tracking obesity related behaviors from childhood …

TRACKING OBESITY RELATED BEHAVIORS FROM CHILDHOOD TO

ADOLESCENCE: THE FUN 5 STUDY

A DISSERTATION SUBMITTED TO THE GRADUATE DIVISION OF THE UNIVERSITY

OF HAWAII AT MANOA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR

THE DEGREE OF

DOCTOR OF PHILOSOPHY

IN

EPIDEMIOLOGY

MAY 2014

By

Md Mahabub ul Anwar

Dissertation Committee:

Andrew Grandinetti, Chairperson

Claudio Nigg

Eric Hurwitz

Dongmei Li

Min Liu

Kamal Masaki

Keywords: Obesity, childhood, adolescent, physical activity, fruit and vegetable

consumption, sedentary behavior, theory of planned behavior, body mass index

i

Table of Contents

Table of contents ................................................................................................................... i

List of tables ........................................................................................................................ iii

List of figures ....................................................................................................................... iv

Abbreviations ....................................................................................................................... v

Abstract .............................................................................................................................. vi

Introduction ........................................................................................................................ 1

Specific Aims ........................................................................................................ 3

Framework of three papers ................................................................................................ 3

Paper 1: Tracking physical activity from childhood to adolescence: the Fun 5 study ..................................................................................................................... 7

Introduction ......................................................................................................... 7

Conceptual model: TPB to explain childhood PA ................................................ 9

Methods ............................................................................................................. 12

Results ................................................................................................................ 17

Discussion........................................................................................................... 26

Strengths and limitations of the study .............................................................. 29

Paper 2: Predictors of fruits and vegetables consumption from childhood to adolescence: the Fun 5 study ............................................................................ 30

Introduction ....................................................................................................... 30

Conceptual model: TPB to explain childhood FVC ............................................. 32

Methods ............................................................................................................. 34

Results ................................................................................................................ 39

Discussion........................................................................................................... 47

Strengths and limitations ................................................................................... 50

ii

Paper 3: Predicting adolescents’ obesity related behaviors: the Fun 5 study ................. 52

Introduction ....................................................................................................... 52

Methods ............................................................................................................. 56

Results ................................................................................................................ 61

Discussion........................................................................................................... 69

Conclusion ......................................................................................................................... 72

References ........................................................................................................................ 75

iii

List of tables

Table 1 Descriptive statistics for respondents, predictors, and outcome variables ............................................................................................................. 18

Table 2 Piecewise growth mixer modeling for growth trajectories in PA over time for cohort 1 and cohort 2 .......................................................................... 21

Table 3 Bivariate correlation among TPB constructs to explain childhood PA .............. 22

Table 4 Goodness of fit indices for the models to explain childhood PA by TPB constructs .......................................................................................................... 23

Table 5 Predicting adolescent obesity by childhood PA, adolescent PA, and demographic variables ....................................................................................... 24


Table 7 Piecewise growth mixer modeling for growth trajectories in FVC over time for cohort 1 and cohort 2 .......................................................................... 43

Table 8 Bivariate correlation among TPB constructs to explain childhood FVC ............ 44

Table 9 Goodness of fit indices for the models to explain childhood FVC by TPB constructs ........................................................................................................... 45

Table 10 Predicting adolescent obesity by childhood FVC, adolescent FVC, and demographic variables ....................................................................................... 46

Table 11 Hypotheses to predict adolescents’ obesity related behaviors ........................ 56


Table 13 Goodness of fit indices for the models to explain adolescent PA, FVC, and SB by childhood PA, FVC, and SB with all paths.......................................... 64

Table 14 Goodness of fit indices for the revised models to explain adolescent PA, FVC, and SB by childhood PA, FVC, and SB with significant paths only ............. 65

Table 15 Goodness of fit indices for the revised models to explain adolescent PA, FVC, and SB by childhood PA, FVC, and SB combining both cohorts ................ 65

iv

List of figures

Figure 1 Model of the three papers framework ................................................................ 6

Figure 2a Observed PA of 10% randomly selected students from Cohort 1 over time .................................................................................................................... 20

Figure 2b Observed PA of 10% randomly selected students from Cohort 2 over time .................................................................................................................... 20

Figure 3a Standardized path coefficients of TPB constructs with 95% confidence interval explaining PA of cohort 1-BL participants (n=251) ............................... 23

Figure 3b Standardized path coefficients of TPB constructs with 95% confidence interval explaining PA of cohort 2-BL participants (n=247) ............................... 24

Figure 4a Observed FVC of 10% randomly selected students from Cohort 1 over time .................................................................................................................... 42

Figure 4b Observed FVC of 10% randomly selected students from Cohort 2 over time .................................................................................................................... 42

Figure 5a Standardized path coefficients of TPB constructs with 95% confidence interval explaining FVC of cohort 1-BL participants (n=248) ............................. 45

Figure 5b Standardized path coefficients of TPB constructs with 95% confidence interval explaining PA of cohort 2-BL participants (n=243) ............................... 46

Figure 6a Standardized path coefficients of childhood behaviors with 95% confidence interval explaining adolescent behaviors of cohort 1 participants in revised model (n=192), significant paths only ........................... 66

Figure 6b Standardized path coefficients of childhood behaviors with 95% confidence interval explaining adolescent behaviors of cohort 2 participants in revised model (n=182), significant paths only ........................... 67

Figure 6c Standardized path coefficients of childhood behaviors with 95% confidence interval explaining adolescent behaviors of participants for both cohorts in revised model (n=374), significant paths only ......................... 68

v

Abbreviations

BL Baseline

BMI Body Mass Index

CDC Centers for Disease Control and Prevention

CFI Comparative Fit Index

CVD Cardiovascular Disease

FVC Fruit and Vegetable Consumption

MVPA Moderate to Vigorous Physical Activity

NHANES National Health and Nutritional Examination Survey

NHOPI Native Hawaiians and other Pacific Islanders

PA Physical Activity

PBC Perceived Behavioral Control

RMSEA Root Mean Squared Error of Approximation

SB Sedentary Behavior

SCT Social Cognitive Theory

SD Standard Deviation

SRMR Standardized Root Mean Square Residual

T1 One-Year Follow-Up

T2 Five-Year Follow-Up

TPB Theory of Planned Behavior

TV Television

WRMR Weighted Root Mean Square Residual

vi

Abstract

Obesity in the United States has increased dramatically, and has become a public

health crisis. Studies documented that healthy lifestyle, including healthy eating,

increased physical activity (PA), and decreased sedentary behavior (SB), can lower the

risk of becoming obese, as well as developing obesity related diseases. In order to

explore the changing pattern of PA and fruit and vegetable consumption (FVC) from

childhood to adolescence, as well as to investigate how childhood PA, FVC, and SB

influence adolescence PA, FVC, and SB, this dissertation utilized the data of Fun 5 study,

which is a longitudinal cohort study. Two cohorts of data were utilized for this

dissertation, where students were followed up for five years. Using piecewise growth

mixer modeling, and random coefficient model, paper one of this dissertation did not

find any significant pattern of PA change from childhood to adolescent. Structural

equation modeling revealed the applicability of theory of planned behavior (TPB) for

childhood PA. Findings also documented that self confidence is the strongest variable to

influence childhood PA followed by attitudes. Paper one did not find any impact of

childhood and adolescent PA on adolescent obesity. Using piecewise growth mixer

modeling, and random coefficient model, paper two revealed a declining pattern of FVC

from childhood to adolescent. Structural equation modeling revealed the applicability of

TPB for childhood FVC. Findings also documented that self confidence is the strongest

variable to influence childhood FVC followed by subjective norms. Paper two did not

find any impact of childhood and adolescent FVC on adolescent obesity. Using structural

equation modeling, paper three of this dissertation revealed that all three childhood

vii

behaviors (PA, FVC, SB) predicted the same adolescent behaviors, which implies that

these behaviors develop at childhood and continue to adolescence. Findings suggest to

target early childhood for intervention activities regarding increasing PA, FVC, and

decreasing SB, as well as target children’s self confidence, subjective norms, and

attitudes towards PA and FVC. Findings also recommend further studies to explore the

change of obesity related behaviors from childhood to adolescent, as well as the

predictive ability of these behaviors to adolescent obesity.

1

INTRODUCTION

Childhood obesity in the United States has increased dramatically over the last three

decades. The percentage of obese (BMI ≥ 95th percentile) children aged 6-11 years has

risen from 7% in 1980 to 18% in 2010, while the percentage of obese adolescents aged

12-19 years increased from 5% to 18% over the same period (1). Results from the 2009-

2010 National Health and Nutritional Examination Survey (NHANES) estimate that about

one third of children and adolescents aged 2-19 years are overweight or obese (BMI ≥

85th percentile) (1). Excess weight puts children at greater risk for elevated cholesterol,

plasma insulin, and systolic blood pressure (2, 3), which are risk factors for chronic

conditions like cardiovascular disease (CVD) (4) and type 2 diabetes (5, 6). Childhood

obesity also increases the risk for negative psycho-social consequences, such as

discrimination and stigmatization, and may lead to low self-esteem, social alienation,

lack of self-confidence, depression, and other mental health risks (7-10); moreover, it

contributes to health care costs. Wang and colleagues (11) predict that the prevalence

of overweight in children will nearly be double by 2030 contributing to approximately

16-18% of the total US healthcare costs.

This increasing prevalence of childhood obesity may be addressed by focusing on

primary prevention to avoid the burden of disease, and economic consequences.

Furthermore, as childhood obesity is a significant predictor of adult obesity (12-15),

identifying the related risk behaviors for early intervention efforts is important to

prevent adult obesity and its associated complications. Studies reveal that lack of

physical activity (PA) (16-19), poor diet (16, 20), and sedentary behavior (SB) (16-19) are

2

key risk factors for obesity. The combination of unhealthy nutrition practices and

decreased PA behaviors affect energy balance, which is the simplest explanation for

weight gain. Several studies have found declining trends of PA from childhood to

adolescence to adulthood (21-23). On the other hand, research has shown that SB,

especially TV watching, increases with age and remains relatively consistent over time

(24, 25), as well as obesity. Viner and Cole (25) report that each additional hour of

watching TV on weekends at 5 years of age increases the risk of adult obesity by 7%.

Studies further reveal that food behavior and food choices, especially fruit and

vegetable consumption (FVC), are established in childhood or adolescence and may

persist into adulthood (26-28). In addition, studies observe that many adolescents have

developed multiple health risk behaviors related to PA, SB, and FVC, and the number of

these health risks increases from childhood to adolescence (29-30).

Among the US populations, Native Hawaiians and other Pacific Islanders (NHOPI)

are especially prone to obesity and its consequences (31-33). Novotny and colleagues

(34) reveal that almost one in three NHOPI children of 5-8 years old is either obese or

overweight. To decrease the prevalence of preventable diseases, it is essential to

understand how childhood health behaviors influence adolescence. However,

longitudinal research of obesity-related behaviors, with long term following up from

childhood to adolescence, targeting the NHOPI populations, especially Hawaii, is very

limited. Therefore, this study will track obesity-related behaviors in the Hawaii

populations in the transition from childhood to adolescence.

3

Specific Aims

The specific aims of this study are:

1. To explore the changing pattern of PA from childhood to adolescence,

2. To explore the changing pattern of FVC from childhood to adolescence,

3. And to investigate how childhood PA, SB, and FVC behaviors influence

adolescence PA, SB, and FVC behaviors.

FRAMEWORK OF THREE PAPERS

Recent literature documents that regular PA improves reductions in body

adiposity, increases aerobic fitness, and contributes to a range of other health benefits

from decreased blood pressure to increased bone mass in adulthood (16-19). PA also

benefits children physically, mentally, emotionally, and socially (7, 8, 10, 17-19). Studies

reveal that time spent in moderate to vigorous physical activity (MVPA) is positively

associated with a lower childhood BMI (35, 36). Considering the health benefits of PA,

the Centers for Disease Control and Prevention (CDC) recommends at least 60 minutes

of MVPA for children every day (37). However, a significant number of US children do

not follow this recommendation (38, 39). According to Beighle (40), less than half of the

6-11 year old children met the recommendation. Moreover, the percentage of children

meeting recommended PA levels declines with age (21-23). In a pooled analysis of 26

studies, Dumith and colleagues (21) reveal a 7% PA decline per year among adolescents.

4

Conversely, several studies document that SB, especially TV watching, is

associated with multiple risk factors for negative health outcomes, including obesity (24,

25, 41-43). Proctor and colleagues (44) reveal that eating meals in front of the TV,

especially, salty snacks, pizza, and soft drinks may influence energy intake. Despite the

negative health consequences, TV watching appears to increase with age (24, 25).

Healthy diets have health benefits and help reduce obesity (45). The Dietary

Guidelines for Americans recommends a diet rich in FV, whole grains, and fat-free or

low-fat dairy products for individuals of 2 years and older. The guidelines also

recommend limiting the intake of solid fats for children, adolescents, and adults (46).

However, most young people are not following these recommendations (46-48).

Moreover, studies further reveal a decreasing pattern of FVC from childhood to

adolescence to adulthood (26-28).

Several studies document that PA, SB, and FVC are associated with each other

(29, 30, 49) and influence each other over time (49, 50). For example, Driskell and

colleagues (30) document that healthy diet, intention to be active, and PA preferences

cluster with PA. Lippke and colleagues (51) also reveal a strong correlation between

nutrition and PA. These associations of obesity risk behaviors imply that a group of risk

behaviors tend to occur together, and may be tied with common underlying factors;

thus an intervention that addresses these behaviors together could potentially be more

effective in the long run. Considering the interactions, researchers (52) suggest that

multiple behavior change interventions may have a greater impact than single behavior

change interventions.

5

Since PA, SB and FVC behaviors change from childhood to adolescents, it is also

important to identify the age when these behaviors change. Findings of such studies

may be helpful to develop age appropriate interventions. It is also important to find out

how these behaviors in childhood influence adolescent behaviors, as during adolescence,

individuals start to make their own choices and develop a personalized lifestyle.

Therefore, this study provided evidence to answer two questions regarding the

changing pattern of health behaviors from childhood to adolescence:

1. Do PA and FVC change from childhood to adolescence?

2. How do childhood PA, SB, and FVC behaviors influence adolescence PA, SB,

and FVC behaviors?

This first paper answered whether PA changed from childhood to adolescents

among Hawaii populations, and examined what accounts for behavior changes using the

constructs of the theory of planned behavior (TPB). In addition, this paper also explored

the influence of PA on obesity over time. The second paper answered the same question

for FVC. Paper #3 explored how these behaviors (PA, SB, FVC) in childhood influenced

adolescent behaviors over time. Figure 1 (next page) illustrates the model of the three

papers framework.

6

Figure 1 Model of the three papers framework

Paper 1: Tracking physical activity from childhood to adolescence

Paper 2: Predictors of fruits and vegetables consumption from childhood to adolescence

Paper 3: Predicting adolescents’ obesity related risk behaviors

Adolescence Childhood

Single behavior

Predictors PA PA Obesity

Predictors FVC FVC Obesity

Multiple behaviors

PA

SB

FVC

PA

SB

FVC

PA

SB

FVC

7

PAPER ONE: TRACKING PHYSICAL ACTIVITY FROM CHILDHOOD TO

ADOLESCENCE: THE FUN 5 STUDY

Introduction

Recent literature documents that regular physical activity (PA) helps to reduce

body adiposity; increase aerobic fitness; decrease blood pressure; and increase bone

mass in adulthood (16-19). PA also benefits children physically, mentally, emotionally,

and socially (7, 8, 10, 17-19). Studies reveal that time spent in moderate to vigorous

physical activity (MVPA) is positively associated with a lower childhood BMI (35, 36). The

Centers for Disease Control and Prevention (CDC) also recommends that children should

engage in MVPA for at least 60 minutes every day (37); however, a significant number of

US children do not follow this recommendation (38, 39). According to Beighle (40), less

than half of 6-11 year old children participate in MVPA for one hour per day. Moreover,

the percentage of children meeting the recommended PA levels declines as they grow

up (21-23). In a pooled analysis of 26 studies, Dumith and colleagues (21) reveal a 7% PA

decline per year among adolescents; and the decline is more prevalent among females

than males (53). Studies have also found that girls (54) and overweight children (55, 56)

have lower overall MVPA levels than boys and normal weight children respectively.

Since lack of PA is one of the risk factors of childhood obesity (16-19), which

increases the risk of remaining obese in adulthood as well as experiencing other long-

term adverse health consequences (57, 58), children are the primary target for PA

interventions. Interventions are based on the belief that PA changes over time (59). PA

8

tracking studies quantify whether a child will maintain his/her relative rank for activities

within his/her cohort over time (60, 61). Tracking PA levels for periods of stability

provides insights as to when the root determinants and early antecedents of PA

behavior changes occur. The tracking studies can also be used to examine whether

childhood PA values help to predict PA levels in later life (62). Such studies also suggest

how to measure behavior risks, and identify targeted interventions for the “at-risk”

children. Moreover, a high degree of association between childhood PA patterns and PA

in later life might suggest that early measurement and intervention could promote

healthy PA levels in later life. This strategy might have long-term implications for good

health as the causal relationship between PA and cardio vascular disease (CVD)

outcomes has been reported in adults (63, 64). Additionally, tracking studies help us to

understand how PA develops naturally as related to health related phenomena. For

example, knowing that PA remains stable from childhood to adolescence would provide

evidence that the root determinants of PA occur in the early life.

There are several PA tracking studies (21,60-62, 65, 66), but very few have

targeted children of ethnic minority groups (59, 67). In the US, the Native Hawaiian and

other Pacific Islander (NHOPI) populations consist of diverse minority groups and are

especially prone to obesity and its consequences (31-33). Novotny, Oshiro, and Wilkens

reveal that almost one third of 5-8 year old NHOPI children are either obese or

overweight (34). Studies also reveal a high prevalence of health risk behaviors and

obesity/overweight among Hawaiian youth (68). According to the 2008 Hawai‘i Physical

Activity and Nutrition Surveillance Report, 71% of middle school children do not follow

9

the recommended 60 minutes of MVPA every day (68). Moreover, the longitudinal

studies that track the PA level of the Hawai‘i population are very limited.

In order to develop effective PA interventions, a better understanding of PA

behavior trends is necessary (69-71). Analyzing data under a strong theoretical

framework might prove to be an efficient way to identify intervention targets and

increase PA more effectively among children (71-74). However, most studies are not

based on a strong theoretical framework (75), such as the theory of planned behavior

(TPB) or the social cognitive theory (SCT). Consequently, the results from such studies

have shown limited success in increasing PA levels (76). Baranowski, Anderson and

Carmack (71) have reviewed 23 PA intervention studies, and reveal the importance of

using behavioral theory for successful interventions. In addition, there are a few

longitudinal studies considering the TPB, especially targeting the Hawai‘i population.

Therefore, the specific objectives of this study were to:

1. track PA changes from childhood to adolescence among the Hawai‘i

population.

2. apply the TPB constructs to explain childhood PA.

The secondary objective of this study was to examine whether childhood and

adolescent PA can predict adolescent obesity.

Conceptual model: TPB to explain childhood PA

The TPB is frequently used to study a broad range of behaviors (77, 78) with a

particular focus on health (78), including PA (79,80). The theory has been used

10

extensively to predict and influence PA in a variety of studies (81). Several meta-

analyses (80, 82, 83) have shown evidence of the applicability of the TPB in the

explanation of PA intentions and behavior. Plotnikoff and colleagues have found that

TPB explains 34% of variance in PA while SCT explains 24% (82). According to the TPB, a

person’s individual motivational factors predict his/her specific behavior (84). A

behavior is predicted by the intention and the perceived behavioral control (PBC) when

it is not completely volitional. The intention is hypothesized to be determined by

attitudes, subjective norms, and PBC. An attitude represents an individual’s assessment

of their own beliefs regarding the target behavior’s agency in producing effective

outcomes. The subjective norms represent a person’s evaluation of whether significant

others want them to engage in the target behavior and include his/her motivation to

comply with the desires of those significant others. The effects of attitude and

subjective norms on the behavior are mediated by the intention. The PBC represents an

individual’s assessment of their own capacity to affect behavioral engagement.

According to the TPB, children with strong intentions are more likely to engage in

PA than the ones with weaker intentions (84). Subjective norm, individual’s attitudes,

and PBC influence the intentions to engage in PA. Children who have positive attitudes

toward PA are more likely to have strong intentions than those with negative attitudes.

For instance, children who have fun (i.e., the experiential aspect of an attitude) being

physically active are more likely to make plans to be more active. Children who perceive

that significant adults (e.g., physical education teachers) expect them to engage in PA,

and are motivated to obtain their teacher’s approval, are likely to have strong intentions

11

to participate in PA. In other words, children who are aware of their teacher’s desire for

them to be active, and want to please their teachers in that regard try to (i.e., develop

an intention to) be more active. Finally, children who have control over their PA level

under any circumstances are likely to have high intentions to perform PA compared to

those with weaker control (85).

This paper explored the changing pattern of PA, and evaluated the TPB

constructs to explain childhood PA among Hawai‘i populations. This paper also

examined the ability of childhood and adolescent PA to predict adolescent obesity. We

hypothesized that:

Hypothesis 1: Higher childhood PA precedes higher adolescent PA

Hypothesis 2: The TPB constructs explain childhood PA

a. children with favorable attitudes, strong perceptions about subjective

norms, and strong PBC, all regarding PA, will express greater intentions to

engage in PA compared to those reporting less positive cognitions

b. PBC in childhood will have a direct positive influence on childhood PA as

well as be mediated by intention

c. children with greater intentions will do more PA compared to those with

weaker intentions

d. Subjective norm or attitude in childhood will not directly predict

childhood PA but be mediated by intention

12

Hypothesis 3: Childhood and adolescent PA can predict adolescent obesity

a. children who have higher PA levels will have a lower BMI in adolescence

compared to those who have lower PA levels

b. adolescent who have higher PA levels will have a lower BMI in

adolescence compared to those who have lower PA levels

Methods

Design and sample

For this study, we used the data of two cohorts from the Fun 5 study (86,87),

which had adopted a longitudinal cohort design, and included three consecutive surveys

- baseline (BL), one-year follow-up (T1), and five-year follow-up (T2). In cohort 1, 259

students from 4th-6th grades (9-12 years old) were randomly selected in 2006 for the BL

survey. Out of 259 students, 172 participated in T1 (2007), and 116 in T2 (2011). In

cohort 2 248 students participated in the BL survey in 2007, 176 in T1 (2008), and 88 in

T2 (2012). We replicated the findings using two cohorts of students.

Description of Fun 5 study

The Fun 5 study is an evidence-based PA and nutrition promotion program,

which has been implemented in over 160 Hawai‘i state-legislated elementary A+ after-

school programs for ten years. The program has been implemented with the aim of

promoting PA and fruit and vegetable consumption (FVC) among children. The program

trained after-school staff on PA, nutrition, and sustainability. The nutrition component

13

included lessons on benefits of FV, managing portion sizes, and sugar intake. The staff

was also encouraged to be good role models in terms of snacking during A+ time, and to

acknowledge when the children brought healthy snacks. To promote FVC, the program

developed and distributed an activity booklet for students, which included word

searches, cross words, coloring pages, and word/picture matching activities addressing

local culturally appropriate FV. In order to increase the children’s PA, the program

offered after school students a variety of organized, non-competitive, non-gender-

specific, and fun physical activities in which children of all skill levels could participate

and experience success (86, 87).

Data collection procedure

After obtaining a written consent from parents, the BL surveys were

administered in after-school classes by the Fun 5 staff at the beginning of the school

year. It took about 15 minutes for the students to complete the survey. At the end of

the school year, T1 was administered in after-school classes by the Fun 5 staff. For the

T2 survey, the research staff mailed the BL participants at the addresses they provided

on the consent form. The mail included a cover letter, a survey form, and a self-

addressed and stamped envelope. The students who did not return the surveys were

reminded twice through mail, once every two weeks. Data collection was stopped after

three months. Each student received a $10 gift card for returning the follow-up survey.

The study was approved by the University of Hawai‘i Human Studies Program.

14

Measures

Demographics Children reported their age, grade, and gender in the BL survey.

At T2, they also reported their height in feet and inches, and weight in pounds. This

information was converted into centimeters and kilograms; BMI percentile was

calculated using the age- and sex-specific growth curves from the CDC (88). They also

reported their ethnicity at T2.

PA The Fun 5 study adopted the Godin Leisure-Time Exercise Questionnaire to

measure the PA of the children (89, 90). To fill in the questionnaire, children had to

indicate the number of days per week, and the minutes per day in which they engaged

in strenuous, moderate, and mild PA (not including physical education). The minutes per

day were listed as 10-minute increments starting with zero minutes and ending with 60+

minutes. All PA levels were defined and examples were given. The number of days

children engaged in each PA level was multiplied by the number of minutes spent in

each PA level each day, resulting in a separate score of minutes per week for strenuous,

moderate, and mild PA. Strenuous PA minutes per week and moderate PA minutes per

week were added together and then divided by seven to calculate MVPA minutes per

day. Only the children who reported both moderate and strenuous activities were

included in the MVPA calculation. The Godin Leisure-Time Exercise Questionnaire has

documented test-retest reliability (r = 0.81) and adequate validity (r = 0.39) when

compared to kilocalories expended per day in a sample of 5th, 8th, and 11th graders

(91). Children reported their PA at BL, T1, and T2.

15

TPB Constructs A five point Likert type scale (from “disagree a lot” to “agree a

lot”) was used to measure the TPB constructs. The intention, attitude, subjective norms,

and PBC were assessed by asking whether the children agree or disagree with these

statements “I plan to be physically active at least five times a week”; “Physical activity is

fun”; “Most people who are important to me think I should be physically active on a

regular basis”; and “I believe I can be physically active every day” respectively. Children

reported their intention, attitude, subjective norms, and PBC towards PA at BL, and T1.

Statistical Analysis

Descriptive statistics were produced for age, sex, ethnicity, BMI, and predictor

and outcome variables. To examine hypothesis 1, we developed piecewise growth mixer

modeling (92-95) of PA during BL to T1, and T1 to T2, with 2-level hierarchical linear

modeling. We also used a random coefficient model to validate our results. These more

complex analytic methods provide both the average trajectory across a group of

individuals and capture heterogeneity in individual trajectories if it exists.

To examine hypothesis 2, structural equation modeling was used. We repeated

the same model for BL data point of two cohorts. The overall fit of the models were

assessed using a number of goodness of fit indices representing absolute, comparative

and residual aspects of fit, specifically: chi-square index, comparative fit index (CFI), root

mean squared error of approximation (RMSEA), and weighted root mean square

residual (WRMR). The larger the probability associated with the chi-square, the better

the fit of the model to the data (96).The CFI >.95 indicates excellent model fit (97). An

16

RMSEA of <.05 is excellent (96). The cut-off point for WRMR is <.90. Hypothesis 2a. was

evaluated by the significance of standardized path coefficients of attitude, subjective

norms, and PBC on intention; and R2 on intention. Hypothesis 2b. was evaluated by the

significance of standardized path coefficients of PBC on intention and childhood PA.

Hypothesis 2c. was examined by the significance of standardized path coefficient of

intention on childhood PA; and R2 on childhood PA. Hypothesis 2d. was examined by the

significance of standardized path coefficients of attitude, subjective norms on childhood

PA.

A three-step multivariate linear regression was used to examine hypothesis 3.

Covariates were entered into step one, childhood PA variable into step two, and

adolescent PA variable into step three. Based on the available variables of the Fun 5

study data set, age, sex, ethnicity were assessed for statistical significance and were

included in each PA level model as covariates if significant at p < .05. In cohort 1, 259

students participated in BL survey and only 116 students followed up till the end of the

study and in cohort 2, out of 248 students, only 88 students followed up till the end of

the study. Comparisons of BL predictor and outcome variables of students who

continued the study and who lost to follow up were examined by t-test. Two cohorts of

data were analyzed separately using the same analytical process.

17

Results

Descriptive statistics

A summary of individual characteristics is provided in Table 1. Average age of

students of cohort 1 was 14.70 (SD=.87), and cohort 2 was 14.61 (SD=.92), almost half

were girls for both cohorts, most of them were from fourth grade (45.2% for cohort 1,

and 42.6% for cohort 2), the largest ethnic group represented was Asian (68.5% for

cohort 1, and 60.5% for cohort 2), and the mean BMI percentile for age for cohort 1 was

58.42 (SD = 27.28), and cohort 2 was 67.35 (SD=25.08). Overall, students reported high-

level attitude, mid level of subjective norms, intention, and PBC for both cohorts (Table

1).

18

Table 1 Descriptive statistics for respondents, predictors, and outcome variables

Characteristics Baseline(BL) One year follow-up(T1) Five year follow-up(T2)

Cohort 1

(2006)

Cohort 2

(2007)

Cohort 1

(2007)

Cohort 2

(2008)

Cohort 1

(2011)

Cohort 2

(2012)

n 259 248 172 176 116 88

Age (mean in

years and SD)

14.70(.87) 14.61(.92)

Sex (in %)

Boys 51.2 52.9

Girls 48.8 47.1

Education (in %)

4th Grade

45.2 42.6

5th Grade

39.0 40.1

6th Grade

15.8 17.4

Ethnicity

Native Hawaiian 15.7 11.1

Asian 68.5 60.5

White 11.1 19.8

Others 4.7 8.6

19

(Table continued)

Predictor and outcome variables Mean(SD)

Attitude 4.23(1.13) 4.45(.97)

Subjective

norms

3.58(1.25) 3.70(1.26)

PBC 3.95(1.18) 3.94(1.10)

Intention 3.79(1.14) 3.81(1.20)

MVPA (in

minutes/day)

42.89(29.12) 38.40(29.13) 44.92(30.78) 40.20(30.95) 47.04(28.98) 45.06(28.68)

BMI

percentile for

age

58.42(27.28) 67.35(25.08)

SD = standard deviation, PBC= perceived behavioral control, MVPA=moderate and vigorous physical

activity

Hypothesis one

Figure 2a, and 2b present findings of observed PA for randomly selected samples

of 10% of students over time from cohort 1 and cohort 2 respectively. Students’

reported PA showed no pattern for change over time. Table 2 presents findings from the

piecewise growth mixer modeling. Mean MVPA for cohort 1 students was 42.66 minutes

per day, which increased 2.59 minutes from time BL to T1, and further increased .08

minutes from time T1 to T2. Mean MVPA for cohort 2 students was 38.21 minutes per

day, which increased 3.13 minutes from time BL to T1, and further increased 2.98

20

minutes from time T1 to T2. However, the increases were not statistically significant for

both cohorts. Results from the random coefficient model revealed similar findings for PA

change over time (table not shown).

Figure 2a Observed PA of 10% randomly selected students from Cohort 1 over time

Figure 2b Observed PA of 10% randomly selected students from Cohort 2 over time

21

Table 2 Piecewise growth mixer modeling for growth trajectories in PA over time for

cohort 1 and cohort 2

Time Coefficient Standard

error t-ratio df p

Cohort 1

Intercept 42.66 1.90 22.44 242 <0.001

BL to T1 2.59 2.69 0.96 242 0.336

T1 to T2 0.08 3.45 0.02 242 0.981

Cohort 2

Intercept 38.21 1.93 19.85 235 <0.001

BL to T1 3.13 2.67 1.17 235 0.243

T1 to T2 2.98 3.54 0.84 235 0.400

Hypothesis two

The TPB model was tested separately for both cohorts. The bivariate correlations

among study variables are provided in table 3. Significant intercorrelations were

observed among TPB constructs, and all the constructs were associated with childhood

PA. The hypothesized model revealed an excellent representation of the data in

practical terms for both cohorts (table 4). Figure 3a, 3b present the path coefficients of

TPB constructs explaining children’s PA and intention at BL of cohort 1 and cohort 2

respectively. Attitude, subjective norms, and PBC explained 28-35% of the variance of

intention to childhood PA, and intention explained 19-25% of the variance of childhood

PA. Path coefficients revealed that attitude, subjective norms, and PBC were positively

related with intention for childhood PA; and PBC, and intention were also positively

22

related to PA. Path coefficients also revealed that subjective norm or attitude did not

directly predict PA, rather were mediated by intention.

Table 3 Bivariate correlation among TPB constructs to explain childhood PA

Cohort 1

Childhood PA Intention Attitude SN PBC

Childhood PA -

Intention .37** -

Attitude .22** .35** -

SN .14* .34** .24** -

PBC .28** .43** .40** .35** -

Cohort 2

Childhood PA Intention Attitude SN PBC

Childhood PA -

Intention .26** -

Attitude .12* .36** -

SN .14* .33** .27** -

PBC .29** .57** .29** .26** -

** = significant at the 0.01 level; * = significant at the 0.05 level; SN = Subjective norms; PBC = Perceived

behavior control

23

Table 4 Goodness of fit indices for the models to explain childhood PA by TPB constructs

Time Chi2 RMSEA

CFI WRMR Value df p Estimate 90%CI Pro<=0.5

Cohort 1

BL 1.58 2 .45 .00 .00-.12 .64 1.0 .35

Cohort 2

BL 1.12 2 .57 .00 .00-.11 .73 1.0 .29

RMSEA= root mean squared error of approximation; CFI= comparative fit index; WRMR= weighted root

mean square residual

Figure 3a Standardized path coefficients of TPB constructs with 95% confidence interval

explaining PA of cohort 1-BL participants (n=251)

PBC=perceived behavioral control

Attitude

PBC

Intention Childhood PA Subjective norms

.21(.10-.31)

.30(.19-.41)

.19(.10-.29)

.22(.01-.42)

.33(.19-.46)

R2 =.28

R2 =.25

24

Figure 3b Standardized path coefficients of TPB constructs with 95% confidence interval

explaining PA of cohort 2-BL participants (n=247)


Hypothesis three

Results from three step linear regression revealed that controlling for gender

and ethnicity, both childhood PA and adolescent PA were not statistically significant

predictors of adolescent BMI (table 5).

Table 5 Predicting adolescent obesity by childhood PA, adolescent PA, and demographic

variables

Variable B Std. Er. β 95%CI of B R2 ΔR2

Cohort 1

Step 1 .06

Sex -108.31 45.57 -.23 -198.74, -17.89

Age 8.95 26.16 .03 -42.97, 60.87

Attitude

PBC

Intention Childhood PA

.25(.12-.38)

.46(.34-.58)

.17(.07-.27)

.21(.02-.40)

.24(.10-.38)

R2 =.35

R2 =.19

Subjective norms

25

(Table continued)

Step 2 .06 .000

Sex -109.47 47.11 -.24 -202.98, -15.96

Age 9.62 27.06 .04 -44.01, 63.32

Childhood PA -.08 .78 -.01 -1.62, 1.46

Step 3 .09 .033

Sex -118.84 46.77 -.26 -211.68, -26.00

Age 10.75 26.72 .04 -42.28, 63.78

Childhood PA .27 .79 .04 -1.29, 1.84

Adolescent PA -1.54 .82 -.19 -3.16, .08

Cohort 2

Step 1 .04

Sex -46.71 49.19 -.11 -144.67, 51.25

Age -36.50 29.40 -.14 -95.06, 22.06

Step 2 .04 .001

Sex -46.87 49.57 -.11 -145.62, 51.89

Age -36.31 29.75 -.14 -95.57, 22.96

Childhood PA -.06 .87 -.01 -1.80, 1.69

Step 3 .06 .020

Sex -50.66 49.45 -.12 -149.19, 47.87

Age -42.36 30.00 -.17 -102.13, 17.41

Childhood PA .26 .90 .03 -1.55, 2.06

Adolescent PA -1.13 .88 -.15 -2.88, .63

26

Discussion

The specific aims of this study were to investigate PA from childhood to

adolescence, and to evaluate if the TPB constructs to explain childhood PA among

children from Hawai’i. The secondary aim was to examine the ability of childhood and

adolescent PA to predict adolescent obesity. While previous studies examined PA over

time, as well as TPB constructs to explain childhood PA, very few studies have sought to

understand these factors among NHOPI. NHOPI populations consist of diverse minority

groups and are especially prone to obesity and its consequences (31-33). Studies also

reveal a high prevalence of health risk behaviors and obesity/overweight among

Hawaiian youth (68).

Piecewise growth mixer modeling and random coefficient modeling revealed no

pattern for PA change from childhood to adolescence among Hawaiian populations. The

FLAME study (98) supports our findings of no significant pattern for change among

preschoolers. Two other longitudinal studies report no change of PA over time among

young children (99, 100). However, several other studies (21, 101, 102) have revealed a

significant PA change over time among children and adolescents. The results in these

studies, on significant patterns of change are contradictory; furthermore, the direction

of change is also contradictory. Several cross-sectional studies have documented that

PA levels generally decline at rates ranging from 1% per year to over 20% per year (103,

104). In a longitudinal study, Kimm and colleagues (105) recruited 2379 females at 9-10

years of age and followed them for 10 years and found a 35% decline in PA levels.

Dumith and colleagues (21) also noted a 7% PA decline per year among adolescents.

27

Conversely, some studies (62, 106) reported small increases (4–9%) in PA levels among

children from 10 to 15 years of age.

Since the results are contradictory, it is important to compare the studies and

conduct further research with proper methods. While almost all tracking studies have

used Spearman rank correlation or Kappa coefficient for statistical analysis, we have

used more advanced statistical methods, like piecewise growth mixer modeling, and

random coefficient modeling. Random coefficient models with participant as a random

effect were used to analyze the repeated measures at each assessment and data for all

three time points simultaneously. These models take into account the underlying

correlations between repeated observations. Both this study and the FLAME study (98)

have used random coefficient modeling to evaluate PA change pattern, and noted

similar results. Moreover, studies based on self-report measures have reported

contradictory results, while, studies that collected objectively measured data on PA

have reported somewhat more consistent results on PA tracking (99, 107, 108). In

addition, a longitudinal study allows investigators to track individuals, to see how their

PA changes over time. Therefore, a further longitudinal study with objectively measured

PA assessment is recommended to confirm the findings.

This study also investigated the TPB constructs to explain childhood PA among

NHOPI populations. Overall, the model was an excellent fit to the data for both cohorts,

meaning that our results support the TPB as an appropriate theoretical model for

explaining childhood PA behaviors. The model accounted for 28-35% of the variance in

intention, and 19-25% of the variance in PA behavior, which is similar to most TPB-PA

28

studies in children and adolescent populations (109-115). Our findings indicate that

when a child holds a positive attitude toward PA, perceives significant others believe he

or she should participate in PA, and feels that he or she has the requisite ability to

participate, that individual will form a strong intention. Similar findings have been

reported by many studies (109, 111, 115). Strong intentions will then lead to more

activity. For both cohorts, PBC was found to make the largest contribution to predicting

PA intention followed by attitude. Results also showed that PBC is more strongly

associated with intention than with PA, which is also reported in several studies (110-

115).

The findings of this study provide us insights for tailoring future intervention

designs for this population. To achieve higher activity levels among children, future

interventions should implement strategies that reflect our findings on the importance of

attitude and PBC. Findings of this study suggest that PA interventions aimed at children

should attempt to provide information highlighting the benefits of participating in

regular PA (109, 116); for example, making PA fun and enjoyable and developing

strategies to show the importance of being physically active. More importantly,

strategies in developing confidence in participating in PA when tired, busy with

homework, and engaged with other friends in sedentary activities, are required to

enhance the efficacy and control for PA behavior change. Indeed, Blanchard et al. (117)

argue that improving self-efficacy along with access to PA facilities for children may

enhance participation in PA in this target population.

29

Strengths and limitations of the study

The longitudinal design is a major strength of this study, which allowed us to

evaluate changes over time and conclude regarding the predictors of change. In addition,

two different cohorts allowed us to replicate our findings and validate our conclusions.

However, there were limitations that should be acknowledged when interpreting the

results. First, data were collected by self-report in a supervised group setting (BL and T1)

and on mail-return questionnaires (T2), which might have introduced social desirability

and recall bias. However, these methods are commonly used in childhood and

adolescent PA studies. Moreover, PA measures have documented reliability and validity

(91). Second, the ethnic composition of this population might limit the generalizability.

Third, like all longitudinal studies, this study experienced lost to follow-up; however, the

BL comparisons between participants who lost to follow up and who continued the

study showed no statistically significant difference in predictor and outcome variables.

These non significant differences however might not be enough to minimize the

selection bias. Fourth, single indicators were used for the TPB constructs, which might

have potentially limited the content validity of the constructs. However, at least one

other study (109) has successfully used single items to measure subjective norms and

intention. Moreover, our findings are similar to other TPB PA studies. Finally, this study

did not collect data on some possible confounding variables, for example, socio-

economic status, parents’ motivation towards PA, environmental factors, etc.

30

PAPER TWO: PREDICTORS OF FRUITS AND VEGETABLES CONSUMPTION FROM

CHILDHOOD TO ADOLESCENCE: THE FUN 5 STUDY

Introduction

Childhood obesity has dramatically increased and become a growing concern of

the United States over the last three decades (1). Previous studies have documented

that healthy eating behavior both in childhood and adolescence is not only important

for proper physical growth and development, but also for preventing health problems

such as obesity (43, 118-120). Furthermore, a poor diet can increase the risk for lung,

esophageal, stomach, colorectal, and prostate cancers (121). Considering the benefits of

a healthy diet, the Dietary Guidelines for Americans recommend a diet rich in fruit and

vegetable, whole grains, and fat-free and low-fat dairy products for individuals 2 years

and older. The guidelines also recommend limiting the intake of solid fats for children,

adolescents, and adults (46). However, most young people are not following these

recommendations (46-48). The state Indicator Report on Fruits and Vegetables, 2013

reveals that US adolescents consume fruit about 1.0 time and vegetables about 1.3

times per day (122). Moreover, studies further observed a decreasing pattern of the

fruit and vegetable consumption (FVC) with age (27, 28, 123).

During adolescence, individuals start to make their own choices and develop a

personalized lifestyle. Many of such choices are related to the risk factors for obesity

such as unhealthy diet patterns, increased TV watching, and lack of physical activity (PA),

which are generally modifiable. Modifying these behaviors at a young age is much easier

31

than at later ages. A longitudinal study on food choice behavior among adolescents

(124) recommends the initiation of healthy diet interventions during childhood as it

identifies that food habits develop during early life.

The Native Hawaiians and other Pacific Islanders (NHOPI) populations are prone

to obesity and its impacts on their health (31-33). Novotny, Oshiro, and Wilkens find that

almost one in three NHOPI children (5-8 years) is either obese or overweight (34).

Evidence further reveals a high prevalence of health risk behaviors and

obesity/overweight among youth in Hawai‘i (68). The state Indicator Report on Fruits

and Vegetables in 2013 reveals that Hawai’i adolescents consume less FV than the

national average (122). According to the 2008 Hawai‘i Physical Activity and Nutrition

Surveillance Report, 78% of middle school children do not consume the recommended

amount (five daily servings) of FV (68). In order to design effective interventions, it is

important to know the factors that influence FVC behavior. Moreover, information

about the tracking of such behavior at different phases of life would be useful in

planning for interventions. However, there are a lack of longitudinal studies addressing

FVC tracking among the Hawaiian populations.

Moreover, most studies lack a strong theoretical framework (75) such as the

theory of planned behavior (TPB) or the social cognitive theory (SCT). Although a

number of interventions have been implemented, most of them have experienced

limited success (76). Therefore, a better understanding of the theoretical constructs that

explain FVC behavior would be helpful to develop effective FVC interventions (70, 72,

74). However, there are few studies that adopt the TPB using longitudinal data,

32

especially targeting the Hawaii populations. Therefore, the specific objectives of this

study were to:

1. track FVC changes from childhood to adolescence among the Hawai‘i

population.

2. apply the TPB constructs to explain childhood FVC.

The secondary objective of this study was to examine whether childhood and

adolescent FVC can predict adolescent obesity.

Conceptual model: TPB to explain childhood FVC

The TPB is frequently used to study a broad range of behaviors (77, 78) with a

particular focus on health (78), including FVC (125-127). Several studies reveal the

applicability of the theory in predicting FVC among children and adolescent (125-127).

According to the TPB, a person’s individual motivational factors predict the performance

of his/her specific behavior (84). A behavior is predicted by the intention and the

perceived behavioral control (PBC) when it is not completely volitional. The intention is

hypothesized to be determined by attitudes, subjective norms, and PBCs. An attitude

represents an individual’s assessment of their own beliefs regarding the target

behavior’s agency in producing effective outcomes. The subjective norms represent a

person’s evaluation of whether significant others want them to engage in the target

behavior and include his/her motivation to comply with the desires of those significant

others. The effects of attitude and subjective norms on the behavior are mediated by

33

the intention. The PBC represents an individual’s assessment of their own capacity on

behavioral engagement.

According to the TPB, children with strong intentions to consume more FV are

more likely to do so than the ones with weaker intentions (84). Subjective norm,

individual’s attitudes, and PBC influence intentions regarding FVC. Children who have

favorable attitudes toward FVC are more likely to have strong intentions than those with

negative attitudes. For instance, children who have fun (i.e., the experiential aspect of

an attitude) eating FV are more likely to make plans to eat more. Children who perceive

that significant adults (e.g., parents) expect them to eat FV, and are motivated to obtain

the parents’ approval, are likely to have strong intentions to eat more FV. In other

words, children who are aware of their parents’ desires for them to eat more FV, and

want to please their parents in that regard, try to (i.e., develop an intention to) eat more

FV. Finally, children who feel in control of their FVC are likely to report strong intentions

to eat FV compared to children with weaker perceptions of control (125-127).

This paper explored the changing pattern of FVC, and evaluated the TPB

constructs to explain childhood FVC among Hawai‘i populations. This paper also

examined the ability of childhood and adolescent FVC to predict adolescent obesity. We

hypothesized that:

Hypothesis 1: Higher childhood FVC precedes higher adolescent FVC

34

Hypothesis 2: The TPB constructs explain childhood FVC

a. children with favorable attitudes, strong perceptions about subjective

norms, and strong PBC, all regarding FVC, will express greater intentions

to consume FVC compared to those reporting less positive cognitions

b. PBC in childhood will have a direct positive influence on childhood FVC as

well as be mediated by intention

c. children with greater intentions will eat more FV compared to those with

weaker intentions

d. Subjective norm or attitude in childhood will not directly predict

childhood FVC but be mediated by intention

Hypothesis 3: Childhood and adolescent FVC can predict adolescent obesity

a. children who have higher FVC levels will have a lower BMI in adolescence

compared to those who have lower FVC levels

b. adolescent who have higher FVC levels will have a lower BMI in

adolescence compared to those who have lower FVC levels

Methods

Design and sample





35








promoting PA and FVC among children. The program trained after-school staff on PA,

nutrition, and sustainability. The nutrition component included lessons on benefits of FV,

managing portion sizes, and sugar intake. The staff was also encouraged to be good role

models in terms of snacking during A+ time, and to acknowledge when the children

brought healthy snacks. To promote FVC, the program developed and distributed an

activity booklet for students, which included word searches, cross words, coloring pages,

and word/picture matching activities addressing local culturally appropriate FV. In order

to increase the children’s PA, the program offered after school students a variety of

organized, non-competitive, non-gender-specific, and fun physical activities in which

children of all skill levels could participate and experience success (86, 87).





36








Measures






FVC To measure FVC, children were asked to report the number of servings of

fruits and the number of servings of vegetables they consumed daily. Responses were

given using whole numbers from zero servings to 10+ servings. Examples of servings

sizes were provided (e.g. one serving of fruit = size of one baseball). These items have

documented validity and reliability in adolescents (128). Servings for fruits and servings

for vegetables were summed for a combined daily fruit and vegetable consumption

score. Only children who reported both fruit and vegetable servings were counted in the

analyses using this variable. Children reported their FVC at BL, T1, and T2.

37

TPB Constructs A five point Likert type scale (from “disagree a lot” to “agree a

lot”) was used to measure the TPB constructs. The intention, attitude, subjective norms,

and PBC were assessed by asking whether the children agree or disagree with these

statements “I plan to eat five or more servings of fruits and vegetables every day”;

“Eating fruits and vegetables is fun”; “Most people who are important to me think I

should eat five or more servings of fruits and vegetables every day; “I believe I can eat

five or more servings of fruits and vegetables every day” respectively. Children reported

their intention, attitude, subjective norms, and PBC towards FVC at BL, and T1.



and outcome variables. To examine hypothesis 1, we developed piecewise growth mixer

modeling (92-95) of FVC during BL to T1, and T1 to T2, with 2-level hierarchical linear

modeling. We also used a random coefficient model to validate our results. These more

complex analytic methods provide both the average trajectory across a group of

individuals and capture heterogeneity in individual trajectories if it exists.

To examine hypothesis 2, structural equation modeling was used. We repeated

the same model for BL data point of two cohorts. The overall fit of the models were



mean squared error of approximation (RMSEA), and weighted root mean square

residual (WRMR). The larger the probability associated with the chi-square, the better

38

the fit of the model to the data (96).The CFI >.95 indicates excellent model fit (97). An

RMSEA of <.05 is excellent (96). The cut-off point for WRMR is <.90. Hypothesis 2a. was

evaluated by the significance of standardized path coefficients of attitude, subjective

norms, and PBC on intention; and R2 on intention. Hypothesis 2b. was evaluated by the

significance of standardized path coefficients of PBC on intention and childhood FVC.

Hypothesis 2c. was examined by the significance of standardized path coefficient of

intention on childhood FVC; and R2 on childhood FVC. Hypothesis 2d. was examined by

the significance of standardized path coefficients of attitude, subjective norms on

childhood FVC.

A three-step multivariate linear regression was used to examine hypothesis 3.

Covariates were entered into step one, childhood FVC into step two, and adolescent FVC

into step three. Based on the available variables of the Fun 5 study data set, age, sex,

ethnicity were assessed for statistical significance and were included in each FVC level

model as covariates if significant at p < .05. In cohort 1, 259 students participated in BL

survey and only 116 students followed up till the end of the study and in cohort 2, out of

248 students, only 88 students followed up till the end of the study. Comparisons of BL

predictor and outcome variables of students who continued the study and who lost to

follow up were examined by t-test. Two cohorts of data were analyzed separately using

the same analytical process.

39

Results







58.42 (SD = 27.28), and cohort 2 was 67.35 (SD=25.08). Overall, students reported mid

level attitude, subjective norms, intention, and PBC for both cohorts (Table 6).

40



Cohort 1

(2006)

Cohort 2

(2007)

Cohort 1

(2007)

Cohort 2

(2008)

Cohort 1

(2011)

Cohort 2

(2012)

n 259 248 172 176 116 88

Age (mean in

years and SD)

14.70(.87) 14.61(.92)

Sex (in %)

Boys 51.2 52.9

Girls 48.8 47.1

Education (in %)

4th Grade

45.2 42.6

5th Grade

39.0 40.1

6th Grade

15.8 17.4

Ethnicity


Asian 68.5 60.5

White 11.1 19.8

Others 4.7 8.6

41

(Table continued)


Attitude 3.33(1.43) 3.73(1.23)

Subjective

norms

3.68(1.27) 3.91(1.25)

PBC 3.62(1.32) 3.72(1.28)

Intention 3.50(1.29) 3.58(1.31)

FVC (servings

per day)

6.84(4.53) 7.40(4.78) 7.04(4.68) 6.65(4.23) 4.64(2.87) 4.07(2.75)

BMI

percentile

for age

58.42(27.28) 67.35(25.08)

SD = standard deviation, PBC= perceived behavioral control, FVC = fruit and vegetable consumption

Hypothesis one

Figure 4a, and 4b present findings of observed FVC for randomly selected

samples of 10% of students over time from cohort 1 and cohort 2 respectively. Students’

reported FVC showed a decreasing pattern for change over time. Table 7 presents

findings from the piecewise growth mixer modeling. Mean FVC for cohort 1 students

was 6.84 servings per day, which slightly increased 0.26 servings per day from time BL to

T1, and decreased 2.50 servings per day from time T1 to T2. Mean FVC for cohort 2

students was 7.41 servings per day, which decreased 0.78 servings per day from time BL

to T1, and further decreased 2.30 servings from time T1 to T2. Decrease in all time

42

points were statistically significant. However, the increase in FVC of cohort 1 from BL to

T1 was not statistically significant. Results from the random coefficient model revealed

similar findings for FVC change over time (table not shown).

Figure 4a Observed FVC of 10% randomly selected students from Cohort 1 over time

Figure 4b Observed FVC of 10% randomly selected students from Cohort 2 over time

43

Table 7 Piecewise growth mixer modeling for growth trajectories in FVC over time for

cohort 1 and cohort 2

Time Coefficient Standard

error t-ratio df p

Cohort 1

Intercept 6.84 0.29 23.60 246 <0.001

BL to T1 0.26 0.36 0.73 246 0.464

T1 to T2 -2.50 0.38 -6.56 246 <0.001

Cohort 2

Intercept 7.41 0.31 23.56 235 <0.001

BL to T1 -0.75 0.36 -2.05 235 0.042

T1 to T2 -2.30 0.41 -5.65 235 <0.001

Hypothesis two

The TPB model was tested separately for both cohorts. The bivariate correlations

among study variables are provided in table 8. Significant intercorrelations were

observed among TPB constructs, and all the constructs were associated with childhood

FVC. The hypothesized model revealed an excellent representation of the data in

practical terms for both cohorts (table 9). Figure 5a, 5b present the path coefficients of

TPB constructs explaining children’s FVC and intention at BL of cohort 1 and cohort 2

respectively. Attitude, subjective norms, and PBC explained 50-51% of the variance of

intention to childhood FVC, and intention explained 15-17% of the variance of childhood

FVC. Path coefficients revealed that attitude, subjective norms, and PBC were positively

related with intention for childhood FVC; and PBC, and intention were also positively

44

related to FVC. Path coefficients also revealed that subjective norm or attitude did not

directly predict FVC, rather were mediated by intention.

Table 8 Bivariate correlation among TPB constructs to explain childhood FVC

Cohort 1

Childhood FVC Intention Attitude SN PBC

Childhood FVC -

Intention .42** -

Attitude .25** .36** -

SN .28** .51** .33** -

PBC .45** .65** .31** .46** -

Cohort 2

Childhood FVC Intention Attitude SN PBC

Childhood FVC -

Intention .32** -

Attitude .26** .43** -

SN .22** .51** .24** -

PBC .34** .60** .39** .41** -

** = significant at the 0.01 level; SN = Subjective norms; PBC = Perceived behavior control

45

Table 9 Goodness of fit indices for the models to explain childhood FVC by TPB

constructs

Time Chi2 RMSEA

CFI WRMR Value df p Estimate 90%CI Pro<=0.5

Cohort 1

BL 0.82 1 .37 .00 .00-.16 .50 1.0 .25

Cohort 2

BL 0.25 1 .62 .00 .00-.14 .71 1.0 .14

RMSEA= root mean squared error of approximation; CFI= comparative fit index; WRMR= weighted root


Figure 5a Standardized path coefficients of TPB constructs with 95% confidence interval

explaining FVC of cohort 1-BL participants (n=248)


Attitude

PBC

Intention Childhood FVC

Subjective norms

.14(.05-.22)

.60(.50-.70)

.27(.16-.38)

.55(.03-.1.08)

.75(.37-1.13)

R2 =.50

R2 =.17

46

Figure 5b Standardized path coefficients of TPB constructs with 95% confidence interval

explaining FVC of cohort 2-BL participants (n=243)


Hypothesis three

Results from three step linear regression revealed that controlling for gender,

both childhood FVC and adolescent FVC were not statistically significant predictors of

adolescent BMI (table 10).

Table 10 Predicting adolescent obesity by childhood FVC, adolescent FVC, and

demographic variables

Variable B Std. Er. β 95%CI of B R2 ΔR2

Cohort 1

Step 1 .06

Sex -120.86 45.82 -.25 -211.70, -30.01

Attitude

PBC

Intention Childhood FVC

.26(.15-.36)

.45(.35-.54)

.37(.21-.49)

.60(.04-1.16)

.60 (.12-1.09)

R2 =.51

R2 =.15

Subjective norms

47

(Table continued)

Step 2 .06 .000

Sex -120.40 46.10 -.25 -211.82, -28.98

Childhood FVC -.96 5.35 -.02 -11.57, 9.65

Step 3 .06 .000

Sex -120.48 46.36 -.25 -212.43, -28.53

Childhood FVC -1.04 5.67 -.02 -12.28, 10.21

Adolescent FVC .37 8.66 .01 -16.81, 17.55

Cohort 2

Step 1 .03

Sex -72.71 49.29 -.16 -170.72, 25.30

Step 2 .03 .008

Sex -68.81 49.62 -.15 -167.50, 29.88

Childhood FVC -4.21 5.18 -.09 -14.51, 6.09

Step 3 .06 .026

Sex -72.83 49.31 -.16 -170.93, 25.26

Childhood FVC -6.42 5.34 -.13 -17.04, 4.21

Adolescent FVC 14.07 9.32 .17 -4.46, 32.61

Discussion

Several studies have revealed tracking FVC from childhood to adolescence (27,

123, 129-134), and some of them have observed decreasing FVC (27, 123, 129-131, 134)

over time. However, none of these studies have examined whether the decrease is

48

statistically significant or not. They just focused on tracking FVC from childhood to

adolescence. They defined tracking as the “stability overtime”. In other word, they

either measured the tracking as the stability in the rank at a group label, or calculated

the proportion of participants that stayed in the same group over time. Since FVC

stability study cannot provide a direction of significant change, findings lack insight to

develop age appropriated intervention, for example, findings do not suggest whether

intervention is necessary to increase FVC and or what age to target. However, two

studies (28, 133) have examined the significant change of FVC over time. But, they have

only used two time points, which also lack the tracking pattern. This study, therefore,

tried to study the FVC from childhood to adolescence with three time points’ data,

which would help to determine the ideal age for intervention.

The specific aims of this study were to investigate FVC from childhood to

adolescence, and to evaluate if the TPB constructs to explain childhood FVC among

children from Hawai’i. The secondary aim was to examine the ability of childhood and

adolescent FVC to predict adolescent obesity. While previous studies examined FVC

over time, as well as TPB constructs to explain childhood FVC, very few studies have

sought to understand these factors among NHOPI. NHOPI populations consist of diverse

minority groups and are especially prone to obesity and its consequences (31-33).

Studies also reveal a high prevalence of health risk behaviors and obesity/overweight

among Hawaiian youth (68).

Piecewise growth mixer modeling and random coefficient modeling revealed

significant decline of FVC from childhood to adolescent for both cohorts. Several studies

49

observed the same results (27, 28, 123, 129-131, 133), although they did not test the

significance. Wang and colleagues (132) noted that food intake and dietary habits

normally change during this time due to individual factors like physiological

development, changes in parental influence and social and environmental changes.

Patterson and colleagues (27) also argued that children are restricted by what parents

and schools provide them to eat, while, adolescents to a lesser degree as both

autonomy and the influence of peers increases. At the same time, young adults are

usually independent and may have more control over their food choice. For example,

soft drinks consumption increases in young adulthood (135, 136). The findings of

decreasing trend of FVC from childhood to adolescence imply that interventions need to

target children to increase their FVC, as well as continue over time. In this regard, it is

important to know the determinants of childhood FVC, which is another aim of this

study.

This study applied TPB constructs to explain childhood FVC among NHOPI

populations. Overall, the model was an excellent fit to the data for both cohorts,

meaning that our results support the TPB as an appropriate theoretical model for

explaining childhood FVC behaviors. The model accounted for 50-51% of the variance in

intention, and 15-17% of the variance in FVC behavior, which is similar to most TPB-FVC

studies in children and adolescent populations (125-127, 137, 138). Our findings indicate

that when a child holds a positive attitude toward FVC, perceives significant others

believe he or she should consume more, and feels that he or she has the requisite ability

to consume, that individual will form a strong intention. Similar findings have been

50

reported by many studies (125-127, 137, 138). Strong intentions will then lead to more

consumption. For both cohorts, PBC was found to make the largest contribution to

predicting FVC intention followed by subjective norms. However, several studies found

attitude is a stronger predictor of FVC intentions compared to subjective norms (125,

137), which contradicts with this study. This might be due to the age of the respondents.

Those studies were conducted on youth, who have more control over their food choice

(27), while this study was conducted on elementary school children. This study revealed

that a subjective norm is a stronger predictor if FVC intention, which is reported in

several studies (125, 139). The subjective norm finding is likely due to the parental

influence of eating behaviors during childhood.

The findings of this study provide us insights for tailoring future intervention

designs for this population. To achieve more FVC among children, future interventions

should implement strategies that reflect our findings on the importance of subjective

norms and PBC. Findings of this study suggest that subjective norms and PBC have

important role on FVC intention, therefore, encouraging parents and schools to enable

access to more FV, providing self regulation behavioral skills for FVC, as well as

encouraging them by informing the benefits of FVC seem beneficial.

Strengths and limitations of the study

The longitudinal design is a major strength of this study, which allowed us to

evaluate changes over time and conclude regarding the predictors of change. In addition,

two different cohorts allowed us to replicate our findings and validate our conclusions.

51

However, there were limitations that should be acknowledged when interpreting the

results. First, data were collected by self-report in a supervised group setting (BL and T1)

and on mail-return questionnaires (T2), which might have introduced social desirability

and recall bias. However, Baranoski and Domel have found that children are able to give

accurate dietary information and are well aware of the foods they have eaten (140).

Moreover, FVC measures have documented reliability and validity (128). Second, the

ethnic composition of this population might limit the generalizability. Third, like all

longitudinal studies, this study experienced lost to follow-up; however, the BL

comparisons between participants who lost to follow up and who continued the study

showed no statistically significant difference in predictor and outcome variables. These

non significant differences however might not be enough to minimize the selection bias.

Fourth, single indicators were used for the TPB constructs, which might have potentially

limited the content validity of the constructs. However, at least one other study (109)

has successfully used single items to measure subjective norms and intention; and

another study (126) to measure PBC. Moreover, our findings are similar to other TPB

FVC studies. Finally, this study did not collect data on some possible confounding

variables, for example, socio-economic status, parents’ motivation towards FVC,

availability of FV at home and school, environmental factors, etc.

52

PAPER THREE: PREDICTING ADOLESCENTS’ OBESITY RELATED BEHAVIORS: THE

FUN 5 STUDY

Introduction

Childhood obesity has dramatically increased and become a growing concern of the

United States over the last three decades (1). There is no single or simple solution to

addressing the childhood obesity epidemic, but a healthy lifestyle, including healthy

eating and physical activity (PA), can lower the risk of becoming obese (140). Recent

literature document that regular PA improves reductions in body adiposity, increases

aerobic fitness, and contributes to a range of other health benefits from decreased

blood pressure to increased bone mass in adulthood (16-19). PA also benefits children

physically, mentally, emotionally, and socially (7, 8, 10, 17-19). Studies reveal that time

spent in moderate to vigorous physical activity (MVPA) is negatively associated with the

change of childhood BMI (35, 36). Considering the health benefits of PA, the Centers for

Disease Control and Prevention (CDC) recommends at least 60 minutes of MVPA for

children every day (37). However, a significant number of US children do not follow this

recommendation (38, 39). According to Beighle (40), less than half of 6-11 year old

children met the recommendation. Moreover, the percentage of children meeting

recommended PA levels declines with age (21-23). In a pooled analysis of 26 studies,

Dumith and colleagues (21) reveal a 7% PA decline per year among adolescents.

Conversely, several studies document that sedentary behavior (SB), especially TV

watching, is associated with multiple risk factors for negative health outcomes, including

53

obesity (24, 25, 41-43). In addition, Proctor and colleagues (44) reveal that eating meals

in front of the TV, especially, salty snacks, pizza, and soft drinks may influence energy

intake. Despite the negative health consequences, TV watching appears to increase with

age (24, 25).

Healthy diets also have positive effects on health, including obesity (45). The

Dietary Guidelines for Americans recommend a diet rich in fruit and vegetable, whole

grains, and fat-free and low-fat dairy products for individuals 2 years and older. The

guidelines also recommend limiting the intake of solid fats for children, adolescents, and

adults (46). However, most young people are not following these recommendations (46-

48). Moreover, studies further reveal a tracking pattern of fruit and vegetable

consumption (FVC) from childhood to adolescence to adulthood (26-28).

Several studies document that PA, SB, and FVC are associated with each other

(29, 30, 49) and influence each other over time (49, 50). For example, Driskell and

colleagues (30) document that healthy diet, intention to be active, and PA preferences

cluster with PA. Lippke and colleagues (51) also reveal a strong correlation between

nutrition and PA. These associations of obesity risk behaviors imply that a group of risk

behaviors tend to occur together, and may be tied with common underlying factors. An

intervention that addresses these behaviors together could potentially be more

effective in the long run. Considering the interactions, researchers (52) suggest that

multiple behavior change interventions may have a greater impact than single behavior

change interventions.

54

Since PA, SB and FVC change over time from childhood to adolescences, it is also

important to identify the age when these behaviors change. Findings of such studies

may be helpful to develop age appropriate interventions. It is also important to find out

how these behaviors interact in changing from childhood to adolescence, as during

adolescence, individuals start to make their own choices and develop a personalized

lifestyle.

Among the US, the native Hawaiian and other Pacific Islander (NHOPI)

populations consists of diverse minority groups and are especially prone to obesity and

its consequences (31-33). Novotny, Oshiro, and Wilkens found that almost one third of 5-

8 year old NHOPI children are either obese or overweight (34). Studies reveal a high

prevalence of health risk behaviors and obesity/overweight among Hawaiian youth (68).

According to the 2008 Hawai‘i Physical Activity and Nutrition Surveillance Report, 71%

of the middle school children do not follow the recommended 60 minutes of MVPA and

most (78 %) do not consume the recommended amount (five daily servings) of FV.

Moreover, this report shows that Pacific Islander and Asian middle school students are

more inactive (those engaging in no MVPA in the past seven days) than all other

ethnicities (68). Failure to meet recommended guidelines for PA, FVC, and SB places

children at an increased risk for overweight/obesity and for chronic conditions such as

cardiovascular disease and diabetes (1). Understanding how multiple health risk

behaviors track overtime is essential to developing effective disease prevention

strategies in populations with high percentages of NHOPI population. Thus, the primary

55

aim of this study was to find out how PA, FVC, and SB influence with each other in

tracking from childhood to adolescence. We hypothesized that:

a. children with higher PA, FVC, and lower SB will have higher PA at adolescents

compared to those with lower PA, FVC, and higher SB. We examined this from BL

to T1, T1 to T2, and BL to T2.

b. children with higher PA, FVC, and lower SB will have higher FVC at adolescents



c. children with higher PA, FVC, and lower SB will have lower SB at adolescents



Table 11 presents the hypotheses in an abridged format.

56

Table 11 Hypotheses to predict adolescents’ obesity related behaviors

Hypotheses BL to T1 Hypotheses T1 to T2 Hypotheses BL to T2

BL T1 T1 T2 BL T2

Hypothesis a

PA + PA PA + PA PA + PA

FVC + PA FVC + PA FVC + PA

SB - PA SB - PA SB - PA

Hypothesis

b

PA + FVC PA + FVC PA + FVC

FVC + FVC FVC + FVC FVC + FVC

SB - FVC SB - FVC SB - FVC

Hypothesis c

PA - SB PA - SB PA - SB

FVC - SB FVC - SB FVC - SB

SB + SB SB + SB SB + SB

BL = baseline, T1 = one-year follow-up, T2 = five-years follow-up, PA = physical activity, FVC = fruits and

vegetable consumption, SB = sedentary behavior

Methods

Design and sample





57








promoting PA and FVC among children. The program trained after-school staff on PA,

nutrition, and sustainability. The nutrition component included lessons on benefits of FV,

managing portion sizes, and sugar intake. The staff was also encouraged to be good role

models in terms of snacking during A+ time, and to acknowledge when the children

brought healthy snacks. To promote FVC, the program developed and distributed an

activity booklet for students, which included word searches, cross words, coloring pages,

and word/picture matching activities addressing local culturally appropriate FV. In order

to increase the children’s PA, the program offered after school students a variety of

organized, non-competitive, non-gender-specific, and fun physical activities in which

children of all skill levels could participate and experience success (86, 87).




58









Measures






PA The Fun 5 study adopted the Godin Leisure-Time Exercise Questionnaire to

measure the PA of the children (89, 90). To fill in the questionnaire, children had to

indicate the number of days per week, and the minutes per day in which they engaged

in strenuous, moderate, and mild PA (not including physical education). The minutes per

day were listed as 10-minute increments starting with zero minutes and ending with 60+

minutes. All PA levels were defined and examples were given. The number of days

children engaged in each PA level was multiplied by the number of minutes spent in

59

each PA level each day, resulting in a separate score of minutes per week for strenuous,

moderate, and mild PA. Strenuous PA minutes per week and moderate PA minutes per

week were added together and then divided by seven to calculate MVPA minutes per

day. Only the children who reported both moderate and strenuous activities were

included in the MVPA calculation. The Godin Leisure-Time Exercise Questionnaire has

documented test-retest reliability (r = 0.81) and adequate validity (r = 0.39) when

compared to kilocalories expended per day in a sample of 5th, 8th, and 11th graders

(91). Children reported their PA at BL, T1, and T2.

FVC To measure FVC, children were asked to report the number of servings of

fruits and the number of servings of vegetables they consumed daily. Responses were

given using whole numbers from zero servings to 10+ servings. Examples of servings

sizes were provided (e.g. one serving of fruit = size of one baseball). These items have

documented validity and reliability in adolescents (128). Servings for fruits and servings

for vegetables were summed for a combined daily fruit and vegetable consumption

score. Only children who reported both fruit and vegetable servings were counted in the

analyses using this variable. Children reported their FVC at BL, T1, and T2.

SB To measure sedentary behavior, children were asked to report the number of

hours per day that they watch TV, play videogames, and use the Internet (not for

homework). Responses were given using whole hours from zero hours to 10+ hours.

This measure has documented test-retest reliabilities in a sample of college students

(141). Children reported their SB at BL, T1, and T2.

60



and outcome variables. Baseline demographic characteristics and outcome variables

were compared between the students who completed three time points and the

students lost to follow up. Assumptions of multivariate normality were assessed by

comparing two estimating methods: maximum likelihood method, and maximum

likelihood robust method. To examine the hypothesis structural equation modeling was

used. We repeated the same model for both cohorts. The overall fit of the models were



mean squared error of approximation (RMSEA), and standardized root mean square

residual (SRMR). The larger the probability associated with the chi-square, the better

the fit of the model to the data (96). The CFI >.95 indicates excellent model fit (97). An

RMSEA of <.05 is excellent (96). The cut-off point for SRMR is <.90. Hypothesis a. was

evaluated by the significance of standardized path coefficients of BL PA, FVC, and SB on

T1 PA; standardized path coefficients of T1 PA, FVC, and SB on T2 PA; and standardized

path coefficients of BL PA, FVC, and SB on T2 PA. Hypothesis b. was evaluated by the

significance of standardized path coefficients of BL PA, FVC, and SB on T1 FVC;

standardized path coefficients of T1 PA, FVC, and SB on T2 FVC; and standardized path

coefficients of BL PA, FVC, and SB on T2 FVC. Hypothesis c. was examined by the

significance of standardized path coefficient of BL PA, FVC, and SB on T1 SB;

61

standardized path coefficients of T1 PA, FVC, and SB on T2 SB; and standardized path

coefficients of BL PA, FVC, and SB on T2 SB.

Results







58.42 (SD = 27.28), and cohort 2 was 67.35 (SD=25.08). On average students of cohort 1

had 42.89 (SD = 29.12) minutes of MVPA per day at BL, 44.92 (SD = 30.78) at T1, and

47.04 (SD = 28.98) at T2. Students of cohort 2 had 38.40 (SD = 29.13), 40.20 (SD = 30.95),

and 45.06 (SD = 28.68) minutes of MVPA at BL, T1, and T2 respectively. Students of

cohort 1 consumed 6.84 (SD = 4.53), 7.04 (SD = 4.68), and 4.64 (SD =2.87) servings of FV

per day at BL, T1and T2 respectively. Whereas, cohort 2 students consumed 7.40 (SD =

4.78), 6.65 (SD = 4.23), and 4.07 (SD = 2.75) servings of FV at same time points

respectively. On average students of cohort 1 watched TV 4.05 (SD = 2.82) hours per day

at BL, 3.76 (SD = 2.99) at T1, and 3.16 (SD = 1.77) at T2. Students of cohort 2 watched

3.99 (SD = 3.27), 4.29 (SD = 3.22), and 3.61 (SD = 2.31) hours per day at BL, T1, and T2

respectively (Table 12). Comparison of maximum likelihood and maximum likelihood

robust estimation revealed that the assumptions of multivariate normality were met.

62



Cohort 1

(2006)

Cohort 2

(2007)

Cohort 1

(2007)

Cohort 2

(2008)

Cohort 1

(2011)

Cohort 2

(2012)

n 259 248 172 176 116 88

Age (mean in

years and SD)

14.70(.87) 14.61(.92)

Sex (in %)

Boys 51.2 52.9

Girls 48.8 47.1

Education (in %)

4th Grade

45.2 42.6

5th Grade

39.0 40.1

6th Grade

15.8 17.4

Ethnicity


Asian 68.5 60.5

White 11.1 19.8

Others 4.7 8.6

63

(Table continued)


MVPA (in

minutes/day)

42.89(29.12) 38.40(29.13) 44.92(30.78) 40.20(30.95) 47.04(28.98) 45.06(28.68)

FVC

(servings per

day)

6.84(4.53) 7.40(4.78) 7.04(4.68) 6.65(4.23) 4.64(2.87) 4.07(2.75)

Hours of TV

watching/day

4.05(2.82) 3.99(3.27) 3.76(2.99) 4.29(3.22) 3.16(1.77) 3.61(2.31)

BMI

percentile for

age

58.42(27.28) 67.35(25.08)

SD = standard deviation, MVPA = moderate to vigorous physical activity, FVC = fruit and vegetable

consumption

Comparative analysis on baseline demographic characteristics and outcome variables

revealed no significant differences between the students who completed three time

points and the students lost to follow up (table not shown).

Hypothesis testing

The hypothesized model revealed a poor fit of the data in practical terms for

both cohorts (table 13). The model was revised with the non significant paths deleted

and the revised model revealed a better fit of the data (table 14). Figure 6a, and 6b

present the revised model’s path coefficients of childhood obesity related behaviors on

adolescent obesity related behaviors for cohort 1 and cohort 2 respectively. In cohort 1,

64

T1 PA was significantly predicted by BL PA and FVC, but not BL SB. For T2 PA, T1 PA, FVC

and SB did not significantly predict T2 PA. However, BL PA significantly predicted T2 PA

(Figure 6a). For cohort 2, T1 PA was significantly predicted by BL PA and FVC like cohort

1, however, the direction of BL FVC to T1 PA was negative. T2 PA was significantly

predicted by T1 PA, and BL FVC. T1 SB did not predict T2 PA (Figure 6b).

Only BL FVC significantly predicted T1 FVC, and T1 FVC predicted T2 FVC in

cohort 1. In cohort 2, BL FVC significantly predicted T1 FVC, but T1 FVC did not predict

T2 FVC. However, BL FVC significantly predicted T2 FVC. Only BL SB significantly

predicted T1 SB, and T1 SB predicted T2 SB in cohort 1. In cohort 2 BL SB significantly

predicted T1 SB, but T1 SB did not predict T2 SB.

Table 13 Goodness of fit indices for the models to explain adolescent PA, FVC, and SB by

childhood PA, FVC, and SB with all paths

Time Chi2 RMSEA

CFI SRMR Value df p Estimate 90%CI Pro<=0.5

Cohort 1 9.47 3 .02 .11 .03-.19 .09 .93 .03

Cohort 2 8.16 3 .04 .10 .02-.18 .13 .96 .03

RMSEA= root mean squared error of approximation; CFI= comparative fit index; SRMR= standardized root


To improve power and interpretability, the possibility of combining the cohorts was

explored. Baseline demographics and all three behaviors were compared between

cohorts at each time points using t-tests. None of the comparisons were significant

65

(p>.05). Thus the two cohorts were merged. Similar to the individual cohort analysis, the

overall model fit was not adequate. The model was revised with non-significant paths

deleted and revealed a better fit of the data (table 15). Figure 6c presents the revised

model’s path coefficients of childhood obesity related behaviors on adolescent obesity

related behaviors for both cohorts combined. BL PA predicted T1 PA and T2 PA. T2 PA

was also predicted by BL FVC. BL FVC predicted T1 FVC and T2 FVC. T2 FVC was also

predicted by T1 FVC. BL SB predicted T1 SB, and T1 SB predicted T2 SB.

Table 14 Goodness of fit indices for the revised models to explain adolescent PA, FVC,

and SB by childhood PA, FVC, and SB with significant paths only

Time Chi2 RMSEA


Cohort 1 24.70 19 .17 .04 .00-.08 .62 .94 .06

Cohort 2 19.81 20 .47 .00 .00-.06 .86 1.00 .06



Table 15 Goodness of fit indices for the model to explain adolescent PA, FVC, and SB by

childhood PA, FVC, and SB combining both cohorts

Time Chi2 RMSEA


With all paths 14.13 3 .01 .10 .05-.16 .05 .95 .03

Significant paths only 23.79 21 .30 .02 .00-.05 .96 .99 .04



66

Figure 6a Standardized path coefficients of childhood behaviors with 95% confidence interval explaining adolescent

behaviors of cohort 1 participants in revised model (n=192), significant paths only

BL = baseline, T1 = one-year follow-up, T2 = five-years follow-up, PA=physical activity, FVC = fruit and vegetable consumption, SB = sedentary activity

BL PA

BL FVC

T1 PA

T1 FVC

T2 PA

T2 FVC

Childhood Adolescence

.30(.17, .43)

.37(.26, .49)

.15(.03, .28)

BL SB T1 SB T2 SB

.41(.30, .52)

R2 = .11

R2 = .17

R2 = .14

.20(.03, .38)

.36(.20, .52)

R2 = .05

R2 = .15

R2 = .04

.23(.09, .37)

67

Figure 6b Standardized path coefficients of childhood behaviors with 95% confidence interval explaining adolescent

behaviors of cohort 2 participants in revised model (n=182), significant paths only


BL PA

BL FVC

T1 PA

T1 FVC

T2 PA

T2 FVC


.35(.24, .47)

.46(.36, .56)

-.18 (-.31, -.06)

BL SB T1 SB T2 SB

.46(.36, .56)

R2 = .13

R2 = .21

R2 = .21

.38(.21, .56)

-.21(-.41, -.01)

.29(.09, .49)

R2 = .24

R2 = .17

.38(.21, .56)

.36(.20, .51)

68

Figure 6c Standardized path coefficients of childhood behaviors with 95% confidence interval explaining adolescent

behaviors of participants for both cohorts in revised model (n=374), significant paths only


BL PA

BL FVC

T1 PA

T1 FVC

T2 PA

T2 FVC


.32(.24, .41)

.42(.34, .49) BL SB T1 SB T2 SB

.42(.35, .50)

R2 = .10

R2 = .18

R2 = .17

.20(.07, .34)

.16(.03, .30)

R2 = .10

R2 = .10

R2 = .04

.16(.04, .29)

.17(.06, .29)

.21(.09, .33)

69

Discussion

With an aim to evaluate the predictability of childhood obesity related behaviors

(PA, FVC, and SB) to adolescent behaviors (PA, FVC, and SB), this study used the data of

the Fun 5 study. While several cross-sectional and longitudinal studies examined the

relationship between PA and FVC (51, 142-158), or SB and FVC (159-162), or PA and SB

(149, 163-168), very few longitudinal study examined these three (PA, FVC, and SB)

behaviors together. Moreover, longitudinal study on the relationship of these three

behaviors targeting NHOPI populations is very limited. NHOPI populations consist of

diverse minority groups and are especially prone to obesity and its consequences (31-

33). Studies also document a high prevalence of health risk behaviors and

obesity/overweight among Hawaiian youth (68).

This study hypothesized and found that high PA, high FVC, and low SB in

childhood predict high PA, high FVC, and low SB in adolescence, which means that all

behaviors themselves are stable from childhood to adolescents. This is consistent with

other studies that have found that childhood PA and FVC are predictive of adolescent PA

and FVC (65, 132, 169, 170), and childhood SB is predictive of adolescent SB (49, 66).

Findings imply that obesity related behaviors are developed in early childhood and

continue to adolescence.

Looking at the cross behavioral prediction, our hypotheses are not largely

confirmed. Only 1 path was significant (BL FVC to T2 PA) out of 18 paths of cross

behavioral prediction. Findings reveal that SB in childhood, especially TV watching is not

related with PA. Other longitudinal studies (165, 166) documented same findings.

70

Robinson and colleagues (165) followed 279 adolescents for two years and observed

that TV watching was not associated with the change of PA levels. Neumark-Sztainer

and colleagues (166) followed 201 adolescents for 8 months and documented similar

results. Anderson and colleagues (163), and Nigg and colleagues (164) also did not find

any meaningful relationship between TV watching and PA. However, several cross-

sectional studies revealed an inverse association between times spent watching TV and

playing video games with PA (149, 167, 168).

Findings of this study also revealed that SB and FVC are not associated with each

other. Olivares and colleagues (162) found similar results between nutritional status and

television viewing. However, several longitudinal studies (159-161) found inverse

relationship between SB and FVC.

This study provides some indication that childhood FVC predicts adolescents PA

(one of two paths was significant). This is consistent with other studies examining the

relationship between FVC and PA that documented significant positive relationship

(142-150), whereas, several others found no relationship or moderate relationship (151-

158). For example, Emmonos and colleagues (150) found that PA were associated with

improved dietary behaviors, including FVC, while Dutton and colleagues (151), in their

analysis of a PA intervention's impact on dietary quality found no effect on FVC or

dietary fat intake. Wilcox and colleagues (152), however revealed significant decreases

in calories, fat, and cholesterol in result of a PA intervention, but no changes for FVC or

fiber intake.

71

This study documented inconclusive results regarding the cross behavioral

prediction from childhood to adolescents. This might be due to small sample size, and a

focus solely on these three behaviors. For example, while predicting adolescent PA, this

study only used childhood PA, FVC, and SB, and did not consider behavioral, social, and

environmental factors that might influence adolescents PA. Therefore, further

longitudinal studies are recommended to examine the predictability of childhood

obesity related behaviors to adolescent behaviors. Findings of which might be useful to

develop age appropriate interventions to fight against obesity epidemic.

The major strength of this study was the longitudinal design, which allowed us to

predict the adolescent behaviors by childhood behaviors. In addition, two different

cohorts allowed us to replicate our findings and validate our conclusions. However,

there were limitations that should be acknowledged when interpreting the results. First,

data were collected by self-report in a supervised group setting (BL and T1) and on mail-

return questionnaires (T2), which might have introduced social desirability and recall

bias. However, Baranoski and Domel have found that children are able to give accurate

dietary information and are well aware of the foods they have eaten (139). Moreover,

PA (91), FVC (128), and SB (141) measures have documented high reliability and validity.

Second, the ethnic composition of this population might limit the generalizability. Third,

like all longitudinal studies, this study experienced loss to follow-up. However, we

compared the baseline characteristics of lost to follow-up students with all students and

found no statistically significant difference in behaviors and demographics. These non

significant differences might not be enough to minimize the selection bias due to lost to

72

follow up. Finally, this study solely focused on PA, FVC, and SB; did not collect data on

some possible confounding variables, for example, socio-economic status, parents’

motivation towards specific behavior, environmental factors, etc.

The results of this study provide evidence that childhood obesity related

behaviors predict same adolescent behaviors, which means that these behaviors

develop in childhood, and continue to be stable through adolescence. Therefore,

interventions in childhood to increase PA, and FVC, and decrease SB are recommended

to fight against the obesity epidemic. This study, however, did not find cross behavioral

prediction from childhood to adolescent, which implies that further study is necessary

to study these relationships over time.

CONCLUSION

Obesity in the United States has increased over the past three decades, and has

become a public health crisis. Studies showed that healthy lifestyle, including healthy

eating, increased PA, and decreased SB, can lower the risk of becoming obese, as well as

developing obesity related disease and negative psycho-social consequences. In order to

explore the changing pattern of PA and FVC from childhood to adolescence, as well as

investigate how childhood PA, FVC, and SB behaviors influence adolescence PA, FVC,

and SB behaviors, this study utilized the data of Fun 5 study. The Fun 5 study is an

evidence-based PA and nutrition promotion program, which has been implemented in

over 160 Hawai‘i state-legislated elementary A+ after-school programs for ten years in

order to increase PA and FVC among elementary school children.

73

This dissertation is divided into three papers. In paper one, we tried to explore

the changing pattern of PA from childhood to adolescence, as well as examine the TPB

constructs to explain childhood PA. Paper one also examined the impact of childhood

and adolescent PA on adolescent obesity. Using piecewise growth mixer modeling, and

random coefficient model, paper one did not find any significant pattern of PA change

from childhood to adolescent. Structural equation modeling revealed the applicability of

TPB. Findings also documented that PBC is the strongest variable to influence childhood

PA followed by attitudes. Paper one did not find any impact of childhood and adolescent

PA on adolescent BMI.

In paper two, we tried to explore the changing pattern of FVC from childhood to

adolescence, as well as examine the TPB constructs to explain childhood FVC. Paper two

also examined the impact of childhood and adolescent FVC on adolescent obesity. Using

piecewise growth mixer modeling, and random coefficient model, paper two revealed a

declining pattern of FVC from childhood to adolescent. Structural equation modeling

revealed the applicability of TPB. Findings also documented that PBC is the strongest

variable to influence childhood FVC followed by subjective norms. Paper two did not

find any impact of childhood and adolescent FVC on adolescent BMI.

In paper three, we tried to examine the predictability of childhood PA, FVC, and

SB on adolescent PA, FVC, and SB. Structural equation modeling revealed that all three

childhood behaviors predicted same adolescent behaviors, which implies that these

behaviors develop at childhood and continue to be stable over time. Paper three,

74

however, did not find cross behavioral prediction, except BL FVC to T2 PA, in this

population.

The overall findings of this dissertation point to the following intervention

strategies to fight against childhood obesity:

• Target early childhood for intervention activities regarding increasing PA, FVC,

and decreasing SB

• Address children’s self confidence, attitudes, and subjective norms towards PA

• Address children’s self confidence, subjective norms, and attitudes towards FVC

This study also recommends:

• Further study to explore the change of obesity related behaviors from childhood

to adolescent, which should be a longitudinal study, with objectively measured

indicators. Advanced statistical technique (piecewise growth mixer modeling,

random coefficient model for example) is recommended for data analysis.

• Further study to examine the relationships of childhood and adolescent PA and

FVC with adolescent obesity. Longitudinal study with objectively measured PA,

FVC, and obesity should be adopted for such kind of study.

• Further study to explore the condition where cross behavioral relationships may

exist.

75

REFERENCES

1. Ogden CL, Carroll MD, Kit BK, et al. Prevalence of obesity and trends in body mass

index among US children and adolescents, 1999–2010. JAMA. 2012;307(5):483-

490.

2. Bao W, Srinivasan SR, Wattigney WA, et al. Persistence of multiple cardiovascular

risk clustering related to syndrome X from childhood to young adulthood: the

Bogalusa Heart Study. Arch Intern Med. 1994;154(16):1842-1847.

3. Wattigney WA, Webber LS, Srinivasan SR, et al. The emergence of clinically

abnormal levels of cardiovascular disease risk factor variables among young adults:

the Bogalusa Heart Study. Prev Med. 1995;24(6):617-626.

4. Freedman DS, Dietz WH, Srinivasan SR, et al. The relation of overweight to

cardiovascular risk factors among children and adolescents: the Bogalusa Heart

Study. Pediatrics. 1999;103(6 Pt 1):1175-1182.

5. Narayan KM, Boyle JP, Thompson TJ, et al. Lifetime risk for diabetes mellitus in the

United States. JAMA. 2003;290(14):1884-1890.

6. Freedman DS, Mei Z, Srinivasan SR, et al. Cardiovascular risk factors and excess

adiposity among overweight children and adolescents: the Bogalusa Heart Study. J

Pediatr. 2007;150(1):12-17.e2.

7. Gortmaker SL, Must A, Perrin JM, et al. Social and economic consequences of

overweight in adolescence and young adulthood. N Engl J Med.

1993;329(14):1008-1012.

8. Griffiths LJ, Parsons TJ, Hill AJ. Self-esteem and quality of life in obese children and

adolescents: a systematic review. Int J Pediatr Obes. 2010;5(4):282-304.

9. Stunkard A, Mendelson M. Obesity and the body image: characteristics of

disturbances in the body image of some obese persons. Am J Psychiatry.

1967;123(10):1296-1300.

10. Wallace WJ, Sheslow D, Hassink S. Obesity in children: a risk for depression. Ann N

Y Acad Sci. 1993;699(1):301-303.

76

11. Wang YC, Beydoun MA, Liang L, et al. Will all Americans become overweight or

obese? Estimating the progression and cost of the US obesity epidemic. Obesity.

2008;16(10):2323-2330.

12. Veltsista A, Kanaka C, Gika A, et al. Tracking of overweight and obesity in Greek

youth. Obes Facts. 2010;3(3):166-172.

13. Whitaker RC, Wright JA, Pepe MS, et al. Predicting obesity in young adulthood

from childhood and parental obesity. N Engl J Med. 1997;337(13):869-873.

14. Freedman DS, Khan LK, Serdula MK, et al. Inter-relationships among childhood BMI,

childhood height, and adult obesity: the Bogalusa Heart Study. Int J Obes Relat

Metab Disord. 2004;28(1):10-16.

15. Guo SS, Chumlea WC. Tracking of body mass index in children in relation to

overweight in adulthood. Am J Clin Nutr 1999;70(1part2):145S-148S.

16. Obarzanek E, Schreiber GB, Crawford PB, et al. Energy intake and physical activity

in relation to indexes of body fat: the National Heart, Lung and Blood Institute

Growth and Health Study. Am J Clin Nutr 1994;60(1):15-22.

17. Penedo FJ, Dahn JR. Exercise and well-being: a review of mental and physical

health benefits associated with physical activity. Curr Opin Psychiatry

2005;18(2):189-93.

18. Warburton DE, Nicol CW, Bredin SS. Health benefits of physical activity: the

evidence. CMAJ 2006;174(6):801-809.

19. Hallal PC, Victora CG, Azevedo MR, et al. Adolescent physical activity and health: a

systematic review. Sports Med 2006;36(12):1019-1030.

20. Wang Y, Ge K, Popkin BM. Why do some overweight children remain overweight,

whereas others do not? Public Health Nutr 2003;6(6):549-558.

21. Dumith SC, Gigante DP, Domingues MR, et al. Physical activity change during

adolescence: a systematic review and a pooled analysis. Int J Epidemiol

2011;40(3):685-698.

http://www.ncbi.nlm.nih.gov.eres.library.manoa.hawaii.edu/pubmed/21872750

77

22. Nadar PR, Bradley RH, Houts RM, et al. Moderate-to-vigorous physical activity

from ages 9 to 15 years. JAMA. 2008;300(3):295-305.

23. Pate RR, Stevens J, Webber LS, et al. Age-related change in physical activity in

adolescent girls. J Adol Health. 2009;44(3):275-282.

24. Gebremariam MK, Totland TH, Andersen LF, et al. Stability and change in screen-

based sedentary behaviours and associated factors among Norwegian children in

the transition between childhood and adolescence. BMC Public Health.

2012;12:104.

25. Viner RM, Cole TJ. Television viewing in early childhood predicts adult body mass

index. J Pediatr. 2005;147(4):429-435.

26. Mikkila V, Rasanen L, Raitakari OT, et al. Consistent dietary patterns identified

from childhood to adulthood: the cardiovascular risk in Young Finns Study. Br J

Nutr. 2005;93(6):923-931.

27. Patterson E, Warnberg J, Kearney J, Sjostrom M. The tracking of dietary intakes of

children and adolescents in Sweden over six years: the European Youth Heart

Study. Int J Behav Nutr Phys Act. 2009;6:91.

28. Demory-Luce D, Morales M, Nicklas T, et al. Changes in food group consumption

patterns from childhood to young adulthood: the Bogalusa Heart Study. J Am Diet

Assoc. 2004;104(11):1684-1691.

29. Sanchez A, Norman GJ, Sallis JF, et al. Patterns and correlates of physical activity

and nutrition behaviors in adolescents. Am J Prev Med. 2007;32(2):124-130.

30. Driskell MM, Dyment S, Mauriello L, et al. Relationships among multiple behaviors

for childhood and adolescent obesity prevention. Prev Med. 2008;46(3):209-215.

31. Hawaii Department of Health. Hawaii Behavioral Risk Factor Surveillance Survey

2009. www.hawaii.gov/health/statistics/ [Accessed June 1, 2013]

32. Mau MK, Sinclair K, Saito EP, et al. Cardiometabolic health disparities in native

Hawaiians and other Pacific Islanders. Epidemiol Rev. 2009;31:113-129.

http://www.hawaii.gov/health/statistics/

78

33. Baruffi G, Hardy C, Waslien C et al. Ethnic differences in the prevalence of

overweight among young children in Hawaii. J Am Diet Assoc. 2004;104(11):1701-

1707.

34. Novotny R, Oshiro CE, Wilkens LR. Prevalence of childhood obesity among young

multiethnic children from a health maintenance organization in Hawaii. Child ObesI.

2013;9(1):35-42.

35. Mitchell JA, Pate RR, España-Romero V, et al. Moderate-To-vigorous physical

activity is associated with decreases in body mass index from ages 9 to 15 years.

Obesity (Silver Spring). 2013;21(3):E280-93.

36. Riddoch CJ, Leary SD, Ness AR, et al. Prospective associations between objective

measures of physical activity and fat mass in 12-14 year old children: the Avon

Longitudinal Study of Parents and Children (ALSPAC). BMJ. 2009;339:b4544.

37. United States Department of Health and Human Services. Physical Activity

Guidelines Advisory Committee Report 2008. Washington, DC: United States

Department of Health and Human Services, 2008.

38. Biddle SJH, Gorely T, Stensel DJ. Health-enhancing physical activity and sedentary

behaviour in children and adolescents. J Sports Sci. 2004;22(8):679-701.

39. Pate RR, Freedson PS, Sallis JF, et al. Compliance with physical activity guidelines:

prevalence in a population of children and youth. Ann Epidemiol. 2002;12(5):303-

308.

40. Beighle, A. Increasing physical activity through recess. Retrieved January, 2012,

from http://www.eric.ed.gov/PDFS/ED541180.pdf

41. Hancox RJ, Milne BJ, Poulton R. Association between child and adolescent

television viewing and adult health: a longitudinal birth cohort study. Lancet.

2004;364:257-262.

42. Robinson TN. Television viewing and childhood obesity. Pediatr Clin North Am.

2001;48(4):1017-1025.

http://www.eric.ed.gov/PDFS/ED541180.pdf

79

43. Marshall SJ, Biddle SJ, Gorely T, et al. Relationships between media use, body

fatness and physical activity in children and youth: a meta-analysis. Int J Obes Relat

Metab Disord. 2004;28(10):1238-1246.

44. Proctor MH, Moore LL, Gao D, et al. Television viewing and change in body fat

from preschool to early adolescence: the Framingham Children’s Study. Int J Obes

Relat Metab Disord. 2003;27(7):827-833.

45. Dietary Guidelines Advisory Committee. Report of the Dietary Guidelines Advisory

Committee on the Dietary Guidelines for Americans, 2010, to the Secretary of

Agriculture and the Secretary of Health and Human Services. Washington, DC: U.S.

Department of Agriculture; 2010.

46. U.S. Department of Agriculture, U.S. Department of Health and Human Services.

Dietary Guidelines for Americans, 2010. 7th Edition. Washington, DC, US

Government Printing Office; 2010.

47. Briefel RR, Johnson CL. Secular trends in dietary intake in the United States. Annu

Rev Nutr. 2004;24:401-431.

48. Forshee RA, Anderson PA, Storey ML. Changes in calcium intake and association

with beverage consumption and demographics: comparing data from CSFII 1994-

1996, 1998 and NHANES 1999–2002. J Am Coll Nutr. 2006;25(2):108-116.

49. Pearson N, Biddle SJH. Sedentary behavior and dietary intake in children,

adolescents, and adults: a systematic review. Am J Prev Med. 2011;41(2):178-188.

50. Taveras EM, Field AE, Berkey CS, et al. Longitudinal relationship between television

viewing and leisure-time physical activity during adolescence. Pediatrics.

2007;119(2):e314-e319.

51. Lippke S, Nigg CR, Maddock JE. Health-promoting and health-risk behaviors:

theory-driven analyses of multiple health behavior change in three international

samples. Int J Behav Med. 2012;19(1):1-13.

52. Nigg CR, Allegrante JP, Ory M. Theory comparison and multiple-behavior research:

common themes advancing health behavior research. Health Edu Res.

2002;17(5):670-679.

80

53. Trost SG, Pate RR, Sallis JF, et al. Age and gender differences in objectively

measured physical activity in youth. Med Sci Sports Exerc. 2002;34(2):350-355.

54. Ridgers ND, Stratton G, Fairclough SJ. Physical activity levels of children during

school playtime. Sports Med. 2006;36(4):359-371.

55. Stratton G, Ridgers ND, Fairclough SJ, et al. Physical activity levels of normal-

weight and overweight girls and boys during primary school recess. Obesity (Silver

Spring). 2007;15(6):1513-1519.

56. Belcher BR, Berrigan D, Dodd KW, et al. Physical activity in US youth: effect of

race/ethnicity, age, gender, and weight status. Med Sci Sports Exerc.

2010;42(12):2211-2221.

57. Nader PR, O’Brien M, Houts R, et al. Identifying risk for obesity in early childhood.

Pediatrics 2006;118(3):e594-601.

58. Must A, Strauss RS. Risks and consequences of childhood and adolescent obesity.

Int J Obes Realt Metab Disord. 1999;23(S2):S2-11.

59. van der Horst K, Paw MJ, Twisk JW, et al. A brief review on correlates of physical

activity and sedentariness in youth. Med Sci Sports Exerc. 2007;39(8):1241-1250.

60. Telama R. Tracking of physical activity from childhood to adulthood: a review.

Obes Facts. 2009;2(3):187-195.

61. Janz KF, Dawson JD, Mahoney LT. Tracking of fitness and activity during puberty:

the Muscatine Study. Med Sci Sports Exerc. 2000;32(7):1250-1257.

62. Malina RM. Tracking of physical activity and physical fitness across the lifespan.

Res Q Exerc Sport. 1996;67(3suppl):48-57.

63. Blair SN, Kohl HW, Barlow CE, et al. Changes in physical fitness and all-cause

mortality. JAMA. 1995;273(14):1093-1098.

64. Blair SN, Kohl HW, Paffenbarger RS, et al. Physical fitness and all-cause mortality: a

prospective study of healthy men and women. JAMA. 1989;262(17):2395-2401.

65. Telama R, Yang X, Viikari J, et al. Physical activity from childhood to adulthood: a

21-year tracking study. Am J Prev Med. 2005;28(3):267-273.

81

66. Janz KF, Burns TL, Levy SM; Iowa Bone Development Study. Tracking of activity and

sedentary behaviors in childhood: the Iowa Bone Development Study. Am J Prev

Med. 2005;29(3):171-178.

67. Gesell SB, Reynolds EB, Ip EH, et al. Social influences on self-reported physical

activity in overweight latino children. Clin Pediatr. 2008;47(8):797-802.

68. Kolodziekski B, Hirokawa R, Benson K, et al. Hawaii physical activity and nutrition

surveillance report 2008. Hawaii: Hawaii State Department of Health, 2008.

69. Jago R, Baranowski T. Non-curricular approaches for increasing physical activity in

youth: a review. Prev Med. 2004;39(1):157-163.

70. Bartholomew LK, Parcel GS, Kok G, et al. Planning Health Promotion Programs: An

Intervention Mapping Approach. John Wiley Sons, San Francisco, Calif, USA, 2006.

71. Baranowski T, Anderson C, Carmack C. Mediating variable framework in physical

activity interventions: how are we doing? How might we do better? Am J Prev Med.

1998;15(4):266-297.

72. Norman P, Conner M. Predicting and changing health behaviour: future directions.

in Predicting Health Behaviour. Conner, M, Norman, P. (Eds). Open University Press,

Beckshire, UK, 2005.

73. Tsorbatzoudis H. Evaluation of a school-based intervention programme to

promote physical activity: an application of the theory of planned behavior.

Percept Mot Skills. 2005;101(3):787-802.

74. Huhman M, Heitzler C, Wong F. The VERB campaign logic model: a tool for

planning and evaluation. Prev Chronic Dis. 2004;1(3): A11.

75. Painter JE, Borba CP, Hynes M, et al. The use of theory in health behavior research

from 2000 to 2005: a systematic review. Ann Behav Med. 2008;35(3):358-62.

76. Dishman RK, Motl RW, Saunders R, et al. Self-efficacy partially mediates the effect

of a school-based physical-activity intervention among adolescent girls. Prev Med.

2004;38(5):628-636.

77. Armitage CJ, Conner M. Efficacy of the theory of planned behavior: a meta-analytic

review. Br J Soc Psychol. 2001;40(4):471-499.

82

78. Godin G, Kok G. The theory of planned behavior: a review of its applications to

health related behaviors. Am J Health Prom. 1996;11(2):87-98.

79. Dzewaltowski DA, Noble JM, Shaw JM. Physical activity participation: social

cognitive theory versus the theories of reasoned action and planned behavior. J

Sport Exerc Psychol. 1990;12: 388-405.

80. Hagger MS, Chatzisarantis NLD, Biddle SJH. A meta-analytic review of the theories

of reasoned action and planned behavior in physical activity: predictive validity

and the contribution of additional variables. J Sport Exerc Psychol. 2002;24:3-32.

81. Nigg CR, Lippke S, Maddock JE. Factorial invariance of the theory of planned

behavior applied to physical activity across gender, age, and ethnic groups. Psychol

Sport Excr. 2009;10:219-225.

82. Plotnikoff RC, Costigan SA, Karunamuni N, et al. Social cognitive theories used to

explain physical activity behavior in adolescents: a systematic review and meta-

analysis. Prev Med. 2013;56(5):245-53.

83. Hausenblas HA, Carron AV, Mack DE. Application of the theories of reasoned

action and planned behavior to exercise behavior: a meta-analysis. J Sport Exerc

Psychol. 1997;19(1):36-51.

84. Ajzen I. The theory of planned behavior. Organizational Behavior and Human

Decision Processes 1991;50:179-211.

85. Martin JJ, McCaughtry N, Hodges-Kulinna P, et al. Predicting physical activity and

cardiorespiratory fitness in African American children. J Sport Exerc Psychol.

2005;27:456-469.

86. Nigg CR, Geller K, Adams P, et al. Successful dissemination of Fun 5 – a physical

activity and nutrition program for children. Transl Behav Med. 2012;2(3):276-285.

87. Battista J, Nigg CR, Chang JA, et al. Elementary after school programs: an

opportunity to promote physical activity for children. Californian J Health Promot.

2005;3:108-118.

88. Prevention CDC. Growth Chart Training: A SAS Program for CDC growth charts.

CDC . 2009.

83

89. Godin G, Jobin J, Bouillon J. Assessment of leisure time exercise behavior by self-

report: a concurrent validity study. Can J Public Health. 1986;77(5):359-362.

90. Godin G, Shephard RJ. A simple method to assess exercise behavior in the

community. Can J Appl Sport Sci. 1985;10(3):141-146.

91. Sallis JF. Epidemiology of physical activity and fitness in children and adolescents.

Crit Rev Food Sci Nutr. 1993;33(4-5):403-408.

92. Colder CR, Mehta P, Balanda K, et al. Identifying trajectories of adolescent

smoking: an application of latent growth mixture modeling. Health Psychol.

2001;20(2):127-35.

93. Seltzer MH, Frank KA, Bryk AS. The metric matters-the sensitivity of conclusions

about growth in student-achievement to choice of metric. Educ Eval Policy Analysis.

1994;16:41-49.

94. Bartholow BD, Sher KJ, Krull JL. Changes in heavy drinking over the third decade of

life as a function of collegiate fraternity and sorority involvement: a prospective,

multilevel analysis. Health Psychol. 2003;22(6):616-626.

95. Li F, Duncan TE, Hops H. Examining developmental trajectories in adolescent

alcohol use using piecewise growth mixture modeling analysis. J Stud Alcohol.

2001;62(2):199-210.

96. Hu L, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis:

conventional criteria versus new alternatives. Structural Equation Modeling.

1999;6(1):1-55.

97. Bentler, PM. Comparative fit indexes in structural models. Psychological Bulletin.

1990;107(2):238-246.

98. Taylor RW, Murdoch L, Carter P, et al. Longitudinal study of physical activity and

inactivity in preschoolers: the FLAME study. Med Sci Sports Exerc. 2009;41(1):96-

102.

99. Pate RR, Baranowski T, Dowda M, et al. Tracking of physical activity in young

children. Med Sci Sports Exerc. 1996;28(1):92-96.

84

100. Wilkin TJ, Mallam KM, Metcalf BS, et al. Variation in physical activity lies with the

child, not his environment: evidence for an ‘activitystat’ in young children

(EarlyBird 16). Int J Obes (Lond). 2006;30(7):1050-1055.

101. Telama R, Leskinen E, Yang X. Stability of habitual physical activity and sport

participation: a longitudinal tracking study. Scand J Med Sci Sports. 1996;6(6):371-

378.

102. Aarnio M, Winter T, Peltonen J, et al. Stability of leisure-time physical activity

during adolescence - a longitudinal study among 16-, 17-and 18-year-old Finnish

youth. Scand J Med Sci Sports. 2002;12(3):179-185.

103. Aaron DJ, Kriska AM, Dearwater SR, et al. The epidemiology of leisure physical

activity in an adolescent population. Med Sci Sports Exerc. 1993;25(7):847-853.

104. Sunnegardt J, Bratteby LE, Sjolin S. Physical activity and sports involvement in 8-

and 13-year-old children in Sweden. Acta Paediatr Scand. 1985;74(6):904912.

105. Kimm SY, Glynn NW, Kriska AM, et al. Longitudinal changes in physical activity in a

biracial cohort during adolescence. Med Sci Sports Exerc. 2000;32(8):1445-1454.

106. Ross JG, Gilbert GG. The National Children and Youth Fitness Study: a summary of

findings. J Phys Educ Rec Dance. 1985;56(1):45-50.

107. Jackson DM, Reilly JJ, Kelly LA, et al. Objectively measured physical activity in a

representative sample of 3- to 4-yearold children. Obes Res. 2003;11(3):420-425.

108. Kelly LA, Reilly JJ, Jackson DM, et al. Tracking physical activity and sedentary

behavior in young children. Pediatr Exerc Sci. 2007;19(1):51-60.

109. Plotnikoff RC, Lubans DR, Costigan SA, et al. A test of the theory of planned

behavior to predict physical activity in an overweight/obese population sample of

adolescents from Alberta, Canada. Health Educ Behav. 2013;40(4):415-25.

110. Araujo-Soares V, McIntyre T, Sniehotta FF. Predicting changes in physical activity

among adolescents: the role of self-efficacy, intention, action planning and coping

planning. Health Edu Res. 2009;24(1):128-139.

111. Craig S, Goldberg J, Dietz WH. Psychosocial correlates of physical activity among

fifth and eighth graders. Prev Med. 1996;25(5):506-513.

85

112. Hagger MS, Chatzisarantis N, Biddle SJH, et al. Antecedents of children’s physical

activity intention and behavior: Predictive validity and longitudinal effects. Psychol

Health. 2001;16(4):391-407.

113. Martin JJ, Oliver K, McCaughtry N. The theory of planned behavior: Predicting

physical activity in Mexican American children. J Sport Exerc Psychol.

2007;29(2):225-238.

114. Mummery WK, Spence JC, Hudec JC. Understanding physical activity intention in

Canadian school children and youth: An application of the theory of planned

behavior. Res Q Exerc Sport. 2000;71(2):116-124.

115. Trost SG, Saunders R, Ward DS. Determinants of physical activity in middle school

children. Am J Health Behav. 2002;26(2):95-102.

116. Boudreau F, Godin G. Using the theory of planned behavior to predict exercise

intention in obese adults. Can J Nurs Res. 2007;39(2):112-125.

117. Blanchard CM, McGannon KR, Spence JC, et al. Social ecological correlates of

physical activity in normal weight, overweight, and obese individuals. Int J Obes

(Lond). 2005;29(6):720-726.

118. Dietz WH. Overweight in childhood and adolescence. N Engl J Med. 2004;350:855-

857.

119. Franks PW, Hanson RL, Knowler WC, et al. Childhood obesity, other cardiovascular

risk factors, and premature death. N Engl J Med. 2010;362(19):485-493.

120. Singh AS, Mulder C, Twisk JWR, et al. Tracking of childhood overweight into

adulthood: a systematic review of the literature. Obes Rev. 2008;9:474-488.

121. Kushi LH, Doyle C, McCullough M, et al. American Cancer Society guidelines on

nutrition and physical activity for cancer prevention: reducing the risk of cancer

with healthy food choices and physical activity. CA Cancer J Clinicians.

2012;62(1):30-67.

122. National Center for Chronic Disease Prevention and Health Promotion, Division of

Nutrition, Physical Activity, and Obesity. State indicator report on fruits and

vegetables 2013. 2013

86

123. te Velde SJ, Twisk JW, Brug J. Tracking of fruit and vegetable consumption from

adolescence into adulthood and its longitudinal association with overweight. Br J

Nutr. 2007;98(2):431-8.

124. Kelder SH, Perry CL, Klepp KI, et al. Longitudinal tracking of adolescent smoking,

physical activity and food choice behaviors. Am J Public Health. 1994;84(7):1121-

1126.

125. Hewitt AM, Stephens C. Healthy eating among 10 - 13-year-old New Zealand

children: understanding choice using the Theory of Planned Behavior and the role

of parental influence. Psychol Health Med. 2007;12(5):526-535.

126. Bazillier C, Verlhiac JF, Mallet P, et al. Predictors of intentions to eat healthily in 8-

9-year-old children. J Cancer Educ. 2011;26(3):572-576.

127. Branscum P, Sharma M. Using the theory of planned behavior to predict two types

of snack food consumption among Midwestern upper elementary children:

implications for practice. Int Q Community Health Educ. 2011-2012;32(1):41-55.

128. Prochaska J, Sallis J, Rupp J, et al. Brief self reported measures for assessing fat,

fruit and vegetable and calcium intake with adolescents [abstract]. Ann Behav Med.

2000;22:S132.

129. Resnicow K, Smith M, Baranowski T, et al. 2-year tracking of children's fruit and

vegetable intake. J Am Diet Assoc. 1998;98(7):785-789.

130. Patterson E, Wärnberg J, Kearney J, et al. The tracking of dietary intakes of children

and adolescents in Sweden over six years: the European Youth Heart Study. Int J

Behav Nutr Phys Act. 2009;6:91.

131. Nelson MC, Neumark-Sztainer D, Hannan PJ, et al. Five-year longitudinal and

secular shifts in adolescent beverage intake: findings from project EAT (Eating

Among Teens)-II. J Am Diet Assoc. 2009;109(2):308-312.

132. Wang Y, Bentley ME, Zhai F, et al. Tracking of dietary intake patterns of Chinese

from childhood to adolescence over a six-year follow-up period. J Nutr.

2002;132(3):430-438.

87

133. Lake AA, Mathers JC, Rugg-Gunn AJ, et al. Longitudinal change in food habits

between adolescence (11-12 years) and adulthood (32-33 years): the ASH30 Study.

J Public Health (Oxf). 2006;28(1):10-16.

134. Oellingrath IM, Svendsen MV, Brantsaeter AL. Tracking of eating patterns and

overweight - a follow-up study of Norwegian schoolchildren from middle

childhood to early adolescence. Nutr J. 2011;10:106.

135. Nelson MC, Neumark-Sztainer D, Hannan PJ, et al. Five-year longitudinal and

secular shifts in adolescent beverage intake: findings from project EAT (Eating

Among Teens)-II. J Am Diet Assoc. 2009;109(2):308-312.

136. Kvaavik E, Andersen LF, Klepp KI. The stability of soft drinks intake from

adolescence to adult age and the association between long-term consumption of

soft drinks and lifestyle factors and body weight. Public Health Nutr.

2005;8(2):149-157.

137. Lautenschlager L, Smith C. Understanding gardening and dietary habits among

youth garden program participants using the Theory of Planned Behavior. Appetite.

2007;49(1):122-30.

138. Berkey CS, Rockett HR, Field AE, et al. Activity, dietary intake, and weight changes

in a longitudinal study of preadolescent and adolescent boys and girls. Pediatrics.

2000;105(4):E56.

139. Dixey R, Sahota P, Serbjit A, et al. Children talking about healthy eating: data from

focus groups with 300 9 - 11-year-olds. Nutrition Bulletin. 2001;26:71-79.

140. Baranowski T, Domel S. A cognitive model of child’s reporting of food intake. Am J

Clin Nutr. 1994;59:2125-2175.

141. Office of the Surgeon General. The Surgeon General's Vision for a Healthy and Fit

Nation. Rockville, MD, U.S. Department of Health and Human Services; 2010.

Available at:

http://www.surgeongeneral.gov/library/obesityvision/obesityvision2010.pdf

http://www.surgeongeneral.gov/library/obesityvision/obesityvision2010.pdf

88

142. Wallace LS, Buckworth J, Kirby TE, et al. Characteristics of exercise behavior among

college students: application of social cognitive theory to predicting stage of

change. Prev Med. 2000;31(5):494-505.

143. Coulson NS, Eiser C, Eiser JR. Diet, smoking and exercise: interrelationships

between adolescent health behaviours. Child Care Health Dev. 1997;23(3):207-216.

144. Emmons KM, McBride CM, Puleo E, et al. Prevalence and predictors of multiple

behavioral risk factors for colon cancer. Prev. Med. 2005;40(4):527-534.

145. Matthews CE, Hebert JR, Ockene IS, et al. Relationship between leisure-time

physical activity and selected dietary variables in the Worcester Area Trial for

Counseling in Hyperlipidemia. Med. Sci. Sports Exerc. 1997;29(9):1199-1207.

146. Eaton CB, McPhillips JB, Gans KM, et al. Cross-sectional relationship between diet

and physical activity in two southeastern New England communities. Am J Prev

Med. 1995;11(4):238-244.

147. Emmons K, Marcus B, Shadel WL, et al. Physical activity: a gateway to improve

dietary behaviors? Ann Behav Med. 1998;20:S069 (abstract).

148. Simoes EJ, Byers T, Coates RJ, et al. The association between leisure-time physical

activity and dietary fat in American adults. Am J Public Health. 1995;85(2):240-244.

149. Pate RR, Heath GW, Dowda M, et al. Associations between physical activity and

other health behaviors in a representative sample of US adolescents. Am J Public

Health. 1996;86(11):1577-1581.

150. Emmons K, Shadel W, Linnan L, et al. A prospective analysis of change in multiple

risk factors for cancer. Cancer Res Ther Control. 1999:8:15-23.

151. Dutton GR, Napolitano MA, Whiteley JA, et al. Is physical activity a gateway

behavior for diet? Findings from a physical activity trial. Prev Med. 2008;46(3):216-

221.

152. Wilcox S, King AC, Castro C, et al. Do changes in physical activity lead to dietary

changes in middle and old age? Am J PrevMed. 2000;18(4):276-283.

153. Zunich EM, Dickinson A. A comparison of physical fitness behaviors of joggers and

nonjoggers. Health Values. 1979;3:296-299.

89

154. Hartung GH, Foreyt JP, Mitchell RE, et al. Relation of diet to high-density-

lipoprotein cholesterol in middle-aged marathon runners, joggers, and inactive

men. N Engl J Med. 1980;302(7):357-361.

155. Leigh JP, Fries JF. Associations among healthy habits, age, gender, and education in

a sample of retirees. Int J Aging Hum Dev. 1992-1993;36(2):139-155.

156. Voorrips LE, van Staveren WA, Hautvast JGAJ. Are physically active elderly women

in a better nutritional condition than their sedentary peers? Eur J Clin Nutr.

1991;45(11):545-552.

157. Blair SN, Goodyear NN, Wynne KL, et al. Comparison of dietary and smoking habit

changes in physical fitness improvers and nonimprovers. Prev Med.

1984;13(4):411-420.

158. Expert Panel on the Identification, Evaluation, and Treatment of Overweight and

Obesity in Adults. Executive summary of the clinical guidelines on the identification,

evaluation, and treatment of overweight and obesity in adults. Arch Intern Med.

1998;158(17):1855-1867.

159. Boynton-Jarrett R, Thomas TN, Peterson KE, et al. Impact of television viewing

patterns on fruit and vegetable consumption among adolescents. Pediatrics.

2003;112(6 Pt 1):1321-1326.

160. Larson NI, Neumark-Sztainer D, Harnack L, et al. Calcium and dairy intake:

longitudinal trends during the transition to young adulthood and correlates of

calcium intake. J Nutr Educ Behav. 2009;41(4):254-260.

161. Larson NI, Neumark-Sztainer D, Story MT, et al. Fast food intake: longitudinal

trends during the transition to young adulthood and correlates of intake. J Adolesc

Health. 2008;43(1):79-86.

162. Olivares S, Kain J, Lera L, et al. Nutritional status, food consumption and physical

activity among chilean school children: a descriptive study. Eur J Clin Nutr.

2004;58(9):1278-1285.

163. Anderson RE, Crespo C, Bartlett SJ, et al. Relationship of physical activity and

television watching with body weight and level of fatness among children: results

90

from the third national health and nutrition examination survey. JAMA.

1998;279(12):938-942.

164. Nigg CR, Allegrante JP, Ory M. Theory comparison and multiple-behavior research:

common themes advancing health behavior research. Health Edu Res. 2002;17(5):

670-679.

165. Robinson TN, Hammer LD, Killen JD, et al. Does television viewing increase obesity

and reduce physical activity? Cross-sectional and longitudinal analyses among girls.

Pediatrics. 1993;91(2):273-280.

166. Neumark-Sztainer D, Story M, Hannan PJ, et al. Factors associated with changes in

physical activity. Arch Pediatr Adolesc Med. 2003;157(8):803-810.

167. Strauss RS, Rodzilsky D, Burack G, et al. Psychosocial correlates of physical activity

in health children. Arch Pediatr Adolesc Med. 2001;155(8):897-902.

168. Lowry R, Wechsler H, Galuska DA, et al. Television viewing and its association with

overweight, sedentary lifestyle, and insufficient consumption of fruits and

vegetables among US high school students: Differences by race, ethnicity, and

gender. J Sch Health. 2002;72(10):413-421.

169. Friedman HS, Martin LR, Tucker JS, et al. Stability of physical activity across the

lifespan. J Health Psychol. 2008;13(8):1092-1104.

170. Cutler GJ, Flood A, Hannan P, et al. Major patterns of dietary intake in adolescents

and their stability over time. J Nutr. 2009;139(2):323-328.

tracking obesity related behaviors from childhood …

Documents