirmcwcmtyc
ehcrcdcahreafa
fto(f
ldiatd
scghtc
oc(pfvbt
vpsasevshhmetia
c(moiosc
tphabs
T
8
nd vegetable intake and dietary fat. In addition, familynfluences and decisional balance variables were associ-ted with dietary outcomes. Thus, some but not all of theypothesized psychosocial correlates were supported inhe total sample.
Analyses indicated somewhat different correlates forach dietary behavior, though healthful household rulesnd behavior-change strategies were significant for bothietary outcomes. Specifically, mean daily servings ofruit and vegetables was associated with family supportnd pros, whereas percentage of energy from total fat wasssociated with cons. It is possible that a strong percep-ion of benefits (ie, pros) is needed to stimulate increasedonsumption of fruits and vegetables, whereas a low per-eption of cons to reducing desirable high-fat foods maye a critical ingredient of change. These food-specific as-ociations of pros and cons deserve further study.Male vs female differences in correlates of dietary be-
avior need to be explored to determine whether inter-entions need to be targeted for each sex. Few significantorrelates were identified for each group, potentially be-ause the sex-specific subgroups reduce variability.mong boys, household rules were related to both dietaryutcomes. Behavior-change strategies were related toruit and vegetable consumption among boys and dietaryat consumption among girls, so it is notable that teach-ng behavior-change strategies should be helpful for bothexes. Dietary fat intake was related to pros for boys andons for girls. This finding may suggest an interesting sexifference in the beliefs underlying dietary fat changes,ut the result needs to be confirmed in other studies.amily support was the only significant correlate of fruitnd vegetable intake for girls.In terms of age, it is reasonable to expect that influ-
nces will be different for younger and older youth. Forounger children, the only significant correlate wasousehold rules, which was related to fruit and vegetable
ntake. It is notable that household rules were parent-eported, so questionnaire responses from young childrenay have limited reliability that reduced observed asso-
iations. In contrast, older adolescents’ dietary behaviorsere related to almost every category of psychosocial
orrelate. This could indicate that older adolescents wereore accurate reporters of the psychosocial characteris-
ics or that the behavior-change theories apply better asoung people become more responsible for their own de-ision making.
There were five significant correlates of fruit and veg-table intake among older adolescents: family support,ousehold rules, change strategies, pros, and self-effi-acy. By contrast, only household rules and cons wereelated to dietary fat intake. This pattern of findingsould indicate that older adolescents are more involved inecisions about fruit and vegetable intake than about fatonsumption. The emergence of self-efficacy as a correlatemong older adolescents also suggests adolescents do notave substantial control about their choices until theyeach a certain age. The general pattern of age differ-nces suggests parent-focused interventions may be mostppropriate for younger adolescents, whereas programsor older adolescents may need to target both parents anddolescents.
It was surprising that peer influences were one of the t20 June 2006 Volume 106 Number 6
ew theory-based variables that were not significant con-ributors because increasing reliance on peer influence isften discussed as a hallmark of adolescent development21). Perhaps the measurement of peer influence needsurther development.
The overall regression analyses did not account for aarge portion of the variance in fruit and vegetable orietary fat consumption. This is common in other studiesnvestigating correlates of complex health behaviors suchs physical activity (22) and reinforces the need to iden-ify other hypothesized correlates of behavior to betteresign future intervention development.This study had several limitations that should be con-
idered when interpreting the findings. The sample, re-ruited through health care provider offices, may limiteneralizability of these findings to those with access toealth care. Also, although diverse in socioeconomic sta-us and ethnicity, the sample may not represent adoles-ents in other regions of the country.Nonetheless, strengths of this research include thor-
ugh assessment of dietary intake via three 24-hour re-alls and multiple measures of psychosocial constructseg, assessed from adolescent and parent). The large sam-le size allowed subgroup analyses that can provide in-ormation on which to base sex- and age-targeted inter-entions. Thus, this study adds further evidence to theody of knowledge about the correlates of dietary pat-erns among adolescent girls and boys.
Behavior-change theories hypothesize that multipleariables and processes influence behavior (8,9), and theresent study of correlates of adolescent dietary behaviorupport those propositions. Especially in the total groupnd the subgroup of older adolescents, psychological andocial variables were related to fruit, vegetable, and di-tary fat intake. There are several implications for inter-entions that need to be tested in randomized controlledtudies. The most consistently supported correlates wereousehold rules and behavior-change strategies. House-old rules, primarily provision of healthful foods, are aeans by which parents can provide a healthful food
nvironment, at least at home. Teaching parents abouthe value of household rules for healthful foods may be anmportant component of interventions for adolescents ofll ages.Teaching behavior-change strategies is central to So-
ial Cognitive Theory (6) and the Transtheoretical Model7), but present results suggest use of these strategiesay be effective only for adolescents age 13 years and
lder who are making more decisions about their dietaryntake. Because pros and cons were related to dietaryutcomes in several subgroups, teaching decision-makingkills to older adolescents may be an effective means ofhanging fruit, vegetable, and fat intake.Our findings suggest that effective dietary-change in-
erventions for adolescents likely require multiple com-onents. Parents need to be encouraged to provideealthful food environments to adolescents of all ages,nd older adolescents need to be taught to use theory-ased behavior change strategies and systematic deci-ion-making skills.
his project was supported by the National Cancer Insti-
ute (grant no. 1 R01 CA081495).
R
1
1
1
1
1
1
1
1
1
1
2
2
2
eferences1. Weinhouse S, Bal DG, Adamson R, Dwyer J, Klein-
man RE, Kritchevsky D, Mettlin CJ, Newell GR, Riv-lin R, Wattenberg L, Winick M. American CancerSociety guidelines on diet, nutrition, and cancer: Thework study group on diet, nutrition, and cancer. CACancer J Clin. 1991;41:334-338.
2. National Cancer Institute. Eat more fruits & vegeta-bles—5 a day for better health. Bethesda, MD: US-DHHS, Public Health Service 1991. NIH PublicationNo. 92-3248.
3. Centers for Disease Control and Prevention. Surveil-lance summaries, May 21, 2004. MMWR Morb MortalWkly Rep. 2004;53(no. SS-2):1-96.
4. US Department of Health and Human Services.Healthy People 2010: Understanding and ImprovingHealth. 2nd ed. Washington, DC: US GovernmentPrinting Office; November 2000.
5. Baranowski T, Lin LS, Wetter DW, Resnicow K,Hearn MD. Theory as mediating variables: Whyaren’t community interventions working as desired?Ann Epidemiol. 1997;7:S89-S95.
6. Bandura A. Human agency in social cognitive theory.Am Psychol. 1989;44:1175-1184.
7. Prochaska JO, Velicer WF. The transtheoreticalmodel of health behavior change. Am J Health Pro-mot. 1997;12:38-48.
8. Birch LL, Fisher JO. Development of eating behav-iors among children and adolescents. Pediatrics.1998;101(3Pt2):539-549.
9. Zive MM, Frank-Spohrer GC, Sallis JF, McKenzieTL, Elder JP, Berry CC, Broyles SL, Nader PR. De-terminants of dietary intake in a sample of white andMexican-American children. J Am Diet Assoc. 1998;98:1282-1289.
0. Cullen KW, Bartholomew LK, Parcel GS, Koehly L.Measuring stage of change for fruit and vegetableconsumption in 9- to 12-year-old girls. J Behav Med.1998;21:241-254.
1. Cullen KW, Baranowski T, Rittenberry L, Cosart C,Hebert D, de Moor C. Child-reported family and peerinfluences on fruit, juice, and vegetable consumption:Reliability and validity measures. Health Educ Res.2001;16:187-200.
2. Woodward DR, Boon JA, Cumming FJ, Ball PJ, Wil-
liams HJ, Hornsby H. Adolescents’ reported usage ofselected foods in relation to their perceptions andsocial norms for those foods. Appetite. 1996;27:109-117.
3. Granner ML, Sargent RG, Calderon KS, Hussey JR,Evans AE, Watkins KW. Factors of fruit and vegeta-ble intake by race, gender, and age among youngadolescents. J Nutr Educ Behav. 2004;36:173-180.
4. Resnicow K, Davis-Hearn M, Smith M, BaranowskiT, Lin LS, Baranowski J, Doyle C, Wang DT. Social-cognitive predictors of fruit and vegetable intake inchildren. Health Psychol. 1997;16:272-276.
5. Schakel SF, Sievert YA, Buzzard IM. Sources of datafor developing and maintaining a nutrient database.J Am Diet Assoc. 1988;88:1268-1271.
6. Hagler AS, Norman GJ, Radick LR, Calfas KJ, SallisJF. Comparability and reliability of paper- and com-puter-based measures of psychosocial constructs foradolescent fruit and vegetable, and dietary fat in-take. J Am Diet Assoc. 2005;105:1758-1764.
7. Velicer WF, DiClemente CC, Prochaska JO, Bran-denburg N. A decisional balance measure for assess-ing and predicting smoking status. J Pers Soc Psy-chol. 1985;48:1279-1289.
8. Rossi SR, Greene GW, Rossi JS, Plummer BA, Ben-isovich SV, Keller S, Velicer WF, Redding CA,Prochaska JO, Pallonen UE, Meier KS. Validation ofdecisional balance and situational temptations mea-sures for dietary fat reduction in a large school-basedpopulation of adolescents. Eat Behav. 2001;2:1-18.
9. Sallis JF, Pinski RB, Grossman RM, Patterson TL,Nader PR. The development of self-efficacy scales forhealth-related diet and exercise behaviors. HealthEduc Res. 1988;3:283-292.
0. Sallis JF, Taylor WC, Dowda M, Freedson PS, PateRR. Correlates of vigorous physical activity for chil-dren in grades 1 through 12: Comparing parent-re-ported and objectively measured physical activity.Pediatric Exercise Sci. 2002;14:30-44.
1. Larson R, Richards MH. Daily companionship in latechildhood and early adolescence: Changing develop-mental contexts. Child Dev. 1991;62:284-300.
2. Sallis JF, Prochaska JJ, Taylor WC. A review ofcorrelates of physical activity of children and adoles-
cents. Med Sci Sports Exerc. 2000;32:963-975.June 2006 ● Journal of the AMERICAN DIETETIC ASSOCIATION 821