copyright 2014, janani r. thapa
TRANSCRIPT
ESSAYS ON
COMMUNITY-BASED BEHAVIOR CHANGE INTERVENTIONS
FOR OBESITY PREVENTION
by
Janani R. Thapa, M.S., M.P.A.
A Dissertation
In
Agricultural and Applied Economics
Submitted to the Graduate Faculty
of Texas Tech University in
Partial Fulfillment of
the Requirements for
the Degree of
DOCTOR OF PHILOSOPHY
Approved
Dr. Conrad Lyford
Chair of Committee
Dr. Jaime Malaga
Dr. Barbara Pence
Dr. Eduardo Segarra
Dr. Ryan Williams
Mark Sheridian
Dean of the Graduate School
August 2014
Copyright 2014, Janani R. Thapa
Texas Tech University, Janani R. Thapa, August 2014
ii
ACKNOWLEDGEMENTS
Earning a Ph.D. degree and serving in academia is a dream come true which
wouldn’t have been possible without the support, patience, and guidance of the following
people. It is to them that I owe my deepest gratitude.
To Dr. Conrad Lyford, my major advisor, for his endless encouragements,
invaluable and uncountable suggestions, constant guidance, and crucial understanding
throughout my Ph.D. study. I highly acknowledge his wisdom, knowledge and
commitment that always inspired and motivated me. I am indebted and will forever
continue to be his advisee. To Dr. Eduardo Segarra for his unending confidence and
support since before I started my degree to this day. To my committee members: Dr.
Jaime Malaga for his countless encouragements, Dr. Barbara Pence for her invaluable
time, and Dr. Ryan Williams for his helpful comments. This dissertation took its current
shape due to the detailed attention and feedback I received from my major advisor and
my committee members. Also, to Dr. Eric Belasco, Tyra Carter, Dr. Conrad Lyford, Dr.
Barent McCool, Dr. Audrey McCool, and Dr. Barbara Pence for their work in bringing
the research grant from the Cancer Prevention Research Institute of Texas.
To my support system: guardian Drs. Devendra and Ratna Khatri, siblings (Sujata
and Siddhartha), father- and mother-in law, very close ones (Beth-Thys, Roji-Kaushal,
Sameer-Simran, Resham-Durga, and Chandra-Sirjana), colleagues (Drs. Adhikari, Nair
and Tiwari), friends (Michael, Sanja, Haiyan, Abbas, Abbes, Dacheng, Kishor, Mira,
Dipti, Bishnu, Shivani, Nilam, Erina, Shailee), office staff (Cindy, DeeAnn and Neal),
advisors (Drs. Nauriddine Abidi, Durga D. Dhakal, Durga M. Gautam, Kathryn March,
Megha Parajulee, and Norman Uphoff), and above all I acknowledge all the faculty of the
Department of Agricultural and Applied Economics at Texas Tech University.
To Rajeev Rajbhandari, my husband, for making finishing so important and for
making every tough times easy and to Rishav, my son, for spending countless hours
without ‘mamu.’ Without them this effort would have worth nothing.
Finally, I dedicate this dissertation to Ms. Subhadra Kumari, my mother. It is only
due to her devotion and perseverance that I have been able to reach this far.
Unfortunately I fell short by few months and lost her just before I could tell her I have
attained what she had set me forth for. This is for her.
Texas Tech University, Janani R. Thapa, August 2014
iii
TABLE OF CONTENTS
ACKNOWLEDGEMENTS ______________________________________________ ii
ABSTRACT ___________________________________________________________ ix
LIST OF TABLES ______________________________________________________x
LIST OF FIGURES ___________________________________________________ xii
CHAPTER I: BEHAVIORAL INTERVENTIONS TO ADDRESS OBESITY AS A
MARKET FAILURE ____________________________________________________1
Abstract _______________________________________________________________1
1.1 Background _________________________________________________________2
1.2 Obesity: a market failure ______________________________________________4
1.2.1 Negative externality in production____________________________________5
1.2.2 Negative externality in consumption __________________________________8
1.3 Addressing the market failure __________________________________________9
1.3.1 Taxation ________________________________________________________9
1.3.2 Behavioral interventions __________________________________________10
1.4 Addressing obesity as an externality ____________________________________12
1.5 Problem statement and research objective _______________________________16
1.5.1 Essay I ________________________________________________________18
1.5.2 Essay II________________________________________________________18
1.5.3 Essay III _______________________________________________________19
Texas Tech University, Janani R. Thapa, August 2014
iv
CHAPTER II: THE IMPACT OF A MULTI-TIERED COMMUNITY-BASED
OBESITY AND CANCER PREVENTION PROGRAM IN A SMALL
COMMUNITY ________________________________________________________20
Abstract ______________________________________________________________20
2.1 Introduction ________________________________________________________21
2.2 The project concept and model ________________________________________23
2.3 Review of literature__________________________________________________24
2.3.1 Obesity and obesity in the US ______________________________________24
2.3.1.1 Obesity and overweight _____________________________________24
2.3.1.2 Obesity in the United States __________________________________25
2.3.2 Obesity and cancer _______________________________________________27
2.3.2.1 Evidence of obesity and cancer _______________________________27
2.3.2.2 Epidemiology _____________________________________________28
2.3.3 Behavior _______________________________________________________28
2.3.3.1 Behavior and cancer ________________________________________28
2.3.3.2 Food behavior ____________________________________________29
2.3.3.3 Theories of behavior change _________________________________30
2.3.4 Community-based participatory research (CBPR) ______________________31
2.3.4.1 Community interventions measuring BMI as the outcome variable ___32
2.4 Conceptual framework _______________________________________________33
2.4.1 The mediating variable model ______________________________________35
Texas Tech University, Janani R. Thapa, August 2014
v
2.5 Methods ___________________________________________________________36
2.5.1 Data __________________________________________________________36
2.5.1.1 Anthropometric measurements _______________________________39
2.5.1.2 Health assessment survey ___________________________________40
2.5.1.3 Project participation survey __________________________________42
2.5.1.4 Missing data ______________________________________________42
2.5.2 Study designs ___________________________________________________43
2.5.3 Data analysis ___________________________________________________45
2.5.3.1 Project impact and sustainability through group comparisons _______45
2.5.3.2 Simple intervention effect on cancer knowledge score _____________47
2.5.3.3 Effect of intervention on BMI ________________________________49
2.6 Results ____________________________________________________________52
2.6.1 Project impact and sustainability through group comparisons _____________52
2.6.1.1 Cancer risk awareness ______________________________________52
2.6.1.2 Cancer risk behavior including nutrition awareness _______________55
2.6.1.3 Food behavior and other cancer risk behavior ____________________56
2.6.2 Simple intervention effect on cancer knowledge score ___________________59
2.6.3 Effect of intervention on BMI ______________________________________62
2.6.3.1 Multiple linear regressions ___________________________________67
2.7 Conclusions ________________________________________________________72
Appendix _____________________________________________________________74
Texas Tech University, Janani R. Thapa, August 2014
vi
CHAPTER III: THE EFFECT ON SUPERMARKET FOOD PURCHASES FROM
POINT OF SALE NUDGES WITH COMMUNITY REINFORCEMENT _______83
Abstract ______________________________________________________________83
3.1 Introduction ________________________________________________________84
3.2 Review of literature__________________________________________________87
3.2.1 Risk factors for obesity ___________________________________________87
3.2.2 Supermarket interventions _________________________________________88
3.2.3 Use of scanner data ______________________________________________89
3.2.4 Influence of personal characteristics in food choice decisions _____________90
3.3 Conceptual framework _______________________________________________91
3.4 Methods ___________________________________________________________94
3.4.1 Research setting _________________________________________________94
3.4.2 Data __________________________________________________________95
3.4.2.1 Produce _________________________________________________95
3.4.2.2 Point of sale nudges ________________________________________96
3.4.2.3 Community awareness and project activities data _________________97
3.4.3 Data analysis ___________________________________________________98
3.4.3.1 Produce data analysis _______________________________________99
3.4.3.2 Point of sale nudges data analysis ____________________________101
3.4.3.3 Community awareness and use of project activities ______________101
Texas Tech University, Janani R. Thapa, August 2014
vii
3.5 Results ___________________________________________________________102
3.5.1 Produce sales __________________________________________________102
3.5.1.1 Paired comparison in intervention community __________________102
3.5.1.2 Change in produce sales and overall comparison ________________104
3.5.2 Point of sale nudges _____________________________________________108
3.5.2.1 Change in sales and overall comparison _______________________108
3.5.3 Community awareness and use of project activities ____________________113
3.5.3.1 Community’s perception on project activities ___________________113
3.5.3.2 Awareness by demographics ________________________________116
3.6 Conclusions _______________________________________________________124
Appendix ____________________________________________________________126
CHAPTER IV: COMMUNITY-BASED BEHAVIOR CHANGE INTERVENTION
FOR OBESITY PREVENTION: A SYSTEMATIC REVIEW AND META-
ANALYSIS __________________________________________________________130
Abstract _____________________________________________________________130
4.1 Introduction _______________________________________________________131
4.2 Review of literature_________________________________________________134
4.3 Conceptual framework ______________________________________________136
Texas Tech University, Janani R. Thapa, August 2014
viii
4.4 Methods __________________________________________________________140
4.4.1 Search strategy _________________________________________________140
4.4.2 Language and data restrictions _____________________________________141
4.4.3 Selection criteria _______________________________________________141
4.4.4 Data extraction and synthesis______________________________________142
4.4.5 Validation of systematic review ___________________________________1423
4.4.6 Meta-analysis __________________________________________________145
4.5 Results ___________________________________________________________145
4.5.1 Trend in community-based intervention evaluation research _____________147
4.5.2 Narrative synthesis ______________________________________________148
4.5.2.1 Small-scale relatively more-intense interventions ________________148
4.5.2.2 Large-scale relatively less-intense interventions _________________151
4.5.3 Meta-analysis __________________________________________________152
4.5.4 Large-scale, less-intense or small-scale, more-intense interventions? ______162
4.6 Conclusions _______________________________________________________163
Appendix ____________________________________________________________165
CHAPTER V: SUMMARY AND CONCLUSIONS _________________________174
BIBLIOGRAPHY _____________________________________________________178
Texas Tech University, Janani R. Thapa, August 2014
ix
ABSTRACT
Unlike the traditional way of addressing obesity as an individual’s concern, it is now
often addressed as a community’s problem for a societal level effect. A conceptual
framework, by studying obesity as a market failure, was developed for the potential of
behavior change interventions in obesity prevention. Then, a multi-tiered supermarket
and community-based obesity prevention model, implemented in a small community, was
evaluated for its effect on 1) cancer knowledge and BMI, and 2) supermarket sales of
fresh fruits and vegetables (produce) and selected healthier items. The first evaluation
was done by comparing food behavior and attitude before and after the project and by
analyzing the intervention effect on cancer knowledge and BMI. The results suggests that
this community-based model has a good potential for bringing long term change in
obesity through targeting the behaviors associated with obesity and should have a
sustainable impact. The second evaluation was done by comparing the change in sales
from the pre-trial to trial period in the intervention store with the change in the control
store. The findings suggest that supermarket interventions have potential in changing
food behavior. Finally, a systematic review and meta-analysis of community-based
behavior change interventions was done. A narrative analysis and meta-analysis suggests
that community-based behavior change interventions have promise for changing
behaviors towards obesity prevention and to some extent changing (decreasing) BMI.
Additionally, the difference in project impact between large scale less intense and small
scale more intense interventions suggests that designers of community-based obesity
prevention intervention should carefully consider between large scale less intense and
small scale more intense interventions for desired outcomes.
Texas Tech University, Janani R. Thapa, August 2014
x
LIST OF TABLES
Table 2.1 Description of explanatory variables ________________________________51
Table 2.2 Comparison of cancer risk awareness in community group _______________53
Table 2.3 Comparison of cancer risk awareness in paired group ___________________53
Table 2.4 Sustainability of cancer risk awareness in the intervention community ______54
Table 2.5 Comparison of project impact indicator variables in the community ________55
Table 2.6 Comparison of project impact indicator variables in the paired group _______56
Table 2.7 Food behavior and other cancer risk behavior _________________________58
Table 2.8 Statistical relationship on cancer knowledge score (CKS) ________________60
Table 2.9 Effect of intervention on cancer knowledge score (N=106) _______________61
Table 2.10 Body mass index (BMI) and waist circumference (WC) outcomes ________63
Table 2.11 Summary statistics of dependent and explanatory variables _____________67
Table 2.12 Statistical relationship on BMI ____________________________________70
Table 2.13 Self-perception of body weight (N=1067) ___________________________71
Table 3.1 Sales of produce during pre-trial and trial period ______________________103
Table 3.3 Change from pre-trial to trial period for produce (mean no. of units sold) __107
Table 3.4 Total percentage change (units sold) from the pre-trial to trial period ______110
Table 3.5 Change from pre-trial to trial period (mean units sold per month) _________112
Table 3.6 Community views of the project during the trial period _________________114
Texas Tech University, Janani R. Thapa, August 2014
xi
Table 3.7 Community views of the project from the follow-up survey _____________116
Table 3.8a Use of NuVal score system by education level_______________________117
Table 3.8b Use of in-store signs by education level in the intervention community ___118
Table 3.9a Use of NuVal score system by income groups _______________________118
Table 3.9b Use of in-store signs by income groups in the intervention community ___119
Table 3.10a Use of NuVal score system by gender ____________________________120
Table 3.10b Use of in-store signs by gender in the intervention community _________120
Table 3.11a Use of NuVal score system by race ______________________________121
Table 3.11b Use of in-store signs by race in the intervention community ___________121
Table 3.12a Use of NuVal score system by language __________________________122
Table 3.12b Use of in-store healthy eating signs by race ________________________122
Table 3.13a Use of NuVal score system by age group __________________________123
Table 3.13b Use of in-store signs by age group _______________________________123
Table 4.2 Body weight comparison from selected interventions __________________155
Table 4.3 BMI comparison in selected interventions ___________________________157
Table 4.4 Data for meta-analysis __________________________________________159
Texas Tech University, Janani R. Thapa, August 2014
xii
LIST OF FIGURES
Figure 1.1a Negative Externality in Production _________________________________6
Figure 1.1b Leftward Shift in the Demand to Decrease the Social Cost _____________11
Figure 1.2 Externality of High-Fat, Sugar and Salt Food Products _________________13
Figure 2.1 Self-reported Obesity among US Adults (BRFSS, 2012) ________________26
Figure 2.2 Norms for Cancer Prevention _____________________________________34
Figure 2.3 Mediating Variable Model _______________________________________35
Figure 2.4 Project Intervention and Control Site _______________________________37
Figure 2.5 Survey Respondents ____________________________________________38
Figure 2.6a Mean BMI Trend in Total Male Respondents ________________________64
Figure 2.6b Mean BMI Trend in Total Female Respondents ______________________64
Figure 2.7a Mean BMI Trend in Male Community Respondents’ Group ____________66
Figure 2.7b Mean BMI Trend in Female Community Respondents’ Group __________66
Figure 3.1 Conceptual Framework for the Supermarket Intervention _______________91
Figure 4.1 The Systematic Review Process Used ______________________________144
Figure 4.2 Trend in Community-based Evaluation Research _____________________147
Figure 4.3 Forest Plot of the Standard Difference in Mean BMI __________________161
Texas Tech University, Janani R. Thapa, August 2014
1
CHAPTER I
BEHAVIORAL INTERVENTIONS TO ADDRESS OBESITY
AS A MARKET FAILURE
Abstract
Obesity in the US is pervasive across gender, ethnic origin, and socioeconomic status.
The cause of obesity at the individual level is the result of genetics, environment and
choices. Additionally evidence increasingly suggests that an environment conducive to
obesity at a personal level results in obesity and the associated energy imbalance. A key
factor affecting energy imbalance is a market system that has increasingly supplied food
high in calories including the foods that are sweeter and saltier. Obesity outcomes from
this market could be considered to be an externality of the demand and supply of high
calorie foods. This paper discusses obesity as a negative externality in food production
and consumption as it develops a framework for behavioral interventions to decrease the
demand of unhealthy foods. This could be used to conceptually evaluate for example
taxation to reduce demand for unhealthy products such as cigarettes, alcohols and
perhaps unhealthy foods that are high in fat, sugar and salt. This paper looks at the
potential of behavioral interventions as a complement to the ongoing efforts in changing
the supply and demand of products that are of concern to public health.
Key words: obesity, market failure, behavior change, community-based.
Texas Tech University, Janani R. Thapa, August 2014
2
1.1 Background
Obesity in the US is pervasive across gender, ethnic origin, and socioeconomic
status. Associated statistics and comorbidity have been cited and reported in many
scholarly works (Guh et al., 2009; Ogden et al., 2012 and NCI, 2013). According to the
Center for Disease Control and Prevention (CDC, 2012), the percent of adults (age 20
years and over) who were obese was 35.9% in 2010. Obesity is among the leading
modifiable behavior risk for morbidity, mortality and disability in Americans (Mokdad et
al., 2004). Obesity prevalence remains high according to the most recent obesity trend
analysis (Ogden et al., 2014).
The cause of obesity at individual level is the result of genetics, environment and
choices. Additionally an increasing area of evidence suggests that an environment
conducive to obesity at a personal level is the result of energy imbalance. An energy
imbalance is when more calorie is consumed than is burned over a number of time
periods (Martinez, 2000; Chou et al., 2002 and Martínez et al., 2002). The World Health
Organization (WHO) has defined energy imbalance as the fundamental cause of obesity
and overweight status (WHO, 2013). Further, WHO (2013) has pointed out that globally
there has been an increased intake of energy dense foods that are high in fat and a
decrease in physical activity due to increasingly sedentary nature of many forms of work,
changing modes of transportation, and increasing urbanization. Research has been
conducted in different disciplines regarding the effect of energy imbalance on obesity:
nutrition and physical activity (Dionne et al., 1997; Seale and Rumpler, 1997;
Murgatroyd et al., 1999 and Mobasheri et al., 2005) and psychology (Williamson and
Stewart, 2005 and Sluis et al., 2010).
Texas Tech University, Janani R. Thapa, August 2014
3
In the realm of demand and supply, a key factor affecting energy imbalance is the
increasing consumption of food high in calories. In particular, foods that are sweeter and
saltier have been increasingly demanded and supplied. As the US food system
transformed from small farm production and home cooking to efficient and specialized
large scale production and manufacturing of ready to eat foods obesity has increased.
With the food system transformation came value addition in foods, keeping cost low and
extensive advertising (Apple, 2003). US consumers adopted with the transformation in
the food system by changing their food (and lifestyle) behavior. Americans now purchase
more of their foods away from home (Stewart, 2011). The sequence of food system
transformation has led to path dependency in food choice decision towards increased
unhealthiness. Swinburn (2011) suggests that the global obesity pandemic is the outcome
of these changes in global food system.
The focus of this paper is looking at obesity as a path dependent outcome of
increasing demand and supply of high calorie foods. Several behavioral and economic
factors affect food consumption decisions (Finkelstein et al., 2005). The market success
of high calorie foods (sweeter and saltier foods) have been said to have contributed to the
obesity epidemic (Finkelstein et al., 2005; Moodie et al., 2006 and Anand and Gray,
2009). A key issue in the debate on the best method of obesity prevention is the need to
identify key points that should be considered in obesity prevention. Evaluating the
problem of obesity from a theoretical approach should lead to potential solutions by
framing the opportunity set. The paper discusses obesity as a market failure and then
discusses community-based behavioral interventions to be a potential method to prevent
Texas Tech University, Janani R. Thapa, August 2014
4
and curb obesity. Studying obesity as at least partly a market failure based on commercial
success has potential to trigger much needed discussion to better combat obesity.
1.2 Obesity: a market failure
Using a market failure approach in health service research is not new. The market
failure concept in chronic disease markets has been used by Watts and Segal (2009). In
obesity research, childhood obesity has been described to be a “market failure” of
obesogenic-calorie dense and activity limiting products by Moodie et al. (2006). This
view has been supported by Lawrence (2009). Lawrence (2009) reported that the
expanded profile of nutrition is exploited to benefit food sales rather than public health.
Further, Swinburn (2009) pointed out that the market failure leading to obesity should be
best addressed by considering food marketing to children as a central policy choice.
Market failure exists when a free market allocation is not efficient. A free and
competitive market allocates resources efficiently. However, in the real world the
markets often are inefficient for various reasons including when there are externalities.
An externality is an unintended consequence, positive or negative, of an economic
activity born by consumers or producers. The concept of obesity as a market failure is
also explained in Moodie et al. (2006). They have argued that the market has sustained
and promoted unhealthy food choices. The current market has led to overconsumption of
food products, particularly unhealthy products. They suggest that significant government
intervention may be needed to correct this market failure.
In addition to what has been proposed by Moodie et al. (2006), an additional
argument to consider is that marketing channels and marketers are more informed than
the consumers because the industry develops sophisticated market research. As stated by
Texas Tech University, Janani R. Thapa, August 2014
5
Nelson (1970) “consumers are continually making choices among products, the
consequences of which they are but dimly aware.” This is still true. The information set
available to consumer before a food purchase decision is made is constrained by the
consumer’s socioeconomic status. Further behaviors are influenced by marketing
strategies including price, place, advertising and promotion.
There is a high social cost of excessive unhealthy food consumption. Obesity can
be considered to be a negative externality of unhealthy food market. There is evidence in
the literature on obesity being a negative externality of production, but it is less
developed. Finkelstein et al. (2005) has examined economic causes such as advancement
in technology behind the causes of obesity epidemic. This paper will discuss obesity as a
negative externality in production and consumption.
1.2.1 Negative externality in production
A negative externality in production occurs when there are costs caused by the
production of a good which are not borne by the producer. That is the private marginal
cost of production which determines the supply is below the social marginal cost. Since
the producer does not bear the full cost factor input and production are higher than the
socially optimal level of production. The effect of a negative externality in production is
described in Figure 1. The case of a negative externality in the production of unhealthy
food products is based on the fact that unhealthy food consumption is at odds with the
assumption of perfect competition, particularly full information and rational decisions
based on preferences. However, assumption of rational decision is affected by several
psychological factors, commonly studied by behavioral economists. Behavioral
economics seeks to explain common and systematic deviations from the behavior implied
Texas Tech University, Janani R. Thapa, August 2014
6
by a fully rational economics model (Just, 2014). Hence, with increasing evidence from
behavioral economics, weight outcomes from consumption of unhealthy food can be
considered negative externality in consumption. Consumers’ lack of awareness and
irrational decisions will be discussed as resulting in a negative externality in
consumption. Moreover, producer’s role in promotion and marketing of foods that are
high in fat sugar and salt play a role and hence it becomes useful to consider obesity as an
externality in production.
Figure 1.1a Negative Externality in Production
In Figure 1.1a, the vertical and the horizontal axis are the price and quantity of
unhealthy foods that are high in fat, sugar and salt. The current competitive market
equilibrium is at 1, is the aggregate demand curve, is the marginal private cost
curve. The quantity demanded is 1 at equilibrium 1.
However, consider the social cost of 1 quantity of unhealthy food items being
produced and consumed. This social cost is not paid by producers and could be
considered to be a net cost to society. represents the supply curve considering this
Texas Tech University, Janani R. Thapa, August 2014
7
marginal social cost. The social cost is the negative health impact of consuming excessive
amounts of food high in fat, sugar and salt. While quantifying this cost is not easy, it
shows the social cost from the production and consumption of unhealthy foods. Triangle
in the figure is the obesity burden and is the cost of the negative externality in
production.
When unhealthy foods are produced the marginal private cost of the producers
determines the supply. However, there are costs caused by the production of these
unhealthy foods, the costs that form the negative externality in production are the social
costs of production. The social costs of unhealthy food production are the costs born to
the society due to obesity for example increasing expenditure in health related issues
caused due to obesity. The economic costs of obesity which allows to estimate triangle A
have been summarized by Mccormick et al. (2007). Similarly, the consequences of
obesity in terms of annual medical costs, lifetime medical costs, nonmedical
expenditures, and occupation choice and wages have been summarized in Finkelstein et
al. (2005). These costs are born directly by the consumers but also by society. Evidence
of the costs of obesity to the society is abundant. In 2008 the national estimated cost of
obesity totaled about $147 billion (Finkelstein et al., 2009).
The negative externality in production due to unhealthy foods, unlike such
externalities in pollution for example comes only after excessive consumption of the
unhealthy food. The argument is if pollution destroying the environment is an externality
then the excessive marketing and consumption of unhealthy foods destroying the food
environment can also be considered to be externality. Hence, obesity is a market failure
Texas Tech University, Janani R. Thapa, August 2014
8
caused both due to negative externality in production and negative externality in
consumption.
1.2.2 Negative externality in consumption
If the marginal social benefit of consumption is less than the marginal private
benefits, it is considered to be a negative externality in consumption. Smoking and
alcohol consumption has been widely explained as a case of negative externality in
consumption. A question is whether foods high in fat, sugar and salt are parallel to
smoking and alcohol consumption? A case of obesity as a negative externality in
consumption was explained in (Camerer et al., 2003). According to Camerer et al. when
consumers make errors, it imposes costs on themselves because the decisions they make
(reflected by their demand) which do not fully reflect the benefits and costs they derive.
Individual consumption also has an external effect on society. The outcomes of
individual decisions are also faced by society in the societal obesity burden (see section
1.2.1).
Texas Tech University, Janani R. Thapa, August 2014
9
1.3 Addressing the market failure
1.3.1 Taxation
In scenarios such as demonstrated in Figure 1.1a, the traditional approach is for
governments to intervene to fix the market failure. A natural response from economists
for externalities is to impose a Pigouvian tax. Using a Pigouvian tax in theory internalizes
the cost of the externality including both externality in consumption and production. The
evidence of imposing tax to reduce the negative externality of production is questionable.
As the result of a tax, consumers can shift their spending towards cheaper substitutes (see
Edwards, 2011). In general, a tax is not effective unless the benefit of the tax offsets the
cost of taxation born by consumers, producers and the tax collectors. In addition taxation
distorts behavior and reduces well-being, especially for low income families from a
regressive tax (Cawley, 2010). Sturm et al. (2010) reported that a greater impact of small
taxes could come from the dedication of the revenues they generate to other obesity
prevention efforts rather than through their direct effect on consumption. Or, at the least
taxation of soda should accompany taxing candy, ice cream and fried foods along with
subsidizing broccoli, gym membership and dental floss (Mankiw, 2010). Other studies
have looked at soda tax and its effect on BMI but have seen an effect small in magnitude
(e.g. Fletcher et al., 2010).
Texas Tech University, Janani R. Thapa, August 2014
10
1.3.2 Behavioral interventions
An alternate or complement to taxation is applying behavioral interventions to
shift supply and demand towards increasing the demand of healthy foods and hence
creating a new supply channel to meet the increasing demand. An approach recently
being used to overcome obesity in health promotion activities is to promote healthy food
habits and physical activity.
The recent trend has been on adopting translational evidence based community-
based participatory models. Community-based participatory models have shown to be
effective in reversing the obesity trend and in disease prevention (e.g. Brownson et al.,
1996; Chou et al., 1998 and Allen et al., 2013). Other studies like Baker and Brownson
(1998), Merzel and D’Afflitti (2003), and Corda et al. (2010) including the American
Cancer Society (2012) have emphasized community-based programs which encompass
multiple interventions as the model with best potential to achieve behavior change that
will reduce person’s health risk. Also, several approaches of behavioral economics have
shown success in combating obesity trends in schools, as reported in Just and Payne
(2009).
The research concept is to shift the demand for unhealthy foods to the left and
encourage demand, supply and consumption of healthier products. This paper develops a
research framework to look at the possibility of providing information and awareness to
encourage the leftward shift in the demand of unhealthy foods. The Dietary Guidelines
for Americans 2010 recommend stronger environmental strategies for improving the
population’s eating practices, including interventions to influence food purchasing
behavior in stores (USDA-USDHH, 2010). The Food Marketing Institute (FMI) reported
Texas Tech University, Janani R. Thapa, August 2014
11
that consumers averaged 2.2 trips per week to the supermarket in 2010 (FMI, 2011).
Hence, supermarkets play an important role in food purchasing (Glanz and Mullis, 1988).
Gittelsohn and Lee (2013) also recommended that the grocery store interventions should
be coordinated with community reinforcement to encourage awareness and to shift
purchases towards healthier products.
Figure 1.1b Leftward Shift in the Demand to Decrease Social Cost
In Figure 1.1b, an integrated effort to shift D to the new demand curve D and at
the new equilibrium 2 where the quantity demanded is 2 can in the long run decrease
the marginal social cost to . This framework shows potential to decrease the obesity
burden to triangle by intervening to shift the demand curve leftward.
Texas Tech University, Janani R. Thapa, August 2014
12
1.4 Addressing obesity as an externality
The relative cost of changing behavior to shift the demand and supply curve is
very important. Additionally the market does not identify obesity or other health risk
from foods high in fat, sugar and salt in the production and marketing of product.
Therefore, an added dimension is needed to explain externality. As such, a two panel
diagram has been used to address obesity as an externality in Figure 1.2. Mccormick et al.
(2007) cited individuals not bearing the full cost of their health costs as a type of
externality of obesity. Moreover, the paper has cited obese people not being in
employment as an externality. The individual bears the foregone earnings, the costs of
unemployment benefit, incapacity benefit and foregone tax revenues are borne by
society. A conceptual model has been developed to look at the unhealthy food market’s
externalities.
Texas Tech University, Janani R. Thapa, August 2014
13
Figure 1.2 Externality of High-Fat, Sugar and Salt Food Products
Texas Tech University, Janani R. Thapa, August 2014
14
In Figure 1.2, the top panel shows the high calorie food’s market equilibrium 1.
At this equilibrium the quantity demanded is 1. The area above the supply curve and
below the demand curve left to the equilibrium is the net welfare, which is the sum of
producers’ surplus and consumers’ surplus in the current market. The new demand curve
which has shifted leftwards (decreased) due to the application of behavior change
interventions is and the equilibrium is 2. The total welfare now has decreased to the
area under and above S, and right to 2. Under the framework of negative externality
of unhealthy food market there are externalities of unhealthy food products’ market
depicted in the bottom panel of Figure1.2.
The bottom panel of figure 1.2 shows the autonomous level of externality 0 in
the form of genetic causes of obesity, other health problems caused regardless of the food
habit. The line E is the supply function of externality that is depicted by the supply of
unhealthy food products in the top panel. As awareness about the effect of the externality
increases, this function can be more elastic. However, unlike the externality effect in
production supply function, in this framework the externality is caused only when what is
produced is consumed. Similarly, the line is the externality with the new decreased
demand . Therefore, the bigger triangle B is the externality caused by 1 level of
production and consumption. The correct welfare measure then will be the welfare in top
panel minus the triangle B in bottom panel. This will give the net welfare that accounts
for obesity as externality. Similar partial equilibrium framework was developed by
Segarra et al. (1991) to develop optimal input use in agriculture.
Texas Tech University, Janani R. Thapa, August 2014
15
Therefore the gain of decreasing the demand of foods high in fat, sugar and salt
can be measured by the gain in externality (which is a decrease in the externalities social
cost). As long as the decrease in externality effect is higher than the welfare loss in the
unhealthy foods market, collaborative approach must be applied to decrease demand of
such foods. How big is the externality effect, what we know? Mccormick et al. (2007)
have listed available estimates of the financial costs of obesity. The two primary costs
they have listed are health costs and employment costs.
The shift in demand shown in Figure 1.1b from to incurs cost. The present
basket of goods that is chosen by consumers is not necessarily the basket chosen under a
full informed decision. The decision is path dependent of the change in food system and
lifestyle behavior. There has been success in influencing this choice and demand set by
translational and community-based research. However from a policy perspective the cost
of these interventions should be less than the decrease in the externality cost. The cost of
intervention on changing the demand should be lower than or equal to decrease in
externality. The interventions are cost effective only as long as the decrease in externality
is higher than the welfare loss in the obesogenic market and the decrease of the
externality should also offset the cost of behavior change intervention. Hence it is very
crucial to develop low cost intervention that has high impact. Behavioral interventions
including the tools’ of behavioral economics have shown success and promise (Thomson
and Ravia, 2011).
An additional option is that government should intervene when market correction
is needed. However, there are cases of failed government intervention suggesting that not
all market forces can be affected by government intervention. A potential way to fix a
Texas Tech University, Janani R. Thapa, August 2014
16
market failure could be by encouraging the market to fix itself by impacting the supply
and demand through behavioral interventions. This dissertation is based on the
conceptual framework of allowing the unhealthy food market to fix itself by intervening
in the area of behavior change to increase the demand for healthy food products which
will cause suppliers to gradually switch selling sugar loaded products by healthier
products. This will also gradually close the gap between the social and private costs of
unhealthy foods being demanded and supplied.
1.5 Problem statement and research objective
Obesity is increasingly seen as a societal problem rather than only an individual’s
problem. It is increasingly viewed as a problem resulting from the interactions of
individual behavior, genetic predisposition and environment (Hernandez et al., 2006).
The environmental factors can be influenced for individuals to improve physical activity
and eating behaviors that encourage individuals to achieve and maintain healthy weight.
One approach to the obesity epidemic is Community-based Participatory Research
(CBPR). In this type of effort, community works together towards a common goal of
obesity prevention. CBPR may be the answer to the need for practical, affordable, and
scalable intervention strategies that effectively induce weight loss and prevent weight
gain.
Texas Tech University, Janani R. Thapa, August 2014
17
Community-based obesity prevention project
A public-private partnership for cancer prevention in rural communities was
implemented in a small West Texas community employing a multi-tiered CBPR
approach. The program was designed with potential to develop a community-based
obesity and cancer prevention model which would be feasible for adoption by small
communities. The model was unique in having a large supermarket serving the
community at the center of program activities. One potential to use the supermarket is to
nudge community’s food choice towards healthy food choices. The supermarket can be a
key part of CBPR-approach also because it serves as a community meeting place and an
information center in sparsely populated areas and in other contexts. A key to success in
the CBPR approach is to enable community residents to more actively participate in the
full effort to promote healthier living.
The three essays in this dissertation address the above discussed issues
independently, each employing its own set of methodologies. A brief description for each
of the essays is provided in the following section, the description will include the need
and appropriateness of each study in tackling the obesity epidemic through community-
based behavior change interventions.
Texas Tech University, Janani R. Thapa, August 2014
18
1.5.1 Essay I
The first essay focuses entirely on the impact of the multi-tiered CBPR model
developed to reduce obesity risk factors by changing cancer knowledge, health attitudes
and food behavior. The objective of the CBPR project was to raise cancer knowledge,
nutrition awareness and change food behavior while building positive health attitudes to
reduce obesity and cancer risk. The model developed in this project accounts for
community heterogeneity and reinforces behavior change at multiple levels. Changing
individual’s decision making within a community context was central to this model.
The model delivered educational interventions to a small community using
several means to encourage healthy eating and healthy weight maintenance. The
interventions included articles in the local newspaper, television, public service
announcements, posters at community sites, presentations to community groups and the
supermarket. Earlier research has shown that the effect of exercise, smoking, occupation,
and race vary by sizeable amounts from high to low BMI quantiles (Belasco et al., 2012).
The impact of this model was analyzed using health promotion impact evaluation
methods.
1.5.2 Essay II
The second essay is on the effect of the project including supermarket
interventions on food purchases. Since the local supermarket is the largest source of daily
food needs it can be an excellent source of information and reinforcement for healthy
behavior. Earlier research on behavioral change has reported reinforcements for behavior
change that include different aspects of the environment are more effective than simple
reinforcement methods. Nudges in the form of shelf talkers were placed on relatively
Texas Tech University, Janani R. Thapa, August 2014
19
healthier items in a particular food category to promote healthy food choice decision. The
supermarket was used to: 1) conduct healthy food demonstrations, 2) put shelf talkers on
the comparatively healthy food product, 3) access food purchase data to analyze purchase
behavior as affected by project intervention, and 4) provide an active site of information
flow including posters on healthy eating that were placed in the store. Thus, the
supermarket was used to nudge participants towards healthy food behavior. The
outcomes of this supermarket intervention were evaluated using comparative analysis.
1.5.3 Essay III
The third essay focuses on the development of a cohesive and assimilative
research paper on the learnings from community-based behavior change interventions. A
systematic review and meta-analysis of research on community-based behavior change
interventions implemented for obesity prevention was done to this end. An assimilation
of literature on impact evaluation of community-based interventions designed to address
obesity has potential in guiding the design, implementation, and evaluation of future
community-based behavior change interventions.
Texas Tech University, Janani R. Thapa, August 2014
20
CHAPTER II
THE IMPACT OF A MULTI-TIERED COMMUNITY-BASED OBESITY AND
CANCER PREVENTION PROGRAM IN A SMALL COMMUNITY
Abstract
There is an increasing interest and effort in reducing obesity but relatively few of these
efforts focus on small communities. A multi-tiered model to reduce obesity risk factors
was developed and implemented to change food behavior and health awareness in a small
community over a one year period. This paper evaluates the project outcomes based on
the following: 1) mean and frequency comparisons of self-reported health behavior and
attitude response before and after the intervention, 2) intervention effects on cancer
knowledge score, and 3) intervention effects on body mass index (BMI). Findings show a
significant increase in obesity as cancer risk awareness and food nutrition awareness as
well as a significant intervention effect on cancer knowledge score. This increased food
nutrition awareness should lead to reduced obesity over time as indicated by a regression
analysis of obesity risk factors on BMI. Overall, this adds to evidence that community-
based interventions such as these have potential in bringing long term change in obesity
and awareness towards cancer inducing behaviors.
Key words: body mass index (BMI), cancer risk, food behavior, health education, multi-
tier approach, obesity, small community
Texas Tech University, Janani R. Thapa, August 2014
21
2.1 Introduction
The prevalence of overweight and obesity in the United States continues to be
high. Ogden et al. (2012) have tracked the prevalence of obesity among the US adults.
According to the report there was no change in the prevalence of obesity among adults or
children from 2007-2008 to 2009-2010. Obesity is also among the leading modifiable
behavior risk for morbidity, mortality and disability in the Americans (Mokdad et al.,
2004). Obesity also causes heart disease and cancer which are also among the top five
preventable causes of death (Yoon, 2014). The prevalence of obesity is now not only in
developed countries but includes emerging economies like China and India that have
reported increased prevalence of obesity. For example, the overweight rate in India
increased by 20% between 1998 and 2005 (Sinha, 2010). The prevalence of obesity and
overweight is increasing to epidemic proportions at an alarming rate in China (Gao et al.,
2011 and Shang et al., 2012).
There is relatively little evidence available in the obesity literature that shows
programs developed and tested to bring healthful obesity prevention services to small
communities and limited success is seen in larger communities. Programs developed and
tested in the past (e.g. Rodgers et al., 1994 and Economos et al., 2007) have been located
in larger urban areas. Interventions with even a moderate success would be an
improvement in small community context. In addition, a key benefit of working in a
small community is the ability to control the media and food environment. Also the
overall community context is much easier, requires fewer actors and much lower cost.
Moreover, innovative public approaches including a variety of environmental initiatives
designed to increase fruit and vegetable consumption have been recommended by
Texas Tech University, Janani R. Thapa, August 2014
22
previous research in obesity prevention (Giang et al., 2008 and Flegal et al., 2012).
Gittelsohn and Lee (2013) also recommended that the grocery store interventions should
be coordinated with community reinforcement to encourage awareness and to hopefully
shift purchase to healthier products.
Obesity has been found to be associated with certain socioeconomic status (SES).
Though there are individual variation, there is a predictable pattern (Swinburn,
2011).There is a significant variation in obesity trend by race and ethnicity (Flegal et al.,
2012). Researchers have also found that residents in small communities are more obese
than their urban counterparts (Blankenau, 2009). Similarly, an individual’s SES affects
timely access to health messages and health care services. The American Cancer Society
has outlined that uninsured patients from ethnic minorities are substantially more likely to
be diagnosed with cancer at a later stage when treatment is extensive and costly
(Gosschalk and Carozza, 2010; Philips et al., 2011and Gong et al., 2012).
A multi-tiered community-based obesity prevention project was launched in
June of 2011 in a small community in West Texas. The objective of the project was to
raise nutrition awareness and reduce obesity induced health risks by changing food
behavior. Multiple community-based and reinforced efforts were used to provide
healthier food and activity choice information and improve health and nutrition
awareness. In particular, the supermarket, given its relatively prominent role in the
community and lack of choice in the community, provided a means to promote healthy
food choices. This paper evaluates the success of this approach and analyzes its outcome
by comparing food behavior and attitude before and after the project and by analyzing the
intervention effect on cancer knowledge and BMI.
Texas Tech University, Janani R. Thapa, August 2014
23
2.2 The project concept and model
Research has shown that the effect of exercise, smoking, occupation, and race
vary by sizeable amounts from high to low BMI-quantiles (Belasco et al., 2012).
Research has also shown that, in the US, higher BMI is ubiquitous for groups with lower
SES and specially minorities (Clarke et al., 2009 and Grabner, 2012). Communities with
a high percentage of minorities are generally medically underserved and have high rates
of disease incidence (Gosschalk and Carozza, 2010 and Gong et al., 2012).
To address this concern, a multi-tiered behavior change model was developed to
account for community heterogeneity and reinforce behavior change at multiple levels.
The concept of changing individual’s decision making within a community context was
central to this behavior change project model. The model delivered educational
interventions using several means to encourage healthy eating and healthy weight
maintenance. This approach included articles in the local newspaper, television, posters at
community sites, presentations to community groups and the supermarket.
Identifying local community leaders was another important component of this
model as suggested by Macaulay et al. (1998), Corda et al. (2010), and Hystad and
Carpiano (2012). As such, community leaders participated in the project advisory board.
Community leaders included leaders from the local school district, churches, city and
county government, extension service, media (local newspaper and television station),
and community organizations, such as 4-H clubs, the Boy and Girl Scouts, and the local
senior center, along with the manager of the local supermarket. These all participated as
local community leaders during the implementation of this project. We have described
Texas Tech University, Janani R. Thapa, August 2014
24
this supermarket based community-based multi-tiered model more fully in McCool et al.
(2013).
In regards to the primary goal of individual behavior change this research model
can be related to behavior change models like the Trans-theoretical Model (TM). Stages
of change are central in TM where views and behavior changes through a process over
time, as individuals move from pre-contemplation stage to maintenance stage
(Zimmerman et al., 2000 and Taylor et al., 2004). The underlying research objective is
attained when individuals believe that benefits of performing a behavior outweigh its
costs and they are persuaded to change their behavior reinforced by the multi-tiered
approach.
2.3 Review of literature
2.3.1 Obesity and obesity in the US
2.3.1.1 Obesity and overweight
Overweight and obesity are both labels for ranges of weight that are greater than
what is generally considered healthy for a given height (CDC, 2013). Overweight and
obesity ranges are determined by using weight and height to calculate a number called
“body mass index (BMI).” BMI is a measure of an adult’s weight in relation to an
individual’s height and is calculated with this formula:
( )
An individual is considered obese when BMI is more than 30, overweight when
BMI ranges from 25 to 30, normal when BMI is less than 25, and a person with a BMI
under 18.5 is considered underweight. This range was used throughout this research
regardless of sex. The National Heart, Lung and Blood Institute guidelines also
Texas Tech University, Janani R. Thapa, August 2014
25
recommends using other two predictors: a) waist circumference (WC) because abdominal
fat is a predictor of risk of obesity-related diseases, and b) other risk factors like high
blood pressure and physical inactivity. WC has also been found to be more strongly
associated with morbidity and mortality than BMI ( Janssen et al., 2004 and Brown, 2009
and Seidell, 2011). However, despite the American Heart Association’s only recent
endorsement of both BMI and WC as primary tools for assessing adiposity, waist
circumference is less commonly used than BMI. Wannamethee et al. (2010) showed that
the association between both WC and BMI and cardio-respiratory fitness is gender
dependent. Moreover, there is less evidence in the literatures about the use of WC as a
single outcome variable. BMI is still used by the Center for Disease Control and
Prevention as a screening tool for its reliability, ease-to-perform and its inexpensive
nature (CDC, 2013). As such, BMI was used in this research as the primary outcome
variable. Changes in WC were analyzed but not as the primary outcome variable.
2.3.1.2 Obesity in the United States
The prevalence of overweight and obesity in United States continues to be high.
According to the Center for Disease Control and Prevention (CDC, 2012), percent of
adults (age 20 years and over) who were obese was 35.9%, and the same group who were
overweight, including obesity was 69.2% (2009 to 2010). Ogden et al. (2012) have
tracked the prevalence of obesity among US adults. According to the report, there was no
change in the prevalence of obesity among adults or children from 2007-2008 to 2009-
2010. Obesity is among the leading modifiable behavior risk for morbidity, mortality and
disability in Americans (Mokdad et al., 2004). Obese individual have a 50 percent higher
risk of dying from cancer than their healthy counterparts (ACS, 2012). Calle et al. (1999)
Texas Tech University, Janani R. Thapa, August 2014
26
reported that the risk of death from causes such as cardiovascular disease, cancer, or other
diseases increased as BMI increased above the healthiest range of 23.5 to 24.9 in men
and 22.0 to 23.4 in women from a 14-year study of a million person cohorts. A study
using National Cancer Institute (NCI) Surveillance, Epidemiology and End Results
(SEER) data estimated that in 2007, about 34,000 new cases of cancer in men (4%) and
50,500 in women (7%) were due to obesity (NCI, 2013). Figure 2.1 shows the self-
reported obesity among US adults, 25 to 30% on average are obese.
Figure 2.1 Self-reported Obesity among US Adults (BRFSS, 2012)
The prevalence of obesity is now not only in developed countries but includes
emerging economies like China and India that have reported increased prevalence of
obesity. The overweight rate in India increased by 20% between 1998 and 2005 (Sinha,
2010). The prevalence of obesity and overweight is increasing to epidemic proportions at
an alarming rate in China (Gao et al., 2011 and Shang et al., 2012). Swinburn (2011)
Texas Tech University, Janani R. Thapa, August 2014
27
reported the global obesity pandemic to be the outcome of the changes in global food
system.
2.3.2 Obesity and cancer
2.3.2.1 Evidence of obesity and cancer
Numerous problems are associated with obesity including cardiovascular disease,
cancer, lost physical activity and active life span etc. This research focuses primarily on
obesity as a cause of preventable cancer. Sturm (2002) has stated that obesity roughly has
the same association with chronic health conditions as does twenty years’ aging and that
obesity is associated with 36% increase in inpatient and outpatient spending and a 77%
increase in medications. In an extensive research, more than 900,000 U.S. adults who
were free from cancer at enrollment in 1982 were tracked (Calle et al., 2003). There were
57,145 deaths due to cancer during 16 years of follow-up. This research concluded that
increased body weight was associated with increased death rates for all cancers combined
and for cancers at multiple specific sites. Bianchini et al. (2002) also showed that excess
body weight is directly associated with risk of cancer at several organ sites, including
colon, breast, endometrium, esophagus, and kidney. The AICR (2007) also found
convincing evidence of association between obesity and cancers of the esophagus,
pancreas, colon and rectum, breast, endometrium, and kidney and a probable association
between obesity and gall bladder cancer. Similarly, Guh et al. (2009) confirmed direct
associations between obesity and cancers of the breast, colon and rectum, endometrium,
esophagus, kidney, ovary, and pancreas using a systematic review and meta-analysis.
Texas Tech University, Janani R. Thapa, August 2014
28
2.3.2.2 Epidemiology
The American Cancer Society has reported that 1 out of every 3 cancer deaths in
the US is linked to excess body fat, poor nutrition and physical inactivity. These are
related and may contribute to cancer risk. Body fat alone contributes to 1 in 5 cancer
death (ACS, 2012). The mechanisms by which obesity induces or promotes tumor-
genesis vary by cancer site. Calle et al. (2003) listed insulin resistance and resultant
chronic level of insulin circulating in the blood relative to glucose, increased bio-
availability of steroid hormones and localized inflammation as the causes for cancer due
to obesity.
Sturm (2002) suggested that despite the consequences of addictive behavior such
as smoking and drinking alcohol is less costly than obesity the former has received more
attention in clinical practice and public health policy. In a study of expected years of life
lost in six preventable cancers (EYLL), Liu et al. (2013) concluded that potential life
years saved by successful prevention would be substantial for lung cancer, breast cancer
and colorectal cancer. According to this report, when considering the annual incidence in
2012, lung cancer would cause the greatest subtotal of EYLL (3,116,000 years) followed
by female breast cancer (1,420,000 years) and colorectal cancer (932,000 years).
2.3.3 Behavior
2.3.3.1 Behavior and cancer
The literature on the potential of behavioral/environmental modification in
preventing cancer cases date back to early 1980s. A classic study by Doll and Peto (1981)
cited by Gotay (2004) concluded that 75% to 80% of cancers in US could have been
avoided and were a result of tobacco and alcohol, physical inactivity, diet and biologic
Texas Tech University, Janani R. Thapa, August 2014
29
factors and occupational and environmental exposures. Research in this area has
recommended weight control and physical activity through healthy lifestyle, low body
weight maintenance and exercising most days of the week for cancer prevention
(Bianchini et al., 2002). Similarly a systematic review done in 2010 has concluded that
sedentary behavior is ubiquitous in contemporary society and has recommended research
on cancer and sedentary behavior to be a priority (Lynch, 2010). Further, the study
concluded that reducing sedentary behavior may be a viable new cancer control strategy.
In this line, according to Klein et al. (2014) behavioral research can address a wide
variety of key processes and outcomes across the cancer control continuum.
Behavior change requires an understanding of how behaviors are learned and
maintained. Available theories of health behaviors are presented in several publications
(e.g. Anderson, 2004 and Shumaker et al., 2009). Avery (2012) concluded from a
systematic review that interventions improved diet more effectively if they were
informed by behavior theory. Gotay (2004) has briefed that theories of health behavior
are based on the premise that different kinds of variables interact to affect behavior, and
modifying these variables and relationships among them can lead to behavior change.
The factors that affect behavior include individual differences such as knowledge,
attitudes, previous experience and personal preference, the social environment, and the
larger community.
2.3.3.2 Food behavior
Numerous studies have shown that food behavior is associated with obesity and
cancer: Ludwig et al. (2001) showed a positive association of obesity and cancer risk
with soft drink consumption and Epstein et al. (2001) showed an inverse relation with
Texas Tech University, Janani R. Thapa, August 2014
30
increased fruit and vegetable consumption. Similarly, Nicklas et al. (1993) showed an
inverse relation with eating breakfast. Jeffery and French (1998) examined the
association between TV viewing, fast food eating, and BMI. They concluded that TV
viewing hours and fast food meals were positively associated with increased energy
intake and BMI.
2.3.3.3 Theories of behavior change
The underlying conclusion of research (Avery et al., 2012) related to behavior and
obesity is that health outcomes are dependent on decisions guiding behavior. The bases
for understanding the decision variables are behavior or social science theories or
conceptual models. The theories most generally used in obesity prevention are
(Baranowski et al., 2003):
1. Knowledge-Attitude-Behavior Model
2. Behavioral Learning Theory
3. Health Belief Model
4. Social Cognitive Theory
5. Theory of Reasoned Action or Theory of Planned Behavior
6. Trans-theoretical Model and Stages of Change
7. Social Marketing
Significant misconceptions related to cancer causation, symptoms, and treatment
including expressed feelings of little preventative control of the disease were reported by
Carpenter and Colwell (1995). They concluded that increased knowledge is associated
with increased self-efficacy for cancer screening. Similarly, Weinrich et al. (1998)
documented the importance of providing educational programs to increase participation
Texas Tech University, Janani R. Thapa, August 2014
31
in prostate cancer screening. Additionally the theory of rational addiction has been used
to justify the lowering of BMI levels of already overweight and obese population by
increasing education level in Miljkovic et al. (2008).
2.3.4 Community-based participatory research (CBPR)
Substantial research has been reported on the effects of community-based
participatory research. Brownson et al. (1996) conducted a study to determine whether a
community-based risk reduction project affected behavioral risk factors for
cardiovascular diseases. They concluded that community-based interventions show
promise in reducing self-reported risk of cardiovascular disease within a relatively brief
period. The success was observed with modest resources.
Chou et al. (1998) reported significant reduction in cigarette use at the initial
follow-up (6 months) and alcohol use at the first 2 follow-ups (up to 1.5 years) in a study
that investigated the secondary prevention effects of a substance abuse primary
prevention program. This study concluded that primary preventions are effective in
reaching and influencing high risk adolescents in a non-stigmatizing manner. Similarly,
Leung et al. (2004) recommended CBPR as a framework that can be applied to gain a
better understanding of the social context in which disease outcomes occur.
Loughlin et al. (1999) reported mixed outcomes of CBPR from a study that
evaluated a four year community-based cardiovascular disease prevention program
among adults aged 18 to 65 years (living in St-Henri, a low-income inner-city
neighborhood in Montreal, Quebec). Though very few community wide program effects
were observed in the study, they reported that several interventions showed promise in
terms of community penetration and impact.
Texas Tech University, Janani R. Thapa, August 2014
32
Allen et al. (2013) reported CBPR to be a feasible intervention in positively
improving parenting contributors to reduce adolescent substance use. This study was
done to assess the feasibility of a family-skills training intervention developed using a
CBPR framework. This study also explored parental traditional values as a modifier of
preliminary effects.
2.3.4.1 Community interventions measuring BMI as the outcome variable
Economos et al. (2007) has reported the finding of a community intervention to
reduce the BMI z-score in children. A community-based environmental change
intervention was conducted in three culturally diverse urban cities of Massachusetts with
one intervention community (Sommerville) and two control communities with similar
demography. This intervention included many groups and individuals within the
community (including children, parents, teachers, school food service providers, city
departments, policy makers, healthcare providers, before and after school programs,
restaurants, and the media). They concluded that a community-based environmental
change intervention decreased BMI z-scores in children at high risk for obesity.
Similarly, Economos et al. (2009) reported lessons for working with restaurants to
improve health from working with community to improve availability of healthy menu
options in Somerville, Massachusetts.
Taylor et al. (2007) reported the findings of a similar research done to determine
the effectiveness of a two year community-based intervention to prevent excessive
weight gain in five to twelve years olds. They concluded that providing activity
coordinators and basic nutrition education in schools can reduce the rate of excessive
weight gain. Similarly, other studies with BMI as the outcome variable have shown
Texas Tech University, Janani R. Thapa, August 2014
33
association between BMI and life-style factors. Kamycheva et al. (2003) reported
negative association between physical activity, smoking and BMI and a positive
association between BMI and coffee.
Using BMI as dependent variable in an ordinary least square regression done with
employing the 1984-1999 BRFSS data, (augmented with state level measures pertaining
to the per capita number of fast food and full service restaurants, the prices of a meal in
each type of restaurant, food consumed at home, cigarettes, and alcohol, and clean indoor
air laws), Chou (2004) reported three major findings: a) a large positive effects associated
with the per capita number of restaurants and the importance of trends in this variable in
explaining the stability of obesity between 1960 and 1978 and the increase since 1978, b)
a downward trend in food prices which than account for part of the upward trend in
weight outcomes, and c) a positive cigarette price effect on cigarette consumption.
2.4 Conceptual framework
Prevention depends on individual behavior change embedded in a social context.
The research concern is to develop, implement, and evaluate a program that addresses
primary cancer prevention in small community where the population currently has
limited access to proper nutritional choices, health care, and educational programs on risk
factors for cancer. An added benefit of working in small community is the ability to
control media and overall community context. Further, working in small community
requires fewer actors and much lower intervention costs. Studies such as these can be
cost effective.
Texas Tech University, Janani R. Thapa, August 2014
34
The issue of access to healthy foods has been central to the work of many
community-based organizations around the country. Researchers have also shown that
only providing health information to people is an insufficient strategy to promote healthy
eating (e.g. Vaandrager and Koelen, 1997). A challenge in cancer prevention is to
achieve long lasting behavior change. A strategy to improve health behavior, including
improved dietary intake and activity level is needed as shown in Figure 2.2. Despite
people’s general awareness of the role of diet in their overall health, it has not led to
improved eating habits. Food purchases and consumption depend not only on the
individual, but also on social, cultural, and environmental factors (Loewenstein, 2000 and
Cohen and Babey, 2012).
Figure 2.2 Norms for Cancer Prevention
Texas Tech University, Janani R. Thapa, August 2014
35
2.4.1 The mediating variable model
A theory that provides a framework for intervention design and how the
intervention works is the mediating variable model (Baranowski et al., 2003). This model
is a cause and effect sequence between an intervention and an outcome. According to the
perspective of mediating variable model, the independent (demographic) variable
influences the mediator variable. The mediating variable in this research are the
categorical outcome variables: health attitude, cancer knowledge and food behavior. The
change in the mediating variables leads to change in anthropometric measurements like
BMI and WC, common desirable outcome variables. The more strongly the mediating
variable is related to the outcome variable the greater the impact of the intervention
(Baranowski et al., 1997 and Baranowski et al., 1998). Changes in obesity status are
expected to occur as a result of changes in mediating variables that influence food and
health behavior decisions. The mediating variable comes from the theories or models
used to understand behavior. The model is shown in Figure 2.3.
Figure 2.3 Mediating Variable Model
Mediating variable
(Health attitudes, food behaviors, cancer knowledge)
Dependent variable (BMI) Independent variable
(Education, SES, age etc.)
Texas Tech University, Janani R. Thapa, August 2014
36
2.5 Methods
2.5.1 Data
Two small communities Muleshoe, TX and Dalhart, TX were selected as project
intervention and control sites respectively, based on their similar demographics.
Muleshoe was selected as the intervention site as this small community met the criteria
for implementing community-based research. A required element in community-based
research is an established and ongoing relationship between key community-based
organizations and the research group (Leung et al., 2004). Analysis was done both under
the assumption that there is no systematic difference due to the pre-treatment variable,
and then when this assumption or no significant difference is relaxed. The average
poverty rate in these communities ranged from 9 to 23% with more than one-third of the
populations being Hispanic. The poverty rate in Dalhart was 9-15%. The population of
Muleshoe was about 4,571 with 17-50% of the population estimated to be Hispanic. The
poverty rate of Muleshoe is 17-23% (TTUHSC, 2010). The project sites are shown in
Figure 2.4.
A multi-tiered behavior reinforcing model was implemented in Muleshoe from
June of 2011 to December of 2012, the intervention site. The data for this section were
drawn from the data collected in five independent surveys: two in the control site and
three in the intervention site. First, a baseline (June of 2011) and second, a post-
intervention survey (June of 2012) was conducted at both the control and the intervention
sites. A third, follow-up survey (January, 2013) was conducted only in the intervention
site to test for sustainability of changes guided by the RE-AIM (Reach, Effectiveness,
Adoption, Implementation, and Maintenance) framework (Glasgow et al., 1999).
Texas Tech University, Janani R. Thapa, August 2014
37
Figure 2.4 Project Intervention and Control Site
There were a grand total of 1,132 respondents in the three surveys. Inclusion
criteria for participation were individuals who were at least 18 years of age, and self-
identified as living in the communities of Muleshoe and Dalhart. They were invited to
participate by mail sent out to a random sample of local population drawn from the local
telephone book. The average show up rate was 22.3%. This random sample was
supplemented by participants invited through flyers and pamphlets distributed in public
areas and printed in local newspapers. The comparison of random mailed participants
(referred as invitees) and the participants recruited through flyers and pamphlets (referred
as volunteers) is shown in Appendix 2.2 through 2.5. Similar recruitment with mail
survey was also done by Crawford et al. (2000). During 2011 survey, there were a total of
382 respondents: 225 from Muleshoe, and 157 from Dalhart. The total participant number
increased to 550 during 2012 survey: 335 from Muleshoe and 215 from Dalhart. Of the
total in Muleshoe and Dalhart, 69 and 39 of them were two year participants respectively.
City of Dalhart (Control)
Hartley and Dallam County
39.4% obesity rate (2011)
6.75% Change (2001 -2009)
City of Muleshoe (Intervention)
Bailey County
41.6% obesity rate (2011)
6.5% Change (2001 -2009)
Texas Tech University, Janani R. Thapa, August 2014
38
During the 2013 survey, there were a total of 204 respondents, all in Muleshoe.
Of the 204 respondents, 102 respondents had participated either in the 2011 or the 2012
survey. The third survey in the intervention site was used for follow-up effects of the
intervention. All the study protocols were approved by the Texas Tech University
Institutional Review Board. The same sets of activities were conducted during the three
surveys. Anthropometric measurement data were collected and the participants completed
a health assessment survey. Detail of survey respondents and their overlap in between
surveys are shown in Figure 2.5.
Figure 2.5 Survey Respondents
Texas Tech University, Janani R. Thapa, August 2014
39
2.5.1.1 Anthropometric measurements
Height, weight, and waist circumference along with demographic information
were taken by trained staff. Participants were given a Project Participation Measurement
Information Form for their reference. Height was measured using a stadiometer
(HM200P PortStad, Portable Stadiometer-WMFS) and weight was measured using a
precision scale (UC-321 Pro FIT Precision Scale, 350 lbs capacity). The stadiometer was
assembled with stabilizer pieces against the wall for the most accurate measurement.
Participants were then asked to remove their shoes and place their heels against the base
of the stadiometer where their height was taken in inches, rounded to the nearest 1/8 inch.
Participants were also asked to remove their jackets and any heavy objects in their
pockets prior to measuring their weight using the precision scale.
Measurements were taken in pounds, rounded to one decimal place. The
participant’s WC was measured using a measure tape (QM2000 Measure Mate – Girth
and Linear Measure Tape CM/IN). We asked participants to lift their shirts to expose
their abdomen region. The measure tape was wrapped around the top of their hip bone,
parallel to the ground (CDC, 2011). For our statistical testing, the measurements were
later converted to meters and kilograms. BMI was computed based on recorded height
and weight. BMI was calculated by dividing the participant’s weight (in kilograms) by
their height (in meters) squared (CDC, 2011). BMI was used to classify individuals into
four main weight categories: underweight (<18.5), ideal (18.5-25), overweight (25-30),
and obese (>30).
Texas Tech University, Janani R. Thapa, August 2014
40
2.5.1.2 The health assessment survey
Self-reported survey questionnaires consisting of two part health assessment
survey were given. The health assessment survey had two parts: 1) Nutrition and Health
Practices Survey, and 2) AMI-HI Fitness Inventory. All the surveys were available in
English and Spanish. A Spanish translator was always available during the surveys. The
first part focused on the participant’s health practices, attitudes and perceptions of cancer
and cancer risk factors and individual dining practices. The demographic questions in the
survey were drafted from the BRFSS 2010 Survey (questions 12.6, 12.7, 12.8, 12.10)
(CDC, BRFSS Questionnaire, 2010). The second part, AIM-HI Fitness Inventory was
created by the American Academy of Family Physicians (AAFP, 2009). This part focused
on the participant’s level of interest in changing food behaviors or their physical activity
levels and dietary intake. The AIM-HI fitness survey was developed and validated by the
American in Motion-Healthy Interventions (AIM-HI) research study, conducted by the
AAFP. All sets of questionnaires were pretested with similar respondents to the intended
survey recipients.
The questionnaire had multiple choice, Likert scale and open ended questions
under the following sub headings:
1. Beliefs regarding cancer - Eight cancer causing factors were identified. Beliefs
regarding overweight and obesity as a cancer causing factor were used as proxy to
changed knowledge.
2. Health practices - Eight questions regarding current health practices with regards to
smoking, snuffing and tanning habits were asked as proxy to cancer prevention behavior.
Texas Tech University, Janani R. Thapa, August 2014
41
3. Socio-demographics - Marital status, number of people in household, education level,
and income range were included. Respondents’ age, gender, ethnicity and language were
collected in a different worksheet along with anthropometric measurements.
4. Activity level - Several indicator questions like number of TV hours watched, yard
work hours etc. were collected.
5. Food behavior - This included several proxies for food behavior including
consumption of high fat food, number of sugary drink and desserts consumed per day,
number of meals eaten in fast food restaurant, number of home prepared meals, and self-
identification of awareness about different food’s nutritional value, self-weight
perception and interest in finding ways to reduce cancer risk.
6. Emotional health - Questions were designed to understand if poor mental and physical
health and stress or depression obstructed usual activity and if participation in some form
of spiritual or cultural activity provided emotional strength to the respondent.
In addition, respondents had an option to select their current attitude and their
degree of openness to change from a close ended multiple choice question on activity
level, food behavior and emotional health. Beliefs, attitude and nutrition knowledge were
considered as mediating or intermediate variables to study the intervention effectiveness
in addition to other direct change variable such as supermarket sales (Lockwood et al.,
2010). Such variables are referred to as psychosocial measures. Psychosocial measures
are considered to be mediating factors for dietary change (MacKinnon and Dwyer, 1993;
Hansen and McNeal, 1996; Baranowski et al., 1997 and Framson et al., 2009). Self-
reported behavioral outcomes have been selected to use for their ease of interpretation
and to reflect an intervention specified goal. They are the most often used basis for
Texas Tech University, Janani R. Thapa, August 2014
42
intervention evaluation (Kristal and Ollberding, 2012).The primary weakness of using the
self-reported behavioral outcome is that the respondents exposed to intervention may bias
their reports of behavior to exaggerate true behavior change (Baranowski et al., 1997 and
Caan et al., 2004). Summary statistics of all categorical variables from 1132 respondents
are shown in Appendix 2.1 and the summary statistics of continuous variables are shown
in the results section.
2.5.1.3 Project participation survey
There was a separate questionnaire for project participation in the post-
intervention and follow-up survey. This questionnaire collected data on the attitude of
program participants towards the project activities. After data collection all participants
were given educational materials regarding the risk factor for cancer including
information about obesity and nutrition.
2.5.1.4 Missing data
Missing data was dealt by imputation or dropping the observation with missing
values as suggested by (Malhotra, 2007). Imputation was done for observations with
missing demographic information like marital status, income education, number of
people in household, number of home prepared meals, and number of meals in a fast food
restaurant. There were 87 questionnaires with missing data during 2011 survey. During
2012 all questionnaire were checked before collection and respondent were asked to fill if
any response were missing. There were only two questionnaires with missing data in
2012. Two questionnaires in 2012 and one questionnaire from 2011 survey was
completely blank. The blank questionnaires were deleted before imputation. Other
Texas Tech University, Janani R. Thapa, August 2014
43
missing data were arbitrary. Missing data imputation was done using SAS (2008), V 9.2,
SAS Institute Inc.
2.5.2 Study designs
As explained in Section 2.5.1, there were respondents who participated in one,
two or three surveys. Due to this, one standard statistical design could not be used in
testing the hypothesis. The statistical design had to be associated with the respondent’s
nature of participation. To accomplish accurate statistical comparison the available data
sets allowed the following three evaluation designs: 1) Experimental two group pretest-
posttest design, 2) One shot case study, and 3) One group pretest-posttest design. These
designs were feasible for the evaluation and statistical comparison of findings according
to Doyle and Ward (2001). The respondent pool was sub-categorized to fit in their
respective study design to allow accurate comparison. The designs are explained in short
as follows:
Texas Tech University, Janani R. Thapa, August 2014
44
1. Experimental: two group pretest-posttest design
This design was used to compare the responses from the intervention and control
site for the sub group of respondents who participated in both the pre- and post-
intervention survey.
2. One-shot case study
This design was used for the sub group of respondents who participated only in
the follow-up survey in the intervention community.
3. One-group pretest-posttest design
This design was used to compare the response of respondents who participated in
both the pre- and post-intervention survey in the intervention community.
Random selection and assigments to
groups
Intervention group
Pretest Intervention Posttest
Random selection and assigments to
groups
Control group
Pretest Intervention Posttest
Intervention Posttest
Pretest Intervention Posttest
Texas Tech University, Janani R. Thapa, August 2014
45
The response from the sub group of respondent who do not fit in the study designs
above are the respondents who participated in only one among the pre-, post-, and follow-
up survey. This group was used to measure the community effect. The community effect
group was used to analyze the effect of project activities outside the health and food
information provided during the survey. This group was also used to isolate the natural
effect by comparing the change between the intervention and the control community
within the community effect group. The first year respondents in the study designs
explained above are also included in the community group for the first survey analysis.
2.5.3 Data analysis
This analysis enables the evaluation of the success of the multi-tiered community-
based approach and analyzes its outcome by computing project impact through changes
in awareness and behavior before and after the project as well as by finding the
intervention effect on cancer knowledge and BMI.
2.5.3.1 Project impact and sustainability through group comparisons
Beliefs, attitude and nutrition knowledge are necessary variables to study
intervention effectiveness. These variables have been used to study the intervention
effectiveness and to conduct impact evaluation of community-based behavior change
interventions for obesity prevention studies (e.g. Sacher et al., 2010a; Nguyen et al., 2012
and Rogers et al., 2013). The variables used in this part of data analysis are:
1. Cancer risk awareness
2. Cancer risk behaviors including nutrition awareness
3. Food behavior and other cancer risk behavior
Texas Tech University, Janani R. Thapa, August 2014
46
The above listed categorical variables were measured in both intervention and
control group during all surveys. The degree to which the group changed their behavior
was compared and analyzed for change and statistical significance. Descriptive statistics,
means and frequency procedures followed by comparison in outcome under each study
design was done. Appropriate test of significance among paired t-test, independent
sample t-test, McNemar’s test and Chi-square test was applied to test the change in the
categorical variable listed above as suitable to each study designs listed in Section 2.5.2.
A paired group comparison was done for paired group within the two group pre-, post-
test design. The p value for categorical variables (variable of interest) for paired group
comparison is from McNemar’s test for proportion for dichotomous response variable. In
cases where the response was not dichotomous variable, the responses were merged to
create dichotomous response.
An important component of the research was to ensure that there is no relapse.
Suitable group comparisons were done to assess the sustainability effect using
comparative analysis from the third year data from the intervention site. A Chi-square test
was done for testing the change in categorical variables in the community group
(community group was described earlier in Section 2.5.2). A Chi-square test for
proportion was done to compare the change between pre- and post-intervention groups
with dichotomous variable used to test sustainability.
Similar methods have been used to find the impact of community-based
interventions in the literature. For example, Goldfinger (2008) and Shaibi et al. (2012)
used a paired t-test for continuous variable for comparing pre-, post-intervention group.
Balcazar (2009a) used a t-test to compare an intervention group with a control group.
Texas Tech University, Janani R. Thapa, August 2014
47
Crawford et al. (2000) used Chi-square test to examine the change in an independent
sample with categorical variables.
2.5.3.2 The simple intervention effect on cancer knowledge score
The method of finding simple intervention effect is finding the difference between
the change in intervention group minus the change in control (or alternative treatment)
group. This is the best and most comprehensible method (Naresh, 2007 and Kristal and
Ollberding, 2012):
( ) ( )
where, is the variable of interest. Subscript and refer respectively to baseline and
follow-up and subscript and refer to intervention and control respectively. Further,
the statistical test of whether or not the intervention effect is different from zero is based
on the standard error of the intervention effect which is defined in the equation below
(Kristal and Ollberding, 2012),
( ( b f)i
i ( b f)c
c)
where, i is the sample size in the intervention group, and c is the sample size in the
control group.
The simple intervention effect was computed while correcting for confounding
variables. The intervention effect is best estimated after adjustment for characteristics that
are associated with outcome variable to overcome randomization errors and increase
statistical power (Kristal and Ollberding, 2012). According to Kristal and Ollberding
(2012) the best approach is to build a regression model with the change from baseline to
follow-up in the outcome measure as the dependent variable. The covariates are the
baseline value of the outcome variable and other confounding variables. A method also
Texas Tech University, Janani R. Thapa, August 2014
48
followed by Kristal et al. (2000) is used to find the simple intervention effect with control
for confounding variables. Analysis of the effects of project intervention were based on
adjusted multiple linear regression model. The dependent variable is the change in cancer
knowledge score. The cancer knowledge score was calculated from the cancer knowledge
questions and serves to quantify cancer risk awareness. Cancer knowledge score is the
sum of affirmative responses to the cancer knowledge question set (see Appendix 2.6)
where there were five questions. Hence, the possible range of cancer knowledge score
was 0 to 5. The regression model consists of an indicator variable in the right arm, which
is the intervention effect. The model used is as follows:
i i 1 i 2 i 3 i i 5 i
i 7 i 8 i 9 i 10 i 11 i i
where, the individual intercept term i, is an unobservable effect from factors outside the
scope of survey and project, ij refers to the change in cancer knowledge score from
before intervention to after intervention for individual in time is dummy for
language (1= english, 0=spanish), is marital status, is cancer survivor dummy
(1=if the respondent was a cancer survivor, 0 otherwise), is dummy for overweight
(1=if the respondent is overweight, 0 otherwise), is the independent variable
representing intervention group (1=intervention, 0=control) and is the variable of interest.
Results for this analysis are given both with and without adjustments for the following
covariates: language, marital status, education, income, age, sex, race, cancer survivor,
and weight status. The computation was done using SAS Enterprise Guide 5.1, SAS
Institute Inc.
Texas Tech University, Janani R. Thapa, August 2014
49
The focus of the project was on changing attitude towards food behavior and the
risk perception towards obesity as a cause of cancer with attitude and perception as
mediating variables. The target variable for overall change by the mediating variables is
BMI, so now we turn to the effect of the intervention on BMI.
2.5.3.3 Effect of intervention on BMI
Body mass index (BMI) was calculated from the data available on height and
weight of respondents from all the survey. The BMI expression explained in Section 2.3,
which is the standard BMI expression for adults, was used to compute BMI from the
available height and weight data. The mean BMI for total respondents during each survey
in control community was compared against mean BMI during each survey in
intervention community. A trend observation was done. BMI comparison was done
separately for male and female respondents.
The overall objective of the project was to improve attitudes and behaviors
towards increased obesity prevention. To find the effect of intervention on BMI, a
multiple linear regression was done. Similar methods have been used to assess the effect
on fruit and vegetable consumption, self-reported, and psychological health of an
introduction of large scale food retailing study in a deprived Scottish community by
Cummins et al. (2005).This analysis enables the evaluation of the success of the multi-
tiered community-based approach by computing the intervention effect on BMI. BMI is
partially determined by a set of covariates, which is defined in the model below:
Texas Tech University, Janani R. Thapa, August 2014
50
ij i 1 ij 2 ij 3 ij ij ij
where, ij refers to the body mass index for individual in time ; is the
demographic variables such as race, gender, age, age2. is the location dummy
(1=intervention and 0=control), includes experimental treatments such as
intervention and year interaction, and includes food and health behavior
variables such as nutrition awareness, nutrition score, satisfaction, activity score etc.. The
individual intercept term i, is an unobservable effect from factors outside the scope of
survey and project.
The selection of variables and derivation of this model come from the literature
on obesity prevention. Demographic variables, satisfaction, activity score, weight
perception are selected based on the citations reported in Goldfinger et al. (2008). Fast
food consumption and physical activity were used to analyze the effect of three year
community-based study on weight control by Crawford et al. (2000). We added the
variables for nutrition awareness (Nutrition awareness) and nutrition score as suggested
by the mediating variable framework explained in Section 2.4.1. The direct outcome
variables are change in the mediating variables (attitude and behavior) to change BMI,
the dependent variable. The explanatory variables are explained in detail in Table 2.1.
The model was tested for model mis-specification using Ramsey’s Reset test using the
AUTOREG functions in SAS. The model was prone to have problems of
heteroscedasticty due to the cross-section and panel nature of data (Greene, 2003).
Breusch-Pagan test was done to test for heteroscedasticty by using the MODEL function
in SAS. All the analysis was done using SAS Enterprise Guide 5.1, SAS Institute Inc.
Texas Tech University, Janani R. Thapa, August 2014
51
Table 2.1 Description of explanatory variables
Hispanic Self-identified race (1=Hispanic, 0=White)
Gender Dummy variable for gender (1=male, 0=female)
Age Age of respondent
Age2 Square of variable age
Nutrition awareness
Self-reported food’s nutritional value awareness (1=at least
moderately aware, 0= not at all or somewhat aware)
Fast food Number of meals at fast food restaurant in the last week
Nutrition score Sum of affirmative response for all food habit question
(Appendix 2.6)
Satisfaction Dummy for self-described satisfaction (1=happy or satisfied,
0=otherwise)
Activity score Sum of affirmative response for all activity question (Appendix
2.7)
Intervention site Control or intervention site (1=intervention site, 0=control site)
Year 2012 Dummy for post-intervention year
Post-intervention Interaction dummy for project intervention
Follow-up Interaction dummy for follow-up response in intervention site
Body weight
perception (BWP)
Dummy (1=perceived weight equals actual weight, 0 if not
equal)
Texas Tech University, Janani R. Thapa, August 2014
52
2.6 Results
2.6.1 Project impact and sustainability through group comparisons
2.6.1.1 Cancer risk awareness
Being aware of the dangers from a particular behavior is essential to changed
behavior. The targeted cancer-risk behavior was primarily obesity, but in addition the
project also focused upon sunburn and tobacco use. The overall knowledge about cancer
risk is shown in Table 2.2 for the intervention and control sites. As shown in Table 2.2,
there was a very high level of knowledge that tobacco use, sunburn and tanning beds are
causes of cancer. There was less knowledge initially that obesity can cause cancer.
Awareness that obesity can cause cancer significantly increased in control and
intervention community during the intervention. The variables were targeted in the
intervention community. However, both intervention and control communities saw
significant improvements in overweight as the cause of cancer. This could be attributed to
the homogenous information provided during the survey to both the control and
intervention community. Awareness on tanning beds as a cause of cancer increased
significantly in the intervention community but decreased significantly in the control
community. Impacts were positive and significant in the intervention community, and
there also was an increasing awareness that sunburn and chewing tobacco are causes of
cancer. There were not comparable positive improvements in awareness outcomes in the
control community for sunburn and chewing tobacco.
Texas Tech University, Janani R. Thapa, August 2014
53
Table 2.2 Comparison of cancer risk awareness in community group analysis
Cancer risk
awareness
indicators
Intervention Control
Pre-
(N=225)
Post-
(N=199) p value
Pre-
(N=157)
Post-
(N=175) p value
Overweight 38% 51% 0.002***
45% 53% 0.038**
Tanning beds 80% 88% 0.004***
93% 86% 0.009***
Sunburn 92% 93% 0.077* 96% 94% 0.075
*
Chewing tobacco 96% 98% 0.043**
97% 97% 0.147
p values are from Chi-square test
*, **
, and ***
significant at 10, 5 and 1% level of significance respectively
As shown in Table 2.3, a second metric on cancer awareness focuses on
individuals who participated in the data collection both before and after the intervention.
They formed the paired group as discussed earlier. This group also showed improvements
in awareness that being overweight and using tanning beds are risk factors for cancer,
significantly in the intervention community.
Table 2.3 Comparison of cancer risk awareness in paired group
Cancer risk awareness indicators
Intervention (N=68) Control (N=39)
Pre- Post- p value Pre- Post- p value
Overweight 32% 51% 0.003***
49% 56% 0.366
Tanning beds 66% 90% 0.000***
87% 87% 1.000
Sunburn 91% 99% 0.059* 95% 92% 0.564
Chewing tobacco 97% 97% 1.000 92% 90% 0.560
p values are from McNemar’s test
*, **
, and ***
significant at 10, 5 and 1% level of significance respectively
Texas Tech University, Janani R. Thapa, August 2014
54
The third year community data from the intervention site was used for
sustainability of the intervention effect. The metrics for focused variables of risk
awareness of obesity and tanning beds were assessed (see Table 2.4). The awareness of
obesity risk remained high six months after the intervention finished. However, the
awareness of tanning bed risk declined significantly. This suggests that the awareness of
cancer risk from obesity has increased significantly and sustainably from the intervention.
However, the awareness in the community from obesity risk (53%) is still much lower
than the awareness of risk for other cancer behaviors (sunburn and tobacco). In addition,
knowledge of cancer risk from tanning beds also appears to be lower and less sustainable.
However, from Table 2.3 and 2.4, awareness on tanning beds as cause of cancer
increased from pre-intervention to follow-up. Therefore, the analysis suggest that both
the obesity and tanning bed cancer risk could be usefully targeted in future public
education as these are below desired high levels of awareness.
Table 2.4 Sustainability of cancer risk awareness in the intervention community
Cancer risk awareness indicators
Intervention community
Post-
(N=199)
Follow-up
(N=91) p value
Believe the use of tanning beds can cause cancer 88% 75% 0.023**
Believe being overweight can cause cancer 51% 52% 0.243
p values are from Chi-square test
*, **
, and ***
significant at 10, 5 and 1% level of significance respectively
Texas Tech University, Janani R. Thapa, August 2014
55
2.6.1.2 Cancer risk behavior including nutrition awareness
In addition to cancer risk awareness, changes in behavior to reduce or avoid
cancer risk were targeted for obesity and to a lesser extent to sunburn and tobacco use.
Actual obesity, sunburns, and tobacco use are considered to be risk outcomes. Targeted
behaviors for improvement were nutrition awareness for obesity, using sunscreen and
wearing appropriate head covering for sunburn prevention.
Table 2.5 shows the project impact to the community. Nutrition awareness
increased significantly across both communities, wearing cowboy hat increased
significantly only in the intervention community. However, use of sunscreen decreased in
both communities. The decrease was significant in the intervention community. The
reluctance to use of sunscreen has been assessed during the second and third survey in the
intervention community.
Table 2.5 Comparison of project impact indicator variables in the community
Project impact
indicators
Intervention Control
Pre-
(N=225)
Post-
(N=199) p value
Pre-
(N=157)
Post-
(N=175) p value
Nutrition Awareness:
very or moderately
aware
44% 55% 0.082* 31% 49%
0.004***
Using sunscreen: always
or most of the time 25% 23% 0.066
* 30% 26% 0.104
Wear Cowboy hat:
always or most of the
time
18% 29% 0.003***
20% 22% 0.121
p values are from Chi-square test
*, **
, and ***
significant at 10, 5 and 1% level of significance respectively
Table 2.6 shows the project impact on the paired group. There appears some
increase in cowboy hat use and nutrition awareness however the effect were not
Texas Tech University, Janani R. Thapa, August 2014
56
significant. This insignificant change in paired group respondents could be better
understood by looking at the demographic differences between respondents in paired
group against those in the community group. The reluctance to sunscreen use as seen
from Table 2.5 could be accounted for the following reasons: 29.86% of the total
respondents reported not to be outside for long to have identified the need to use
sunscreen, 24.26% reported availability of sunscreen as the primary reason for not using
sunscreen, 10% reported time in using sunscreen as an issue and 8% reported costs of
sunscreen as an issue.
Table 2.6 Comparison of project impact indicator variables in the paired group
Project impact indicators
Intervention (N=68) Control (N=39)
Pre- Post- p value Pre- Post- p value
Nutrition Awareness:
very or moderately
aware 53% 63% 0.194 79% 74% 0.366
Wear sunscreen: always
or most of the time 32% 31% 0.782 36% 38% 0.564
Wear Cowboy hat:
always or most of the
time 24% 29% 0.285 21% 21% 1.000
p values are from McNemar’s test
*, **
, and ***
significant at 10, 5 and 1% level of significance respectively
2.6.1.3 Food behavior and other cancer risk behavior
An evaluation of changed in food behavior was done by using the amount/
frequency of fruits and vegetables consumed and changed attitude towards own dietary
habit as key variables. Sunburn was measured by the frequency of getting severely
sunburned because sunburn is the greatest risk factor for getting melanoma skin cancer.
Table 2.7 shows the frequency of response from total respondents, community group and
paired group in control and intervention community during pre-intervention and post-
intervention period. For the intervention community, the percentage in Table 2.7 is
Texas Tech University, Janani R. Thapa, August 2014
57
calculated in N=225 for total and community group during pre-intervention. And, N= 332
and N=199 for total and community group respectively during post intervention. N= 69
for paired group. For the control community, N=157 for total and community group
during pre-intervention. And N= 214 and N=175 for total and community group
respectively during post intervention. N= 39 for paired group. This measure was
unchanged in both the intervention and control communities. Tobacco use was measured
by the percentage of tobacco smokers and those who use smokeless tobacco. These
measures appeared to be unchanged by the project (Table 2.7).
Texas Tech University, Janani R. Thapa, August 2014
58
Table 2.7 Food behavior and other cancer risk behavior
Response Intervention Control
Pre- Post- Pre- Post-
I'm eating healthy at this time
Total 34% 34% 38% 31%
Community 34% 26% 38% 29%
Paired group 45% 54% 38% 44%
3-4 or 5 or more servings of fruits and vegetables
Total 44% 42% 45% 43%
Community 44% 40% 45% 41%
Paired group 49% 51% 59% 54%
Smoking
Total 13% 13% 13% 15%
Community 13% 12% 13% 15%
Paired group 1% 15% 10% 15%
Use smokeless tobacco (% doing)
Total 4% 3% 6% 2%
Community 4% 4% 6% 2%
Paired group 3% 1% 3% 3%
Get sunburned when outside 30 minutes (always or most of the time)
Total 17% 16% 11% 18%
Community 17% 16% 11% 18%
Paired group 13% 15% 13% 18%
Use tanning beds
Total 26% 24% 32% 23%
Community 26% 28% 32% 23%
Paired group 16% 19% 26% 21%
Texas Tech University, Janani R. Thapa, August 2014
59
2.6.2 The simple intervention effect on cancer knowledge score
The cancer knowledge score (CKS) is the sum of affirmative response to cancer
knowledge questions as explained in Section 2.5.3.2. The score ranged from 0 to 5. The
mean score was 3.853 ±0.935 and 4.368±0.731 before and after the intervention
respectively in the intervention community and the score was 4.230±0.872 and
4.179±1.233 before and after the intervention in the control community (see Table 2.9).
The score range was in the intervention community. So the project was successful in
increasing the score from a lower level to a higher level compared to the score in the
control community. The multiple linear regression output for analyzing the simple
intervention effect with control for the confounding variables is shown in Table 2.8. As
shown in Table 2.8 the intervention had significant impact on cancer knowledge score.
Texas Tech University, Janani R. Thapa, August 2014
60
Table 2.8 Statistical relationship on cancer knowledge score (CKS)
Independent variables Coefficient p value
Intercept 0.490 0.687
Before intervention CKS -0.474***
<.001
English language dummy 0.199 0.474
Marital status -0.106 0.119
Education 0.104 0.286
Income -0.017 0.828
Age 0.006 0.174
Female(dummy) 0.106 0.555
Race 0.080 0.335
Cancer survivor (dummy) 0.035 0.879
Overweight (dummy) 0.243 0.202
Interest in reducing cancer risk (dummy) 0.392 0.634
Intervention (dummy) 0.528***
0.004
R square = 0.372
MSE = 0.793
*, **
, and ***
significant at 10, 5 and 1% level of significance respectively
Texas Tech University, Janani R. Thapa, August 2014
61
Further, Table 2.9 shows the pre- and post-intervention cancer knowledge score
for the intervention and control community. Table 2.9 also shows the raw and adjusted
intervention effect on the cancer knowledge score calculated from the regression model.
The intervention effect was significant. The score was higher by 0.530 point in the
intervention community. Hence, the approach was successful in increasing knowledge of
cancer risk factors. A multi-tiered community-based participatory approach in obesity
prevention efforts shows promise for changing cancer risk awareness of the community.
Table 2.9 Effect of intervention on cancer knowledge score (N=106)
Cancer knowledge score N Pre-intervention Post-intervention
Intervention (x ± SD) 69 3.853 ± 0.935 4.368 ± 0.731
Control (x ± SD) 38 4.230 ± 0.872 4.179 ± 1.233
Intervention effect (x ± SE)
Raw (p value)
0.040 ± 0.006 (<.001)
Adjusted (p value)
0.530 ± 0.179 (0.004)
Texas Tech University, Janani R. Thapa, August 2014
62
2.6.3 Effect of intervention on BMI
Obesity was measured by body mass index (BMI). Along with BMI, waist
circumference (WC) was measured for all respondents. The mean BMI and WC for the
respondents from total respondents and community group during the three surveys in
intervention and two surveys in the control community are shown in Table 2.10. As seen
in the table the total respondents form the community group in the pre-intervention
survey. Further, paired group comparison was not done for BMI because BMI was not
the direct intervention outcome variable unlike behavior and attitude. We were interested
in seeing the overall effect rather than BMI change in paired group. Research targeting
BMI has concluded that identification of modifiable influences can lead to long-term
weight loss maintenance (Svetkey et al., 2012).
Texas Tech University, Janani R. Thapa, August 2014
63
Table 2.10 Body mass index (BMI) and waist circumference (WC) outcomes
Intervention community Pre- Post- Follow-up
Mean WC Male Female Male Female Male Female
Total
42.15 39.81
42.04 40.15 40.77 39.69
Community group 42.12 40.09 39.49 38.98
Mean BMI Male Female Male Female Male Female
Total
30.5 31.76
30.01 31.08 27.99 29.52
Community group 30.53 31.43 27.12 30.24
Control community Pre- Post-
Mean WC Male Female Male Female
Total
41.91 37.46
43.31 38.73
Community group 43.21 38.68
Mean BMI Male Female Male Female
Total
29.97 29.97
30.31 28.65
Community group 30.31 28.79
Figure 2.6a and 2.6b shows the mean BMI trend in the total male and female
respondents respectively in the intervention and control community. These measures
show a small trend towards the desired lower levels in the intervention community
against the control for male respondents (Figure 2.6a). A similar plot for female
respondents shows parallel decrease in BMI in both the control and the intervention
community (Figure 2.6b). This suggests that the intervention may be having an effect.
However, the change was insignificant.
Texas Tech University, Janani R. Thapa, August 2014
64
Figure 2.6a Mean BMI Trend in Total Male Respondents
Figure 2.6b Mean BMI Trend in Total Female Respondents
Texas Tech University, Janani R. Thapa, August 2014
65
Similarly, Figure 2.7a and 2.7b shows the mean BMI trend in the community
male and female respondents’ group respectively in the intervention and control
community. These measures also show a small trend towards the desired lower levels in
the intervention community against the control for male respondents (Figure 2.7a). A
similar plot for female respondents shows parallel decrease in BMI in both the control
and the intervention community (Figure 2.7b). These findings point to the differential
reactiveness to project intervention by gender. Patterns of weight change is different
between the genders (Kimokoti et al., 2012).
Texas Tech University, Janani R. Thapa, August 2014
66
Figure 2.7a Mean BMI Trend in Male Community Respondents’ Group
Figure 2.7b Mean BMI Trend in Female Community Respondents’ Group
Texas Tech University, Janani R. Thapa, August 2014
67
2.6.3.1 Multiple linear regression
A multiple linear regression was done to find the effect of intervention on BMI as
explained in section 2.5.3.3. The variables have been explained in Table 2.1. Table 2.11
shows the summary statistics of the variable used in this regression analysis.
Table 2.11 Summary statistics of dependent and explanatory variables
Variable Mean Std. Dev. Minimum Maximum
BMI 30.41 11.67 13.75 60.39
Hispanic 0.45 0.50 0 1
Gender 0.34 0.47 0 1
Age 52.11 18.57 18 97
Age2 3059.45 1975.96 324 9409
Nutrition awareness 0.56 0.50 0 1
Fast food 1.44 0.06 0 21
Nutrition score 3.31 1.89 0 10
Satisfaction 0.68 0.64 0 4
Activity score 1.08 1.00 0 4
Intervention site 1.32 0.47 1 2
Year 2012 0.48 0.50 0 1
Post-intervention 0.30 0.46 0 1
Follow-up 0.18 0.39 0 1
BWP 0.46 0.50 0 1
Texas Tech University, Janani R. Thapa, August 2014
68
The model was tested for mis-specification and heteroscedasticity by using the
Ramsey RESET test and the Breusch-Pagan test respectively. The Ramsey RESET test
statistic was insignificant. However, the Breusch-Pagan statistic (38.12) was highly
significant (p value = 0.0005) and the null hypothesis of no heteroscedasticty was
rejected. This implied that the standard errors of the parameter estimates could be
incorrect and thus any inference derived from them could be misleading. Therefore, the
model was modified to yield accurate inference as discussed below.
When the variables male, age, age2, nutrition score and fast food were dropped the
new model did not show heteroscedasticity. The Breusch-Pagan test statistic (11.16) was
insignificant (p value = 0.265), and the Ramsey RESET test statistic (0.739) was also
insignificant (p value = 0.390). Therefore the standard errors of the parameter estimates
from the new model are correct and thus the inferences derived from them are not
misleading. However, the variables that were dropped will be included in the model by
following appropriate methods to treat heteroscedasticity. The regression coefficients
from the multiple linear regressions on the new model are shown in Table 2.12 with BMI
as dependent variable.
The coefficient for dummy variable Hispanic was positive and significant,
suggesting that the BMI for Hispanic would be higher by 1.839 compared to the Whites.
Other races were dropped before analysis of this model (6 respondents). Similarly,
dummy variables follow-up and BWP were significant and negative. Negative and
significant coefficient for follow-up suggested if the respondent participated in the
project for longer the BMI was significantly lower. Negative and significant for BWP
suggested individuals who had correct self-weight perception (Table 2.13) would have
Texas Tech University, Janani R. Thapa, August 2014
69
significantly lower BMI by 2.453. Weight perception can play an important role in food
behavior and health attitude (ter Bogt et al., 2006). Further activity score and nutrition
awareness was also negatively significant. One unit increase in activity score decreased
BMI by 0.665, and nutrition awareness decreased BMI by 1.193.
The variable of interest in this regression to isolate the intervention effect
controlling for other variables was post-intervention variable and follow-up variable. The
post-intervention variable was positively insignificant. But the significant follow-up
variable despite the insignificant intervention site and post-intervention interaction year
variable is suggestive that having increased access to healthy living messages and
exposure to a supermarket environment that promote healthy living can have a negative
relation with respondents’ BMI with longer exposure to the project activities. This last
result might suggest the potential effect if the intervention went beyond one year.
Therefore, negatively significant coefficients for variable follow-up suggest that a longer
term project intervention will have the required effect on BMI. Also the negative
significant relation of BWP with BMI led us to look at BWP more closely.
Texas Tech University, Janani R. Thapa, August 2014
70
Table 2.12 Statistical relationship on BMI
Independent variables Coefficient Std. error p value
Intercept 32.607***
1.213 <.001
Hispanic 1.839***
0.433 <.001
Nutrition awareness -1.193***
0.418 0.004
Satisfaction -0.308 0.326 0.346
Activity score -0.665***
0.207 0.001
Intervention site 0.062 0.749 0.935
Year 2012 -0.959 0.753 0.203
Post-intervention 0.590 0.958 0.538
Follow-up -2.052***
0.662 0.002
BWP -2.453***
0.425 <.001
R square = 0.083
MSE = 44.02
*, **
, and ***
significant at 10, 5 and 1% level of significance respectively
Texas Tech University, Janani R. Thapa, August 2014
71
Table 2.13 shows the perception of weight status versus the actual weight for all
the respondents across all surveys. As shown in Table 2.13, 39% of overweight
respondent perceived their weight to be normal and 65% of obese respondent perceived
their weight to be overweight and 9% obese perceived their weight to be normal. Self-
perception has been explained in relation to cognitive dissonance1 in psychology
literature. While some literature cite cognitive dissonance as a hindrance to obesity
prevention (Stellefson, 2006), others have used it in treating obesity stigma (Ciao and
Latner, 2011).
Table 2.13 Self-perception of body weight (N=1067)
Actual weight
Frequency of perceived weight within each weight category
Underweight Normal Overweight Obese Not-sure Total (N)
Underweight 27% 60% 7% 0% 7% 15
Normal 8% 64% 4% 0% 23% 227
Overweight 1% 39% 32% 0% 28% 347
Obese 2% 9% 65% 11% 13% 478
1 Cognitive dissonance refers to a situation involving conflicting attitudes, beliefs and
behaviors (Festinger, 1962).
Texas Tech University, Janani R. Thapa, August 2014
72
2.7 Conclusions
This paper evaluates the project outcomes of a community-based multi-tiered
behavior change intervention for obesity prevention based on cancer risk awareness and
cancer risk behavior including nutrition awareness. Intervention effects on cancer
knowledge score and the project impact on BMI were calculated. Mean and frequency
comparison of self-reported categorical variables of interest such as cancer risk
awareness, cancer risk behavior including nutrition awareness and food behavior showed
significant improvement against the null hypothesis of no difference before and after the
intervention and separately between the control and intervention community.
The results suggest that the intervention was successful in significantly and
sustainably changing the attitude toward unhealthy weight status as risk for cancer.
However, the awareness in the community from obesity risk was low compared to
awareness for other cancer risk behaviors such as smoking. The analysis suggest that both
the obesity and tanning bed cancer risk could be usefully targeted in future public
education as these are below high levels of awareness.
The paper also reports a significant project impact on cancer knowledge. The
intervention increased the cancer knowledge score by 0.53 points. Further a regression on
BMI showed the required direction of variables in affecting BMI such as activity score,
nutrition awareness etc. If the respondent had participated in the follow-up survey the
BMI would significantly decrease. These findings in the effect on BMI over time with
increased nutrition and cancer risk awareness shows promise using a longer term
intervention. Therefore, increased food nutrition awareness should lead to reduced
obesity over time as suggested by the regression analysis of obesity risk factors on BMI.
Texas Tech University, Janani R. Thapa, August 2014
73
Overall, this study adds to evidence that shows community-based interventions such as
these have strong potential in bringing long term change in obesity and cancer inducing
behaviors and BMI.
Texas Tech University, Janani R. Thapa, August 2014
74
Appendix
Appendix 2.1 Frequency of total response
What are your beliefs regarding cancer?
Do you believe the following can
cause cancer? Yes No
Don't
know Missing
Tap water 17% 56% 26% 1%
Tanning bed 84% 5% 11% 1%
Sunburn 92% 3% 5% 1%
Over weight 48% 20% 32% 1%
Alcohol 56% 14% 30% 1%
Chewing tobacco 95% 2% 3% 1%
Smoking tobacco 97% 1% 2% 1%
Caffeine 24% 30% 45% 1%
What are some of your health practices?
Do you currently smoke cigarettes? 15% 85%
1%
Do you currently use smokeless
tobacco products? 3% 96%
1%
Have you or anyone in your family
ever used a tanning bed? 23% 75%
1%
Marital status
Married 64%
Divorced 10%
Widowed 9%
Separated 4%
Never married 12%
Missing 1%
Education
Elementary or some high school 22%
High school graduate 37%
Some college or technical school 22%
College graduate 19%
Missing 1%
Income
Less than $20,000 37%
$20,000 to $35,000 25%
$35,000 to $50,000 13%
$50,000 to $75,000 13%
over $75,000 12%
Missing 1%
Texas Tech University, Janani R. Thapa, August 2014
75
Self-reporting on nutrition awareness
Not at all 5%
Somewhat 38%
Moderately 36%
Very 19%
Missing 1%
Wear sunscreen when outside for more than 30 minutes
Always 10%
Most of the time 17%
Some of the time 18%
Occasionally 16%
Never 38%
Missing 1%
Wear a cowboy or broad brimmed hat when outside for more than 30 minutes
Always 10%
Most of the time 13%
Some of the time 15%
Occasionally 12%
Never 48%
Missing 1%
When outside for more than 60 minutes how often do you get sunburn?
Always 8%
Most of the time 8%
Some of the time 22%
Occasionally 32%
Never 29%
Missing 1%
Self-weight perception
Underweight 3%
Healthy weight 31%
Over weight 39%
Obese 5%
Not sure 20%
Missing 1%
Interest in finding ways to reduce cancer
Very interested 73%
Somewhat interested 24%
Not interested 2%
Missing 1%
Texas Tech University, Janani R. Thapa, August 2014
76
Appendix 2.2 Comparing invitee’s vs. volunteer’s response
Response item Invitee Volunteer p value
Tell us about yourself
What is your marital status?
Married 60% 67% 0.472
Divorced 16% 10% 0.364
Widowed 12% 7% 0.371
Separated 0% 1% 0.662
Never Married 12% 15% 0.683
What is the highest grade or year of school you completed?
Elementary 8% 12% 0.553
High school graduate 40% 34% 0.570
Some college or technical school 32% 26% 0.518
College graduate 20% 28% 0.406
What is your annual household income from all sources?
Less than $20,000 20% 22% 0.826
$20,000 to $35,000 32% 21% 0.239
$35,000 -$ 50,000 8% 17% 0.238
$50,000 - $75000 16% 18% 0.794
over $75000 24% 21% 0.756
Gender
Male 40% 39% 0.955
Female 60% 61% 0.955
Race
White 63% 80% 0.042
Hispanic 17% 9% 0.189
Black American 3% 1% 0.480
Pacific Islander 0% 0%
India/Alaskan 0% 1% 0.640
Multi-racial 17% 9% 0.189
Texas Tech University, Janani R. Thapa, August 2014
77
Appendix 2.3 Comparing response in intervention site in baseline survey
Response item Invitee Volunteer p value
Tell us about yourself
What is your marital status?
Married 58% 68% 0.211
Divorced 13% 7% 0.158
Widowed 9% 9% 0.883
Separated 8% 5% 0.534
Never Married 11% 11% 0.966
What is the highest grade or year of school you completed?
Elementary 30% 26% 0.522
High school graduate 34% 37% 0.703
Some college or technical school 25% 20% 0.529
College graduate 11% 17% 0.323
What is your annual household income from all sources?
Less than $20,000 53% 48% 0.535
$20,000 to $35,000 17% 23% 0.324
$35,000 -$ 50,000 13% 10% 0.502
$50,000 - $75000 13% 12% 0.859
over $75000 4% 6% 0.469
Gender
Male 26% 29% 0.691
Female 74% 71% 0.691
Race
White 30% 24% 0.274
Hispanic 35% 38% 0.555
Black American 0% 0% NA
Pacific Islander 0% 0% NA
India/Alaskan 1% 0% 0.065
Multi-racial 35% 38% 0.555
Texas Tech University, Janani R. Thapa, August 2014
78
Appendix 2.4 Comparing response from intervention site in post-intervention survey
Response item Invitee Volunteer p value
Tell us about yourself
What is your marital status?
Married 69% 62% 0.249
Divorced 8% 14% 0.205
Widowed 15% 4% 0.002
Separated 1% 6% 0.103
Never Married 6% 14% 0.067
What is the highest grade or year of school you completed?
Elementary 18% 25% 0.259
High school graduate 33% 36% 0.673
Some college or technical school 19% 23% 0.531
College graduate 29% 16% 0.019
What is your annual household income from all sources?
Less than $20,000 21% 35% 0.026
$20,000 to $35,000 19% 30% 0.090
$35,000 -$ 50,000 25% 14% 0.027
$50,000 - $75000 10% 14% 0.341
over $75000 25% 7% 0.000
Gender
Male 40% 35% 0.389
Female 60% 65% 0.389
Race
White 79% 40% 0.000
Hispanic 19% 58% 0.000
Black American 0% 1% 0.538
Pacific Islander 0% 0% NA
India/Alaskan 0% 0% NA
Multi-racial 1% 2% 0.709
Texas Tech University, Janani R. Thapa, August 2014
79
Appendix 2.5 Comparing response from control site in post-intervention survey
Response item Invitees Volunteers p value
Tell us about yourself
What is your marital status?
Married 39% 78% 0.000
Divorced 14% 7% 0.177
Widowed 18% 1% 0.000
Separated 7% 4% 0.472
Never Married 21% 9% 0.067
What is the highest grade or year of school you completed?
Elementary 29% 23% 0.524
High school graduate 32% 36% 0.709
Some college or technical school 29% 27% 0.866
College graduate 11% 14% 0.623
What is your annual household income from all sources?
Less than $20,000 39% 23% 0.070
$20,000 to $35,000 29% 36% 0.460
$35,000 -$ 50,000 7% 14% 0.350
$50,000 - $75000 18% 13% 0.478
over $75000 7% 15% 0.275
Gender
Male 29% 36% 0.460
Female 71% 64% 0.460
Race
White 61% 41% 0.048
Hispanic 32% 55% 0.028
Black American 0% 1% 0.663
Pacific Islander 0% 0% NA
India/Alaskan 0% 1% 0.536
Multiracial 7% 3% 0.235
Texas Tech University, Janani R. Thapa, August 2014
80
Appendix 2.6 Cancer knowledge question set
What are your beliefs regarding cancer?
Please select the one choice that best describes you
1. Do you believe the use of tanning beds can cause cancer?
a. Yes b. No c. Don’t know
2. Do you believe getting sunburned can cause cancer?
a. Yes b. No c. Don’t know
3. Do you believe being overweight can cause cancer?
a. Yes b. No c. Don’t know
4. Do you believe chewing tobacco/using snuff can cause cancer?
a. Yes b. No c. Don’t know
5. Do you believe smoking tobacco products can cause cancer?
a. Yes b. No c. Don’t know
Texas Tech University, Janani R. Thapa, August 2014
81
Appendix 2.7 Food habit question set
How well do you eat?
I'm eating healthy at this time.
I'm ready to make some changes to eat healthier.
I'm not sure …… but I'm ready to talk about it.
I'm not interested in changing the way I eat at this time.
How many servings of fruits or vegetables do you eat each day?
5 or more
3 to 4
2 or less
How many servings of whole grain do you eat each day?
3 or more
2
1 or less
How many times a week do you eat lean protein like chicken, turkey, fish, tofu or beans?
6 or more
3 to 5
2 or less
How many times a week do you eat high fat foods like fried food, pastries or chips?
1 or less
2 to 3
4 or more
How many times a week do you eat fast food meals or snacks?
1 or less
2 to 3
4 or more
How much margarine, butter or meat fat (lard) do you use?
very little
Some
a lot
How many sugary drinks do you drink each day?
None
1 to 2
3 or more
How many times a week do you eat desserts or other sweets?
3 or less
4 to 6
7 or more
How often do you eat when not hungry, e.g. out of habit or for emotional reasons?
Rarely
Sometimes
Often
All the time
Texas Tech University, Janani R. Thapa, August 2014
82
Appendix 2.8 Physical activity question set
How active are you?
I'm physically active don't need help to be more active.
I'm ready to get more active and would like help.
I'm not sure but I'm ready to talk about it.
I'm not very active and not interested in being more active at this time.
How many hours each day do you spend watching TV or videos or on the computer?
Less than 1
1 to 2
more than 2
How many times a week do you do yard or house work or duties on the job that cause
you to work up a sweat?
4 or more
1 to 3
less than 1
How many times a week do you get out for a brisk walk for 10 minutes or more?
4 or more
1 to 3
less than 1
How many times a week do you participate in sports or an exercise program?
4 or more
1 to 3
less than 1
Texas Tech University, Janani R. Thapa, August 2014
83
CHAPTER III
THE EFFECT ON SUPERMARKET FOOD PURCHASES FROM POINT OF
SALE NUDGES WITH COMMUNITY REINFORCEMENT
Abstract
A project was designed with a supermarket in a small community to provide point of sale
signs or nudges. These signs encouraged consumers to purchase healthier foods in main
supermarket categories accompanied, in the intervention community and store, by overall
healthy choice signs and messages to increase the sale of produce and other healthier
food items. The community education reinforcement program included newspaper
articles, public service announcements on television and in-person healthy eating classes.
Relative to the control store, the program had a statistically significant impact on
increasing purchases within the grain pasta and sauce category among the nudged item’s
categories, and a significant increase in the purchase of green leafy produce category. An
overall increase in the purchase of several nudged item’s categories and produce
categories shows promise in supermarket based community interventions for behavior
change. However, the lack of significant increase in purchases across all categories shows
the challenges of changing food purchase behavior.
Key words: supermarket, nudge, community reinforcement, obesity, food choice
Texas Tech University, Janani R. Thapa, August 2014
84
3.1 Introduction
The Dietary Guidelines for Americans 2010 recommend stronger environmental
strategies for improving the population’s eating practices, including interventions to
influence food purchasing behaviors in stores (USDA-USDHH, 2010). A number of
research projects have focused on the importance of supermarket accessibility in relation
to obesity and obesity related health problems (Spence et al., 2009). However, very few
research has focused on using supermarkets to promote health despite the fact that
supermarkets play an important role in food purchasing (Glanz and Mullis, 1988). An
average consumer makes on an average 2.2 trips per week to the supermarket (FMI,
2011). Supermarket can play a vital role in community-based participatory obesity
prevention efforts. A key recommendation is to use the supermarket in healthy diet
promotion by linking supermarket profits with healthy diets and by making healthy foods
more appealing and available (RWJF, 2011).
The use of supermarkets as a potential site for effective consumer education was
researched substantially in the late 1980s and early 1990s (e.g. Ernst et al., 1986 and
Scott et al., 1991). Light et al. (1989) reported on a feasibility test of the supermarket as a
site for consumer nutrition education program “Eat for Health.” This was a joint research
study by the National Cancer Institute (NCI) and Giant Food Inc., a regional supermarket
chain in the Washington-Baltimore area. Despite the effectiveness of supermarket based
programs like “Eat for Health” (e.g. Scott et al., 1991 and Song et al., 2009) there is little
use of supermarket interventions to prevent and reverse the trends in obesity despite the
potential. Also, a systematic review by Escaron et al. (2013) reported overall
effectiveness from grocery interventions while showing a need for more rigorous testing
Texas Tech University, Janani R. Thapa, August 2014
85
of interventions. Additionally, no other supermarket intervention that specifically
employs point of sales signs (referred hereafter as shelf talker) can be found after Light
et al. (1989) and Patterson et al. (1992).
While obesity is reaching epidemic levels, several studies on community-based
models have shown success in reversing the obesity trend and in preventing disease (e.g.
Brownson et al., 1996; Chou et al., 1998 and Allen et al., 2013). Other studies like Baker
and Brownson (1998), Merzel and D’Afflitti (2003), and Corda et al. (2010) emphasized
community-based programs which encompass multiple interventions as the model with
best potential to achieve behavior change that will reduce person’s health risk. This
research trend has led to advocacy for community-based efforts by such organization as
the American Cancer Society (ACS, 2012).
One approach less used in grocery stores, also suggested by Gittelsohn and Lee
(2013), is to coordinate the grocery store interventions with community reinforcement to
encourage awareness and to hopefully shift purchase to healthier products. The design of
the current program was to develop a community-based cancer risk reduction program. It
was intended to develop an effective feasible program that could be adopted relatively
easily by small communities but could also be applicable in a broader context. We have
described the model in McCool et al. (2013). The project was implemented in a major
supermarket in a small community. The store and the community was a good setting to
implement the intervention due to high rates of obesity in the community and because the
community was small the message over the media regarding healthy eating habit could be
relatively easily and inexpensively controlled.
Texas Tech University, Janani R. Thapa, August 2014
86
The project model focused on incorporating obesity risk education in the local
supermarket. The supermarket focus was designed with an objective to encourage the
development of a community environment supportive of lifestyles that could reduce
residents risk for obesity and cancer. Since the local supermarket is the largest source of
daily food needs it can be an excellent source of information and reinforcement for
healthy behavior. The supermarket was used for following purposes:
to conduct healthy food demonstrations,
to put shelf talkers on comparatively healthy food product,
as an active site of information flow by using posters regarding healthy
eating were placed in the store, and
to access food purchase data to evaluate the effect of the project on food
purchase behavior.
The goal of this essay is to conduct an outcome evaluation by measuring the
achievement of program objectives. The specific objectives of the analysis are to see if
the point of sale nudges and supermarket health promotion intervention could increase
the consumption of relatively healthier items and fresh fruits and vegetables. The focus is
on quantitative outcomes and overall on whether the supermarket intervention was
effective in changing food purchase behavior. The objective was addressed by using
supermarket sales data. The use of supermarket sales data have been recommended also
by Tin et al. (2007) from a review of 22 studies that used supermarket sales data to
supplement food and nutrition monitoring methods.
Texas Tech University, Janani R. Thapa, August 2014
87
3.2 Review of literature
3.2.1 Risk factors for obesity
The causes of obesity are complex. There is also a disparity in obesity incidence
rates among population groups. According to the American Cancer Society disparities
predominantly arise from inequities in work, wealth, income, education, housing, and
overall standard of living, as well as social barriers to high-quality cancer prevention,
early detection, and treatment services (ACS, 2012). There appears to be little difference
in the cancer incidence and mortality rates of rural and urban populations in the United
States. But, evidence suggests that cancer tends to be diagnosed at a more advanced stage
among rural populations (Monroe et al., 1992; Gosschalk and Carozza, 2003; Philips et
al., 2011 and Gong et al., 2012). The American Cancer Society also suggested that
patients from ethnic minorities are substantially more likely to be diagnosed with cancer
at a later stage when treatment can be more extensive and more costly (ACS, 2012).
Access to food is another factor in being able to select healthy foods. Kumanyika
et al. (2008) and Wang and Beydoun (2007) have pointed to income limitations as well as
the accessibility of chain supermarkets to limit a persons’ ability to acquire foods of
higher nutrition value. The relative costs of foods with high and low nutrient density and
caloric values may be another factor contributing to the unhealthy choice. Several
research studies (Drewnowski, 2003; Drewnowski and Specter, 2004; Drewnowski and
Darmon, 2005; Ard et al., 2007 and Maillot et al., 2007) suggest that the current market
prices encourage unhealthy eating. Products with high fat and sugar and low in other
nutrients are relatively less expensive than foods such as fruits and vegetables and whole
grain products.
Texas Tech University, Janani R. Thapa, August 2014
88
3.2.2 Supermarket interventions
A recent report published by the Robert Wood Johnson Foundation (RWJF, 2011)
has emphasized harnessing the power of supermarkets to help reverse childhood obesity.
The report was developed on the following two guidelines:
1. “Marketing strategies to encourage healthy eating must improve the bottom
line, or at least be revenue neutral for food retailers and manufacturers, and
2. Public health researchers and food retailers and manufacturers should work
together to study how grocery store environments and marketing strategies
affect shoppers’ purchases and preferences.”
Light et al. (1989) reported a feasibility test of supermarket as a site for consumer
nutrition education program “Eat for Health”. The reported supermarket nutrition
intervention was built on previously successful collaborative experiences between
Federal agencies and supermarket and on the experience of other researchers who had
conducted point-of-purchase studies. The added components of the “Eat for health”
program were the scope of the project, its length, its extensive advertising, and the scale
and depth of the evaluation. The details of the method and analysis of the impact of the
NCI-Giant Food Eat for Health Study can also be found in (Patterson et al., 1992). Other
studies of health promotion through supermarket intervention have followed though no
publicly reported efforts have used point of sale signs to promote healthy food habit since
Light et al. (1989).
Escaron et al. (2013) excellently summarized the studies published from 1978
through 2012 on supermarket and grocery store interventions (N=33) to promote
healthful eating. This paper identified 58 articles and characterized 38 supermarket
Texas Tech University, Janani R. Thapa, August 2014
89
interventions. Some other studies that were not included in Escaron et al. (2013) also
studied supermarket intervention. Two cross-sectional time series models, estimated with
a variance components procedure, showed the effectiveness of coupons, assuming habit
and non-habit persistence along with consumer demographics (Ward and Davis, 1978).
Similarly, Cummins et al. (2005) reported evidence for an improvement in psychological
health of those directly engaged with the intervention although a net intervention effect
on fruit and vegetable consumption was not found.
3.2.3 Use of scanner data
An evaluation of the feasibility and applicability of supermarket sales data in
population food and nutrition monitoring has been done by Tin et al. (2007). In this
study, eighteen studies that collected supermarket sales data for various population food
and nutrition monitoring purposes and four feasibility studies were reviewed. The
findings supported the feasibility of using supermarket sales data to monitor population
food purchasing patterns. Further, the study showed that it is possible to use various kinds
of sales data (directly collected check out scanner data, commercially available data sets,
and grocery receipts) in population nutrition monitoring.
Using sales data have some advantage over traditional survey methods. Since
sales data are objective, they are relatively free form recall bias or deliberate
misinformation commonly encountered in traditional surveys. Use of supermarket sales
data was recommended as an indirect measure of intervention effectiveness (Kristal and
Ollberding, 2012). Similarly, Andreyeva and Luedicke (2013) used scanner sales data
obtained from supermarket chain to assess how the Women Infant Children revisions
affected purchases of bread and rice among WIC-Participating households in Connecticut
Texas Tech University, Janani R. Thapa, August 2014
90
and Massachusetts. In that study the main outcome variables were total weight of bread,
rice, and tortillas (in ounces) purchased by a household in a given month.
3.2.4 Influence of personal characteristics in food choice decisions
A trivariate probit model estimation of the effect of respondent’s personal
characteristics on decision to participate in attaining healthy weight was done by Chen
and Huffman (2009). The personal characteristics included were education, reading food
labels (signaling concern for good health), adjusted family income, opportunity cost of
time, occupation, marital status, race and ethnicity. This study concluded that men and
women who read more food labels are more likely to participate in moderate or vigorous
physical exercise and women are less likely to be obese. In a previous study, Bender and
Derby (1992) found that US label readers are more likely to be female, older, more
educated and on restriction diets.
Research on the use of nutrition facts label in Honolulu reported that one half of
the shoppers used nutrition facts label and an additional 18% reported using labels
sometimes. The frequency of use did not differ by age, but Caucasians reported using
labels more often than all of the others ethnic groups (Dooley et al., 1998). In a survey
conducted by the Food Marketing Institute (FMI), 31% of the food shoppers interviewed
either said they did not know what Percent Daily Values meant or defined it incorrectly
(FMI, 2013).
Texas Tech University, Janani R. Thapa, August 2014
91
3.3 Conceptual framework
The dictionary definition of persuasion is to cause someone to believe or
convince. Eysenck and Keane (2005) identified persuasion as an information processing
activity, in which thoughts are actively manipulated to create new beliefs and attitudes.
Heath and Fairchild (2007) defined persuasion as any activity which changes the attitudes
of the recipient. Two different types of persuasion are defined by Heath (2006): rational
and emotional. Performance claims, promotions, offers and the like employ both rational
and emotional persuasion. Moreover repeated positive experience from consuming a
product leads to strong brand loyalty. Promotions, convenience, advertising etc. to some
extent have control over individual’s choice.
Figure 3.1 Conceptual Framework for the Supermarket Intervention
Texas Tech University, Janani R. Thapa, August 2014
92
Figure 3.1 above shows the conceptual framework for the supermarket
intervention. A traditional view considered human behavior to be largely a rational
choice but the rational choice view is narrow (Cawley, 2004). Non-economic factors also
affect decision explained by behavioral economics. Behavioral economics assumes that
people depart from rationality in systematic ways (Kahneman, 2003). The role of
psychology in decision making have been explained in Damasio (1994) as cited in
Loewenstein (2000).
Success has been reported in applying behavior economics to increase healthy
food choice and consumption. A growing body of research suggests that people respond
to contextual cues without conscious thought or decision making, perception also affects
decision. Distraction as an external cue is reported to have major effect on the food
selected, the amount consumed and the eater’s perception (Just et al., 2007). Self-
attribution: when people feel that they have made their own decisions, has been recorded
to provide greater satisfaction to the consumer (Just and Wansink, 2009). Choice
architecture (where in choices are affected without letting the decision makers know that
their decision have been influenced) has been shown to work (Just et al., 2007; Just and
Wansink, 2009; Hanks et al., 2012a; Wansink et al., 2012 a, b and Van Kleef et al.,
2013). The ability to act automatically (e.g. if people touch something very hot, they will
withdraw their hands before they have time to make conscious decision to do so) is a
protective measure. However, it extends to eating and food choices (Cohen and Babey,
2012).
Texas Tech University, Janani R. Thapa, August 2014
93
Consumer behavior in food choice decisions can be usefully characterized in two
ways: a) prescriptive that can be persuaded, and b) rapid decision based on heuristic
devices (appearance, shapes, sizes, logos, brands etc.). The most important factor in a
supermarket intervention success is how it influences the consumer. Persuasion and
reinforcement mechanism during post facto choice behavior was identified also by Sheth
(1974). Both the prescriptive and rapid decision approaches are based on behavior. There
are specific behaviors initiated automatically by contextual cues that were previously
congruent with the performance of behaviors (Orbell and Verplanken, 2010).
This research applies both prescriptive and rapid food choice decision theories to
change the behavior of consumers towards healthier food choice decision. The shelf
talkers that were placed near relatively healthier items is proposed to help rapid decision
makers by altering the heuristic device by providing healthy living visual cue. Providing
shelf talkers has been listed under contextual influences in Cohen and Babey (2012). On
the other hand, the overall multi-tiered health promotion approach supports prescriptive
decision making based on persuasion.
Further, research on behavioral change in many different areas (e.g. smoking
cessation, nutrition, and exercise) has focused on the multiple aspects involved in
promoting behavioral change (e.g., an individual’s readiness to change, economic factors,
barriers to change). Drawing on literature across disciplines (economics, business,
psychology, and marketing) it is clear that behavioral change is influenced by a multitude
of factors (Fisher et al., 2002 and Rose et al., 2009). Likewise, reinforcements for
behavior change should come from many different aspects of the environment to be most
effective. Economos et al. (2007) demonstrated how the community context can be
Texas Tech University, Janani R. Thapa, August 2014
94
modified and reinforcements can be offered to prevent weight gain. Embedding
reinforcements within multiple levels of a person’s environment provides a
comprehensive effort towards addressing obesity risk factors.
3.4 Methods
3.4.1 Research setting
A small community of Muleshoe, TX was the setting to implement the project.
The supermarket intervention was a part of a multi-tiered cancer prevention model with a
substantial focus on obesity reduction. The activities in the supermarket were
implemented in partnership with the management team of a local supermarket located in
the intervention community. The selected supermarket was United Supermarket LLC.
This supermarket is the largest supermarket of its size and sale volume in the city
followed by small chain stores. The store intervention can be categorized into three types:
1. Point of sale signs or nudges to encourage consumers to purchase healthier
foods in main supermarket categories,
2. An overall emphasis and communication to promote purchase and
consumption of fresh fruits and vegetables (hereafter referred as produce) and
other identified healthier food items, and
3. Community efforts to reduce obesity. The point of sale nudges and healthy
eating messages were reinforced by a community education program
including newspaper articles, food demonstrations, and public service
announcements on television and in-person healthy eating classes.
The supermarket intervention to promote purchase and consumption of produce
were carried out from January of 2012 to December of 2012 (hereafter referred as the
Texas Tech University, Janani R. Thapa, August 2014
95
trial period). Products were selected by their relatively higher NuVal score within each
category targeted. The NuVal system places a single numeric score 1-100 based upon the
nutritional characteristics of the product and allows us to compare foods using a single
metric to establish the relative healthfulness (Katz et al., 2010). The shelf talker nudges
were checked once every month for placement accuracy. A total of 402 items were
identified for placing shelf talkers. The data for this paper is the sales data of the 402
product with shelf talkers from the intervention store and the sales data of same products
from the control store.
3.4.2 Data
The data set for this research was different for the produce and the identified
healthier food items. Each data set has been explained separately under produce and point
of sales nudges sections below. Further, the demographic data used to analyze the
awareness and attitude towards healthy food promotion in supermarket are described
under the demographics and project activities data in the sections below.
3.4.2.1 Produce
The data set includes monthly sales data (measured in units sold per product or in
ounces per produce item) and their price, from January 2011 to December 2011 (hereafter
referred to as the pre-trial period) and the trial period from the intervention and control
store. There were a total of 2240 produce items being sold over the period of 24 months.
However, many of the produce items were either discontinued or seasonal sales. The data
set was cleaned to include items that were consistently sold over the 24 months (12
month of pre-intervention and 12 month of post-intervention) period with Universal
Product Codes (UPC). Thus the analysis is done in following two parts:
Texas Tech University, Janani R. Thapa, August 2014
96
1. Continued items: Includes sales data for only 10 categories of produce: apples,
avocados, bananas, beans, bell-pepper, berries, carrots, citrus, cucumber, and eggplant
that were sold across the 24 months,
2. Continued and discontinued items: The dataset with or without continued sale
of a unique UPC coded item categorized into the following fourteen categories: apples,
avocados, bananas, beans, broccoli, citrus, corn, cucumber, grapes, green-leafy, other
fruits, other vegetables, potatoes and salad. However, the bean and potato categories were
dropped from the analysis. The bean category were dropped because it contained many
dried bean items as well and the project promotion was on increasing sales of fresh fruits
and vegetables and not other forms of vegetables including dry beans. Similarly potatoes
was also dropped because there is an increasing debate among nutritionist that potato is
biologically a vegetable but unlike other vegetable it is a starchy food like grains.
3.4.2.2 Point of sale nudges
The data for this paper is the monthly sales data of the 402 products with shelf
talkers, from the intervened store and the control store. The control store was in a
location with demographics similar to intervention store location. The data includes
monthly sales data (measured in units sold per item) from January 2011 to December
2012 from both stores. The data were obtained from computerized cash registers, which
record all individual purchases by UPC. None of the selected items were on promotion or
discounted in price differently at any time during the analysis period. A similar
comparison of sales data with two stores in matching geographic region and demography
was accomplished by Sacks et al. (2010). The 402 selected items were categorized into
12 categories based on the Food Marketing Institute Product Category List, published in
Texas Tech University, Janani R. Thapa, August 2014
97
2012 (FMI, 2012). The 12 categories with the number of items in each category in
parenthesis are: Bakery and Bread (27), Baking (24), Beverages (29), Soda (4), Breakfast
(48), Can foods (70), Dairy and Cheese (34), Frozen (41), Grain Pasta and Sauce (10),
Meat and Seafood (39), Pantry (41), Snacks (35). Monthly sales data has been reported to
have the same variation compared to weekly data by Narhinen et al. (1998).
3.4.2.3 Community awareness and project activities data
Two small communities: Muleshoe and Dalhart in the state of Texas were
selected, as project intervention and control sites respectively, based on their similar
demographics. Muleshoe was selected as the intervention site as this small community
met the criteria for implementing community-based research. The average poverty rate in
these communities ranged from 9% to 23% and more than one third of the populations
are Hispanic. The poverty rate of Dalhart is 9-15%. The population of Muleshoe is about
4,571 with 17-50% of the population estimated to be Hispanic. The poverty rate of
Muleshoe is 17-23% (TTUHSC, 2010). The data for this section were drawn from the
data collected in two independent surveys among a total of three surveys undertaken
during the project period. The demographic data used in this chapter were from the post-
intervention survey (June of 2012) that was conducted at both control and intervention
site and from a follow-up (January of 2013) survey that was conducted only in the
intervention site. The project activity data used in the analysis is from the intervention
site from the post-intervention and follow-up survey.
There were a total of 550 respondents during 2012 survey: 335 from Muleshoe
(Intervention) and 215 from Dalhart (Control). Of the total respondents in Muleshoe and
Dalhart, 69 and 39 of them had participated in the baseline pre-intervention survey.
Texas Tech University, Janani R. Thapa, August 2014
98
Inclusion criteria for participation were for individuals who were at least 18 years of age,
and self-identified as living in the communities of Muleshoe and Dalhart. They were
invited to participate by mail sent out to a random sample of local population drawn from
the local telephone book. The average show up rate was 22.3%. This random sample was
supplemented by participants invited through flyers and pamphlets distributed in public
areas and printed in local newspapers. Similar recruitment with mail survey was also
done by Crawford et al. (2000). The demography of mail recruited respondents and the
respondents recruited through flyers and pamphlets were not significantly different
(Appendix 3.1 and 3.2) in both the comparison and intervention site. Appendix 3.3 also
shows that the response on other indicator variables used in this analysis was not
significantly different between the respondents recruited through flyers and pamphlets in
the intervention site during the post-intervention survey.
3.4.3 Data analysis
The data analysis was done separately for the two data sets: 1) produce data, and
2) point of sale nudges data. Similarly, the data set included household characteristics of
respondent who shopped from this supermarket. However the sales data was not
available at individual level. Hence, the analysis is limited to comparing the change in
sales by category in the intervention and control supermarket during the pre-trial and trial
period. Similar tests to compare change in sales in units sold per item has been done to
analyze supermarket intervention (Sacks et al., 2011). Further, the demographic data
available from the questionnaire survey was used to compare the attitude toward project
activities conducted in the supermarket and to find attitudes towards in-store healthy food
choice messages and nudges by demographics.
Texas Tech University, Janani R. Thapa, August 2014
99
The analysis of this data set however was limited due to lack of category sales
data to compute proportional sale as done by Narhinen et al. (1998). Since the data
available from the supermarket was a complete set of sales data in the produce aisle,
computation of category sales data would have been possible. However, the lack of
standard categories within the produce section would make the result inconsistent. Hence
the category sales were not computed. Further, the interest of the research was to increase
the sales of fresh fruits and vegetables overall and not a particular item with a fruit or a
vegetable category. The interest in putting shelf talkers on healthier items was to
increase proportional sales of the nudged item within category. However the lack of
category sales data limited our analysis to looking at the sales difference with control
store during the pre-trial and trial period.
All the analysis was done using SAS Enterprise Guide 5.1, SAS Institute Inc. The
sales data in units sold per month was standardized to compute the units sold per unit
dollar before analysis to account for the price differential between the pre- and post-
intervention period, the price across all months was same in both the comparison and
intervention stores.
3.4.3.1 Produce data analysis
Paired comparison for pre- and post-intervention sales in intervention community
The difference between the monthly sales data on produce purchases during the
pre-trial and trial period was tested using paired t-tests for the data set of items that were
sold each month from January 2011 to December 2012 in the intervention store.
Texas Tech University, Janani R. Thapa, August 2014
100
Change in produce sales an overall comparison
In the paired comparison if a brand of oranges in the citrus category with a specific
UPC code would have been continued to be sold but if it was also substituted (not
replaced) by a new brand of oranges with a different UPC code within the citrus category.
In cleaning the data for this analysis the item with the different UPC code that was started
to be sold later in the year would be deleted (as a discontinued item) and hence the
deleted item would cause the total sale of citrus to decrease by including only the item
with the specific UPC code in the analysis. This caused an unusual fluctuation in sales
data. For this reason the analysis was done for all items in the category following the
paired comparison. The sales data on produce purchases with or without continued sale of
a particular item was used to test the mean difference of sales against a null hypothesis of
zero difference by using independent t-test for the dataset.
To undertake an overall comparison, all the items available in the produce data set
were categorized as explained in 3.4.2.1 (second part). The overall change in the sales of
produce was computed by double difference method. The method of finding simple
intervention effect is finding the difference between the changes in intervention mean
minus the change in comparison mean. This is a typical way to measure the change and is
a comprehensible method (Naresh, 2007 and Kristal and Ollberding, 2012):
( ) ( )
where, is the variable of interest. Subscript and refer to pre-trial and trial
respectively, and subscript and are for intervention and comparison.
Further, the percentage change in sum of units sold per month for each category
was computed for both intervention and control store. The overall percentage change in
Texas Tech University, Janani R. Thapa, August 2014
101
produce sale as found from the formula above was computed from the pre-trial sale in the
intervention community. The mean sales during the pre-trial and trial period were
compared using independent t-test, separately for the intervention and control site.
3.4.3.2 Point of sale nudges data analysis
The change in sales data in units sold per month was summed across each
category of identified food items from pre-trial to trial period. The percentage change in
the sum of units sold per month for each category was computed for both intervention
and control store. However, the change in mean sales in units sold per month from the
pre-trial to trial period was tested using independent t-test separately for the intervention
and control store. Unlike for the produce data set, no paired tests were done for the point
of sale nudges data.
The overall percentage change in produce sale was computed from the pre-trial
sale in the intervention community. The overall effect of the project intervention in sales
of point of sale nudges item was computed by double difference method. A method of
finding simple intervention effect by finding the difference between the changes in
intervention mean minus the change in comparison mean as explained in 3.4.3.1 was
used.
3.4.3.3 Community awareness and use of project activities
Intervention awareness and attitude towards intervention (shelf talkers and NuVal
scores) of the respondents who self-reported regular shopping in the supermarket was
also analyzed by demographics.
Texas Tech University, Janani R. Thapa, August 2014
102
3.5 Results
3.5.1 Produce sales
3.5.1.1 Paired comparison in intervention community
The mean units of produce sold per month during the pre-trial and trial period in
the intervention community, the change in unit sales per month and the corresponding p
value from a paired t-test between sales during the pre-trial and trial period is shown in
Table 3.1. A paired t-test was used because the comparison was done for the produce
items that were consistently sold over the 24 months (12 month of pre-intervention and
12 month of post-intervention period followed by UPC. Further if data for any month was
missing in one of the year the data for concurrent month was deleted from the other year
to make a paired observation. As shown in Table 3.1, the sales of bananas, carrots,
cucumber and eggplant increased significantly, while there was a significant decrease in
sales of citrus. For the reasons explained in Section 3.4.3.1, the paired comparison in
Table 3.1 shows large fluctuations in the sale of common fruits like citrus.
Texas Tech University, Janani R. Thapa, August 2014
103
Table 3.1 Sales of produce during pre-trial and trial periods
Category
Pre-trial Trial
Mean SD Mean SD Change p value
Apples 288.39 781.33 337.68 860.25 49.29 0.166
Avocados 4395.39 4039.28 4143.43 3392.26 -251.96 0.567
Bananas 7702.25 7921.46 8056.30 8280.44 354.05 0.028**
Beans 13.97 22.24 21.97 44.78 8.00 0.181
Bell pepper 345.42 590.39 358.78 581.00 13.35 0.502
Berries 9.51 33.24 15.42 41.40 5.91 0.178
Carrots 66.40 113.09 73.64 131.51 7.23 0.055*
Citrus 4072.78 8411.63 3014.71 6452.85 -1058.07 0.014**
Cucumber 54.93 93.80 134.98 202.26 80.05 0.037**
Eggplant 11.82 2.87 17.15 6.08 5.33 0.026**
p value from paired t-test, * and
** significant at 10 and 5% level of significance
respectively
Texas Tech University, Janani R. Thapa, August 2014
104
3.5.1.2 Change in produce sales and overall comparison
The change in produce sales was computed by finding the change in sum of
produce sales in units sold per month in pre-trial period and trial period in both the
control and intervention store. The percentage change in the sum of produce sales in units
sold per month and the overall change found by double difference method as explained in
Section 3.4.3.1 is shown in Table 3.2. The table reports the percentage change in sales
(monthly units sold summed across each category) by category for all the fresh produce
items sold in the supermarket. The change % for the intervention and control store is the
change from pre-trial period. As shown in Table 3.2, the sale of avocados, citrus, and
grapes decreased in both the intervention and control store. In addition, in the control
store the sale of apples, broccoli, corn, and salad also decreased in the trial period from
pre-trial period.
Table 3.2 also reports the overall change in units sold (%). The overall change is
derived from change in intervention and control store. There was an overall growth in the
sales of apples, bananas, broccoli, corn, green leafy, other fruits, and salad. The results
also showed there was less decline in the sales within category avocado, citrus, and
grapes. The overall decrease in sales was observed in categories cucumbers and other
vegetables. The decrease in the sales of cucumbers was very high (21%) and could not be
explained with the available data set.
Further, as shown in Table 3.2 the total number of units sold from pre-trial period
to the trial period decreased in both stores. The total number of units sold decreased by
12% in the intervention store and by 16% in the control store. Using the double
difference method the overall change in the monthly units sold was an increase of 4% in
Texas Tech University, Janani R. Thapa, August 2014
105
the sales during the trial period in the intervention store. The overall sales increase was
highest for the categories apples, broccoli and green leafy vegetables.
Table 3.2 Change in units sold from pre-trial period to trial period and overall change
Produce category
Change in units sold (%)
Intervention Control Overall
Apples 6 -13 19
Avocados -6 -9 3
Bananas 5 2 3
Broccoli 9 -5 14
Citrus -19 -25 6
Corn 7 -2 9
Cucumber 0 21 -21
Grapes -34 -36 3
Green leafy 62 39 23
Other fruits 7 1 6
Other vegetables 0 3 -3
Salad 2 -3 5
Total -12 -16 4
Texas Tech University, Janani R. Thapa, August 2014
106
Independent t-test was used to compare the sales during the pre-trial period with
the sales during the trial period in both intervention and control store. Table 3.3 shows
the statistics from independent t-test for mean difference from pre-trial to trial period
with a null hypothesis of no difference in sales in units sold per month between the pre-
trial period and trial period separately in the intervention and control store. As shown in
Table 3.3, mean sales in units sold per month was significant only for green leafy
produce (at 10% level of significance in the control store and 5% level of significance in
the intervention store).
The analysis reported in Tables 3.2 and 3.3 supports the inference that the project
had a positive impact on purchase of fresh produce. The change in produce purchase in
green leafy produce category was significant and the overall effect was positive.
Texas Tech University, Janani R. Thapa, August 2014
107
Table 3.3 Change from pre-trial to trial period for produce (mean no. of units sold)
Produce
category
Control Intervention Mean change
Mean (SD) Mean (SD)
Pre-trial Trial Pre-trial Trial Control Intervention
Apples 173.30 150.55 103.17 109.44 -22.75 6.27
(995.86) (827.37) (569.42) (575.57)
Avocados 537.76 491.70 462.74 436.18 -46.06 -26.56
(2038.89) (1638.06) (1865.65) (1670.32)
Bananas 1194.61 1213.24 770.26 805.70 18.63 35.44
(5219.32) (5263.09) (3376.41) (3530.45)
Broccoli 17.82 16.62 9.27 9.99 -1.20 0.72
(78.43) (71.51) (42.56) (44.84)
Citrus 1239.49 929.60 1126.16 907.45 -309.89 -218.71
(6440.96) (4249.20) (5604.14) (4364.26)
Corn 299.20 293.04 200.50 215.27 -6.16 14.77
(1611.94) (1462.32) (1047.24) (1090.05)
Cucumber 248.05 298.95 217.01 217.02 50.90 0.01
(875.28) (1042.21) (830.63) (712.06)
Grapes 124.85 80.50 67.52 44.85 -44.35 -22.67
(519.57) (256.33) (242.98) (140.45)
Green 2.28 3.14 2.19 3.53 0.86
* 1.34
**
(7.04) (8.17) (8.89) (11.78)
Fruits 37.72 37.82 26.36 28.07 0.10 1.71
(277.87) (268.70) (189.74) (192.28)
Vege. 57.55 58.94 36.77 36.73 1.39 -0.04
(491.31) (356.57) (337.67) (242.44)
Salad 41.98 40.95 31.77 31.47 -1.03 -0.30
(324.91) (313.90) (298.97) (289.50)
p values from t-test, * and
** significant at 10 and 5% level of significance respectively
Texas Tech University, Janani R. Thapa, August 2014
108
3.5.2 Point of sale nudges
3.5.2.1 Change in sales and overall comparison
The sum of sales in units sold per month was higher in the control store than in
the intervention store for all categories during both the pre-trial and trial periods. The
change in sales of nudged item (items with shelf talkers) was computed by finding the
change in sum of sales of nudged item in units sold per month from pre-trial period to
trial period in both comparison and intervention stores. The overall change was found by
using the double difference method as explained in Section 3.4.3.1. The results are shown
in Table 3.4. The table reports the percentage change in sales (monthly units sold
summed across each category) by category for all nudged item and the change %. The
change % is the change from pre-trial period. The overall change in units sold (%) is the
overall change from the units sold in the pre-trial period.
As shown in Table 3.4, the total units sold of the nudged items decreased from the
pre-trial period to the trial period in both stores. The total units sold of nudged items
decreased by 4% and 5% respectively in the intervention and control store. The project
expectation was for these items’ sale to be increased within their respective category. Due
to lack of category sales data it could not be clarified if the drop in sales was throughout
the category or in the nudged items only. However, the availability of sales data and
hence the comparison of sales with the control store provides some idea on the effect of
intervention.
The change in monthly sales from pre-trial to trial period was not homogenous
across the categories. The sale of nudged items in bakery and bread, can food, dairy,
drinks excluding soda, soda, and meat and seafood category decreased in both the
Texas Tech University, Janani R. Thapa, August 2014
109
intervention and control store. The sale of nudged items in breakfast category decreased
only in the control store. There was an overall growth in the sales of nudged items in
categories breakfast, frozen, and grain and pasta. The results also showed there was less
decline in the sales of nudged items in categories can food, dairy, drinks excluding soda,
meat and seafood, and soda. The overall decrease in sales was observed in the sales of
nudged items in categories bakery and bread, baking, snacks and a higher decline in the
intervention store in the category pantry. From this comparison of units sold, the category
grain pasta and sauce followed by the categories of (a) canned food, (b) frozen, and (c)
meat and seafood shows a potential positive nudging effect.
Texas Tech University, Janani R. Thapa, August 2014
110
Table 3.4 Total percentage change (units sold) from the pre-trial to trial period
Point of sale shelf talker category Change in units sold (%)
Intervention Control Overall
Bakery and bread -14 -4 -10
Baking 25 29 -4
Breakfast 2 0 2
Can food -3 -8 11
Dairy -10 -14 4
Drinks excluding soda -4 -5 1
Frozen 13 1 12
Grain and Pasta 44 15 29
Meat and seafood -4 -10 6
Pantry -11 -6 -5
Snacks 4 14 -10
Soda -1 0 -1
Total -4 -5 1
Texas Tech University, Janani R. Thapa, August 2014
111
An independent t-test was used to compare the sales of nudged items during the
pre-trial period with the sales of nudged items during the trial period, separately for the
control and intervention stores. Table 3.5 shows the statistics from independent t-test for
mean difference from pre-trial to trial period with a null hypothesis of no difference in
sales in units sold per month between the pre-trial period and trial period separately in the
intervention and control store. As shown in Table 3.5, mean sales in units sold per month
was significantly higher in the intervention store only for nudged items in category grain
and pasta (at 5% level of significance). From this t-test the category grain pasta and sauce
shows a potential positive nudging effect.
Texas Tech University, Janani R. Thapa, August 2014
112
Table 3.5 Change from pre-trial to trial period (mean units sold per month)
Category Control Intervention
Mean (SD) Mean (SD) Mean change
Pre-trial Trial Pre-trial Trial Control Intervention
Bakery and bread 54.26 51.87 42.57 36.84
-2.39 -5.73 (86.56) (88.84) (65.61) (57.79)
Baking 9.14 11.80 6.43 8.04
2.66 1.61 (17.42) (22.85) (12.72) (16.41)
Breakfast
5.69 5.67 3.90 3.99 -0.02 0.09
(7.41) (7.33) (4.80) (4.61)
Can food
48.47 44.80 27.06 26.26 -3.67 -0.80
(131.38) (119.34) (71.63) (66.45)
Dairy
34.39 29.63 29.61 26.52 -4.76 -3.09
(108.91) (83.54) (93.42) (84.22)
Drinks excluding soda
16.39 15.51 13.21 12.63 -0.88 -0.58
(44.47) (38.24) (45.45) (43.16)
Frozen
9.18 9.31 5.42 6.11 0.13 0.69
(10.28) (11) (6.38) (7.06)
Grain and Pasta
13.94 16.07 8.61 12.42 2.13 3.81
**
(12.45) (16.84) (7.74) (17.40)
Meat and seafood
8.11 7.27 5.72 5.46 -0.84 -0.26
(12.61) (11.91) (8.98) (7.68)
Pantry
16.89 15.91 10.56 9.41 -0.98 -1.15
(24.18) (25.37) (15.30) (17.07)
Snacks
9.23 10.51 8.50 8.88 1.28 0.38
(14.24) (15.62) (12.19) (12.34)
Soda 11.95 11.91 7.48 7.44
-0.04 -0.04 (7.71) (7.59) (5.19) (5.86)
p values from t-test, * and
** significant at 10 and 5% level of significance respectively
The results are suggestive that the effect of a healthy nudge intervention will be
greatest when it focuses upon products with already existing product promotion
information. Specifically, it is likely the case that whole grain pastas and healthier sauces
were successfully encouraged because of the existing programs to encourage
consumption of whole grains (e.g. http://choosemyplate.gov/food-groups/grains-
why.html). The increased consumption of whole grains has been a common
encouragement by dietitians for a number of years.
Texas Tech University, Janani R. Thapa, August 2014
113
3.5.3 Community awareness and use of project activities
3.5.3.1 Community’s perception on project activities
In doing the analysis to find the effect of supermarket intervention in healthy food
choice decision the current analysis was restricted by the unavailability of data to analyze
the direct effect of intervention at an individual level. However, an important indicator of
success and usefulness of the project is whether community members recognized and
found project activities to be helpful. With this end in mind an analysis of respondent’s
attitude towards project activities was done by using the data available from the
questionnaire survey done during the trial period and the follow-up survey as described in
Section 3.4.2.3. From the survey during the trial period, 85% of the respondents in the
intervention site had seen project health messages in the project period. This shows the
overall awareness of health messages in the community to be 85%.
The results of respondents’ attitude towards each project activities during the trial
period are shown in Table 3.5. As shown in the table, community members were most
aware of the shelf talkers in the supermarket (38%) followed by information from
community education classes (25%) and nutrition education classes (5%). Fifty-four
percent of the respondents reported to have tasted the foods during the food tastings
organized by the project in the supermarket. Project activities were viewed very
positively by community members. Of the respondents’ aware of information and
programs provided to 4-H, boy and girl scouts, youth and other groups on cancer
prevention 97% reported it to be helpful. Of the respondents who had seen blue or green
signs: shelf talkers in the supermarket, 80% reported it to be helpful. Similarly, 85% of
Texas Tech University, Janani R. Thapa, August 2014
114
the respondents’ who had considered NuVal score before purchasing food item reported
the score system to be helpful.
Table 3.6 Community views of the project during the trial period
Value use of information provided (N=276) yes proportion
1. Aware of information and programs provided to 4-H, boy and girl
scouts, youth and other groups on cancer prevention
70 25%
If yes, agree (strongly agree or agree) these programs have helped
learn more about cancer risk reduction and healthy living
68 97%
2. Participate in monthly education classes at County Electric Co-op 14 5%
If yes, agree (strongly agree or agree) participation has helped to
learn more about cancer
14 100%
3. Taste foods at the food tasting at the Supermarket 148 54%
Tried the recipes received at food tasting in the Supermarket 53 36%
4. Seen Cut your risk boost your health (shelf talkers) 106 38%
If yes, agree (strongly agree or agree) that the signs were helpful 85 80%
5. Consider NuVal score before purchasing food item 39 14%
If yes, higher NuVal score should be selected 28 62%
If yes, NuVal score is helpful (very helpful or helpful) 33 85%
Texas Tech University, Janani R. Thapa, August 2014
115
The results of respondents’ attitude towards each project activities during the
post-trial period from the follow-up survey are shown in Table 3.7. As shown in the table,
community members were most aware of the shelf talkers in the supermarket (48%)
followed by information from community education classes (28%) and nutrition
education classes (5%). Sixty percent of the respondents reported to have tasted the foods
during the food tastings organized by the project in the supermarket. Of the respondents’
aware of information and programs provided to 4-H, boy and girl scouts, youth and other
groups on cancer prevention 54% reported it to be helpful. Of the respondents who had
seen blue or green signs: shelf talkers in the supermarket, 67% reported it to be helpful.
Similarly, 5 % of the respondents’ who had considered NuVal score before purchasing
food item reported the score system to be helpful. Hence all the project activities were
viewed positively by the community members, less positively during the follow-up
period compared to during the trial period.
Texas Tech University, Janani R. Thapa, August 2014
116
Table 3.7 Community views of the project from the follow-up survey
Value use of information provided (N=200) yes proportion
1. Aware of information and programs provided to 4-H, boy and girl
scouts, youth and other groups on cancer prevention
57 28%
If yes, agree (strongly agree or agree) these programs have helped
learn more about cancer risk reduction and healthy living
31 54%
2. Participate in monthly education classes at County Electric Co-op 11 5%
If yes, agree (strongly agree or agree) participation has helped to
learn more about cancer
8 73%
3. Taste foods at the food tasting at United Supermarket 120 60%
Tried the recipes received at food tasting in the Supermarket 64 53%
4. Seen Cut your risk boost your health (shelf talkers) 96 48%
If yes, agree (strongly agree or agree) that the signs were helpful 64 67%
5. Consider NuVal score before purchasing food item 41 20%
If yes, higher NuVal score should be selected 16 39%
If yes, NuVal score is helpful (very helpful or helpful) 22 54%
3.5.3.2 Awareness by demographics
From the available data set it was also possible to analyze the intervention
awareness and attitude towards intervention (shelf talkers and NuVal scores) of the
respondents who self-reported regular shopping in the supermarket was also analyzed by
demographics. This section explains the intervention awareness and attitude towards
intervention by several demographic variables (education, income, race, age, gender, and
language).
Texas Tech University, Janani R. Thapa, August 2014
117
Education
Table 3.8a shows the awareness and use of NuVal score by level of education.
The highest proportion of respondents who considered NuVal score system during
purchase was high school graduate, 38% and 50% respectively in intervention and
control community. The highest proportion of respondent who used NuVal score system
during purchase and considered it helpful was also high school graduates, 29% and 38%
respectively in intervention and control community.
Table 3.8a Use of NuVal score system by education level
NuVal score
Consider during purchase (%) Consider helpful (%)
Intervention Control Intervention Control
Elementary 12 7 24 7
High school graduate 38 50 29 38
Some college or technical school 25 27 22 34
College graduate 25 17 24 21
Table 3.8b shows the awareness and use of healthy eating signs placed by the
project by the respondents’ level of education in the intervention site. The highest
proportion of respondents who considered healthy eating signs during purchase was high
school graduate (39%). The highest proportion of respondent who used healthy eating
signs during purchase and considered it helpful was also high school graduates (39%).
Texas Tech University, Janani R. Thapa, August 2014
118
Table 3.8b Use of in-store signs by education level in the intervention community
Store signs Consider during purchase (%) Consider helpful (%)
Elementary 21% 24%
High school graduate 39% 39%
Some college or technical school 21% 21%
College graduate 19% 16%
Income
Table 3.9a shows the awareness and use of NuVal score system by income. The
highest proportion of respondents who considered NuVal score system during purchase
was in the income range of $20,000 to $35,000, 23% and 33% respectively in
intervention and control community. The highest proportion of respondent who used
NuVal score system during purchase and considered it helpful were in the income range
less than $20,000 in intervention cite (34%) and in the range of $20,000 to $35,000
(38%) in the control community.
Table 3.9a Use of NuVal score system by income groups
NuVal score
Consider during purchase (%) Consider helpful (%)
Intervention Control Intervention Control
Less than $20,000 31% 27% 34% 24%
$20,000 - $35,000 23% 33% 21% 38%
$35,000 -$ 50,000 19% 10% 19% 3%
$50,000 - $75000 10% 20% 9% 17%
over $75000 17% 10% 17% 17%
Texas Tech University, Janani R. Thapa, August 2014
119
Table 3.9b shows the awareness and use of healthy eating signs placed by the
project by the respondents’ level of income in the intervention community. The highest
proportion of respondents who considered healthy eating signs during purchase in the
intervention community were in the income level less than $20,000 (42%). The highest
proportion of respondent who used healthy eating signs during purchase and considered it
helpful was also in the income level less than $20,000 (45%).
Table 3.9b Use of in-store signs by income groups in the intervention community
Store signs Consider during purchase (%) Consider helpful (%)
Less than $20,000 42% 45%
$20,000 - $35,000 20% 22%
$35,000 -$ 50,000 14% 14%
$50,000 - $75000 13% 10%
over $75000 10% 9%
Gender
Table 3.10a shows the awareness and use of NuVal score system by gender. The
highest proportion of respondents who considered NuVal score system during purchase
were female, 75% and 63% respectively in intervention and control community. The
highest proportion of respondent who used NuVal score system during purchase and
considered it helpful were also female, 76% and 59% in the intervention and control
community respectively.
Texas Tech University, Janani R. Thapa, August 2014
120
Table 3.10a Use of NuVal score system by gender
NuVal Score
Consider during purchase (%) Consider helpful (%)
Intervention Control Intervention Control
Male 25% 37% 24% 41%
Female 75% 63% 76% 59%
Table 3.10b shows the awareness and use of healthy eating signs placed by gender
in the intervention community. The highest proportion of respondents who considered
healthy eating signs during purchase in the intervention community were female (70%).
The highest proportions of respondent who used healthy eating signs during purchase and
considered it helpful were also female (73%).
Table 3.10b Use of in-store signs by gender in the intervention community
Store signs Consider during purchase (%) Consider helpful (%)
Male 30% 27%
Female 70% 73%
Race
Table 3.11a shows the awareness and use of NuVal score system by race. The
highest proportion of respondents who considered NuVal score system during purchase
were white, 56% and 50% respectively in intervention and control community. The
highest proportion of respondent who used NuVal score system during purchase and
considered it helpful were also white, 52% and 48% in the intervention and control
community respectively.
Texas Tech University, Janani R. Thapa, August 2014
121
Table 3.11a Use of NuVal score system by race
NuVal score
Consider during purchase (%) Consider helpful (%)
Intervention Control Intervention Control
White 56% 50% 52% 48%
Hispanic 44% 43% 48% 45%
Other 0% 7% 0% 7%
Table 3.11b shows the awareness and use of healthy eating signs placed by race in
the intervention community. The highest proportion of respondents who considered
healthy eating signs during purchase in the intervention community were white (53%).
The highest proportions of respondent who used healthy eating signs during purchase and
considered it helpful was very close, 49% Whites and 50% Hispanics.
Table 3.11b Use of in-store signs by race in the intervention community
Store signs Consider during purchase (%) Consider helpful (%)
White 53% 49%
Hispanic 45% 50%
Other 1% 2%
Texas Tech University, Janani R. Thapa, August 2014
122
Language
Table 3.12a shows the awareness and use of NuVal score system by language.
The highest proportion of respondents who considered NuVal score system during
purchase spoke English language, 92% and 80% respectively in intervention and control
community. The highest proportion of respondent who used NuVal score system during
purchase and considered it helpful also spoke English, 86% and 83% in the intervention
and control community respectively.
Table 3.12a Use of NuVal score system by language
NuVal score
Consider during purchase (%) Consider helpful (%)
Intervention Control Intervention Control
English 92% 80% 86% 83%
Spanish 8% 20% 14% 17%
Table 3.12b shows the awareness and use of healthy eating signs placed by
language in the intervention community. The highest proportion of respondents who
considered healthy eating signs during purchase in the intervention community spoke
English (93%). The highest proportions of respondent who used healthy eating signs
during purchase and considered it helpful were also spoke English (92%).
Table 3.12b Use of in-store healthy eating signs by race
Store signs Consider during purchase (%) Consider helpful (%)
English 93% 92%
Spanish 7% 8%
Texas Tech University, Janani R. Thapa, August 2014
123
Age
Table 3.13a shows the awareness and use of NuVal score system by age group.
The highest proportion of respondents who considered NuVal score system during
purchase were aged more than 30, 92% and 86% respectively in intervention and control
community. The highest proportion of respondent who used NuVal score system during
purchase and considered it helpful were also in the age group more than 30, 86% and
72% in the intervention and control community respectively.
Table 3.13a Use of NuVal score system by age group
NuVal score
Consider during purchase (%) Consider helpful (%)
Intervention Control Intervention Control
Less than or equal to 30 8% 23% 14% 28%
More than 30 92% 77% 86% 72%
Table 3.13b shows the awareness and use of healthy eating signs placed by age in
the intervention community. The highest proportion of respondents who considered
healthy eating signs during purchase in the intervention community was more than 30
years old (86%). The highest proportions of respondent who used healthy eating signs
during purchase and considered it helpful were also in the age group more than 30 years
old (85%).
Table 3.13b Use of in-store signs by age group
Store signs Consider during purchase (%) Consider helpful (%)
Less than or equal to 30 14% 15%
More than 30 86% 85%
Texas Tech University, Janani R. Thapa, August 2014
124
3.6 Conclusions
Community-based models that incorporate the supermarket in program activities
are rare despite its potential effectiveness. A community-based cancer risk reduction
program was implemented with a major supermarket in a small community. The small
community setting allowed control over the media and a supermarket based intervention
provided a key source of information and reinforcement for healthy behavior. In the
project model, the supermarket was used a) to conduct healthy food demonstration, b) to
put shelf talkers with healthy eating message in comparatively healthy food products, c)
to provide healthy eating information, and d) to access food purchase data to evaluate the
effect of the project on food purchase behavior. This research paper reports the project
effect on food purchase behavior based on monthly sales in units sold per month for
produce items and healthier items identified with shelf talkers.
The overall change analysis that employed a double difference method and
comparative analysis suggest that the project had a positive impact on purchase of fresh
produce. The overall increase (or less decrease) was positive. On looking at the change
by category the change in produce purchase in green leafy produce category was
significant. Similarly, the overall change was positive in the sales of healthier items
identified with shelf talkers. On looking at the change by categories the change was
positive and significant for the category grain pasta and sauce. Thus the findings suggest
that supermarket interventions have potential in changing food behavior. However, the
lack of positive change across all categories also suggests the challenges associated with
changing food behavior. Another analysis of attitude towards project activities showed
Texas Tech University, Janani R. Thapa, August 2014
125
that that the general awareness was 85% and majority either strongly agreed or agreed
that the project activities were helpful.
The analysis reported in this paper was restricted by the lack of household level
data on food and produce items’ purchase. The data also lacked category level sales to
compute change in the market share of identified items with shelf talkers. However, the
demographic data available from the associated survey undertaken by the project was
used to analyze the supermarket intervention awareness and attitude towards awareness
by demographics. The results suggest that high school graduate in less than $20,000
income per year, English language speaking female over the age of 30 were most aware
of the interventions and had positive attitude towards the interventions. The data also did
not allow isolation of the shelf talkers effect from the community-based healthy message
reinforcement effect.
Texas Tech University, Janani R. Thapa, August 2014
126
Appendix
Appendix 3.1 Response comparison in intervention site from post-intervention survey
Response item Invitee Volunteer p value
What is your marital status?
Married 69% 62% 0.249
Divorced 8% 14% 0.205
Widowed 15% 4% 0.002
Separated 1% 6% 0.103
Never Married 6% 14% 0.067
What is the highest grade or year of school you completed?
Elementary 18% 25% 0.259
High school graduate 33% 36% 0.673
Some college or technical school 19% 23% 0.531
College graduate 29% 16% 0.019
What is your annual household income from all sources?
Less than $20,000 21% 35% 0.026
$20,000 to $35,000 19% 30% 0.090
$35,000 -$ 50,000 25% 14% 0.027
$50,000 - $75000 10% 14% 0.341
over $75000 25% 7% 0.000
Gender
Male 40% 35% 0.389
Female 60% 65% 0.389
Race
White 79% 40% 0.000
Hispanic 19% 58% 0.000
p values from independent t-test
Texas Tech University, Janani R. Thapa, August 2014
127
Appendix 3.2 Response comparison from control site during post-intervention survey
Response item Invitees Volunteers p value
What is your marital status?
Married 39% 78% 0.000
Divorced 14% 7% 0.177
Widowed 18% 1% 0.000
Separated 7% 4% 0.472
Never Married 21% 9% 0.067
What is the highest grade or year of school you completed?
Elementary 29% 23% 0.524
High school graduate 32% 36% 0.709
Some college or technical school 29% 27% 0.866
College graduate 11% 14% 0.623
What is your annual household income from all sources?
Less than $20,000 39% 23% 0.070
$20,000 to $35,000 29% 36% 0.460
$35,000 -$ 50,000 7% 14% 0.350
$50,000 - $75000 18% 13% 0.478
over $75000 7% 15% 0.275
Gender
Male 29% 36% 0.460
Female 71% 64% 0.460
Race
White 61% 41% 0.048
Hispanic 32% 55% 0.028
p value from independent t-test
Texas Tech University, Janani R. Thapa, August 2014
128
Appendix 3.3 Comparing response in intervention site during post-intervention survey
Aware of information and programs provided to 4-H, boy and girl scouts, youth and
other groups on cancer prevention.
Yes 21% 26% 0.448
No 79% 74% 0.448
If yes, these programs have helped you or your families learn more about cancer
prevention and healthy living.
Strongly agree 13% 30% 0.205
Agree 73% 62% 0.412
Neutral 13% 9% 0.582
Disagree 0% 0% NA
Strongly disagree 0% 0% NA
Participate in monthly education classes at Bailey County Electric Co-op
Yes 7% 3% 0.173
No 93% 97% 0.173
If yes, participation has helped to learn more about
cancer
Strongly agree 40% 33% 0.819
Agree 60% 67% 0.819
Neutral 0% 0%
Disagree 0% 0%
Strongly disagree 0% 0%
Do you shop at united supermarket?
Yes 94% 89% 0.211
No 6% 10% 0.254
Don’t know 0% 1% 0.537
Have you participated in food tastings?
Yes 52% 51% 0.851
No 48% 48% 0.972
Don’t know 0% 1% 0.382
Have you tried the recipes?
Yes 52% 51% 0.913
No 48% 49% 0.913
Have you seen the signs “cut your risk boost your health” in United Supermarket?
Yes 30% 40% 0.137
No 30% 32% 0.760
Have seen 24% 20% 0.526
Don’t know 17% 9% 0.054
If yes, have you found them helpful in making healthy food choice
Strongly agree 10% 28% 0.075
Agree 67% 53% 0.255
Neutral 24% 19% 0.621
Disagree 0% 0% NA
Strongly disagree 0% 0% NA
Texas Tech University, Janani R. Thapa, August 2014
129
Do you consider a food item’s NuVal score before purchasing an item?
Yes 18% 14% 0.443
No 37% 44% 0.293
Don’t know 45% 42% 0.626
Do you select items with higher or lower Vu-Val score to select the best nutritional
value?
Higher 37% 42% 0.507
Lower 39% 30% 0.218
Don’t consider NuVal 16% 20% 0.456
Don’t know 9% 8% 0.886
Do you find NuVal score helpful in making healthy food choices?
Strongly agree 11% 17% 0.468
Agree 42% 32% 0.404
Neutral 47% 46% 0.918
Disagree 0% 2% 0.581
Strongly disagree 0% 3% 0.432
p value from independent t-test
Texas Tech University, Janani R. Thapa, August 2014
130
CHAPTER IV
COMMUNITY-BASED BEHAVIOR CHANGE INTERVENTION FOR OBESITY
PREVENTION: A SYSTEMATIC REVIEW AND META-ANALYSIS
Abstract
This chapter reports the findings of a systematic review of research on community-based
behavior change interventions implemented for obesity prevention. Obesity is a pressing
international public health problem where community-based improvement efforts are
being used to improve obesity outcomes by improving energy balance related behavior
(EBRBs). An assimilation of the literature on impact evaluation of community-based
behavior change interventions designed to address obesity shows potential in guiding the
design, implementation, and evaluation of future community-based behavior change
interventions. Both the narrative synthesis and meta-analysis revealed that community-
based behavior change intervention have a strong impact in decreasing obesity and
energy balance related behaviors but this impact varies substantially. The overall fixed
and random effect of such interventions showed significant but modest decrease in BMI.
Further, this research leads to a much needed consideration during the design of effective
community-based behavior change interventions. The findings of this review suggest that
the designers of community-based interventions should carefully consider between large-
scale, less-intense vs. small-scale more-intense interventions for weight loss and obesity
prevention.
Key words: community-based, behavior, obesity prevention, body mass index (BMI),
energy balance, systematic review, meta-analysis
Texas Tech University, Janani R. Thapa, August 2014
131
4.1 Introduction
Obesity was often considered to be primarily an individual’s problem and some
success in the areas of diet and physical activity on an individual basis was observed (e.g.
Glanz et al., 1998). However, attaining obesity prevention or weight loss without relapse
has posed challenge in obesity prevention and at a societal level has been insufficient. It
is more likely that individual behavior change will be sustained if it is supported by
environment that reinforces healthy choices (Economos and Irish-Hauser, 2007). An
individual’s behavior is framed through individual differences (knowledge and attitude,
previous experiences, and personal preferences), the social environment, and the larger
community (Gotay, 2004). The increasing prevalence and increasing obesity trends
across communities led researchers to target obesity as a societal problem rather than
solely individual’s problem. A rationale for community-based interventions has also been
developed by Economos and Irish-Hauser (2007).
Behavior change has been a key strategy, as the obesity prevention paradigm
shifts from individual based intervention to community-based interventions. The behavior
change is geared towards attaining a decrease in body weight by community-based
interventions for obesity prevention. A generic conceptual framework for understanding
how interventions may promote behavior change has been provided in Baranowski et al.
(2009). Behavior change in obesity prevention can be based upon the concept of energy
balance. Energy balance is achieved when calories being consumed are equal to calories
expended. The World Health Organization (WHO) defined the fundamental cause of
obesity and overweight to be an energy imbalance between calories consumed and
Texas Tech University, Janani R. Thapa, August 2014
132
calories expended (WHO, 2013 and Baranowski et al., 2003). Further, WHO (2013)
pointed out that globally there has been:
an increased intake of energy dense foods that are high in fat, and
a decrease in physical activity due to the increasingly sedentary nature of many
forms of work, changing modes of transportation, and increasing urbanization.
According to US Department of Health and Human Services, a person gains weight when
he/she consumes more calories from food than the body uses through its normal functions
i.e. the basal metabolic rate (BMR) and physical activity (Miljkovic et al., 2008).
In this chapter, obesity prevention is studied with a focus on community-based
behavior change interventions. A key to obesity prevention is behavior change. However,
informing people of the health risks associated with their behavior is not sufficient
(Jeffery, 1995). Multiple behavior interventions can have significant effect in behavior
change towards obesity prevention ( Johnson et al., 2008). A behavioral science research
in weight gain has recommended four key topics related to obesity prevention and
physical activity promotion from a systematic review (Wing et al., 2001). The topics are:
1) environmental factors, 2) adoption and maintenance of healthful eating, physical
activity and weight, 3) etiology of eating and physical activity, and 4) multiple behavior
change interventions.
Related systematic review and meta-analysis in the area of community-based
behavior change interventions is Seo and Sa (2008). But Seo and Sa reviewed psycho-
behavioral obesity interventions among a limited group (US multiethnic and minority
adults). Similar review to find the potential to prevent or delay type 2 diabetes with
intensive lifestyle interventions was done by Satterfield et al. (2003), however the focus
Texas Tech University, Janani R. Thapa, August 2014
133
was on diabetes prevention. Another search of community-based intervention and
subsequent systematic review was done for community-based intervention to reduce
overweight and obesity by Gao et al. (2008) but the interventions were limited to China.
These are the only literature of the kind reported in this paper found in the obesity
prevention literature. The existent literature in the area of obesity prevention has not
looked specifically at community-based behavior change interventions.
A major question still exists on how to make community-based interventions
more effective, efficient and replicable. This question can partly be answered by
assimilating the intervention effect of community-based interventions in modifying
behavior and changing weight status towards obesity prevention. To accomplish this goal
a systematic review and a meta-analysis was undertaken. The systematic review method
has been cited as the best form of synthesizing evidence collected in the literature
(Wright et al., 2007). There are systematic reviews on what works best in an energy
balance related behavioral intervention. For example Stralen et al. (2011) has concluded
that self-efficacy and intention are relevant in physical activity interventions.
The current research provides a significant contribution to obesity prevention
literature by synthesizing the impact of community-based obesity prevention research
from different disciplines. This research will contribute in developing a trans-
disciplinary, translational obesity prevention program. The development of such program
has also been stressed by Kremers et al. (2006) and (Schmitz and Jeffery, 2000). A shift
from the traditional uni-disciplinary research to translational, trans-disciplinary research
has potential to accelerate both discovery and its translation to practice, and eventually
policy (Nebelling, 2012). This review intends to help recommend an effective
Texas Tech University, Janani R. Thapa, August 2014
134
community-based intervention program, the necessity of which was also pointed by
Kristal and Ollberding (2012), by looking at the evidence from previous community-
based behavior change interventions.
4.2 Review of literature
Looking at obesity from an energy balance perspective requires looking at two
components energy intake and energy expenditure. Energy intake can be influenced or
determined by socioeconomic to psychological factors. Evidence of diverse determinants
of energy intake is provided below.
Drewnowski (2003) discussed diet structure with Engel’s law “wealth and poverty
have profound effects on diet structure, nutrition and health.” According to Engel’s law
the proportion of income spent on food diminishes as income increases. Drewnowski
(2003) suggested that diet structure changes as well. Income and the macronutrient
composition of the diet are linked at the aggregate and most likely the individual level.
Furthermore, people in the higher income nations consume more added sugars and fats
than do people in lower income nations.
The availability and accessibility of fast foods, snacks and soft drinks, food
advertising and food marketing have been addressed by environmental studies (e.g. Kant
and Schatzkin, 1994; Harnack et al., 1999; French et al., 2001; Ludwig et al., 2001; Zizza
et al., 2001 and Swinburn et al., 2004 ). Cohen and Babey (2012) reviewed evidence that
dietary behaviors are in large part the consequence of automatic response to contextual
food cue. Restaurants and grocery stores are the primary setting from which people
obtain food. These settings are often designed to maximize sales of food by strategically
Texas Tech University, Janani R. Thapa, August 2014
135
placing and promoting items to encourage impulse purchases of high margin food
products.
A study that investigated the association between walkability and obesity
stratified by neighborhood race and socioeconomic status (SES) among adults residing in
Baltimore City by Casagrande et al. (2010) concluded that among some neighborhoods,
high walkability was associated with less obesity. Other studies like Jp (2011) reported
that factors governing energy intakes influence the energy balance far more powerfully
than factors determining resting energy expenditure. Pagliassotti (1997) reported that an
imbalance between energy intake and energy expenditure can explain approximately 80%
of the variance in body weight gain in a dietary model of obesity.
Thus prevention of obesity will require the study of how behavioral and
environmental factors have interacted to produce positive energy balance and weight
gain. Reversing the epidemic of obesity will require modifying some combination of
these factors to help the population achieve energy balance and a healthy body weight.
While body weight is strongly influenced by biological and behavioral factors, changes in
the environment promoting positive energy balance have been most responsible for the
obesity epidemic.
The best strategy for reversing the obesity epidemic is to focus on preventing
positive energy balance in the population through small changes in diet and physical
activity that take advantage of our biological systems for regulating energy balance. By
simultaneously addressing the environment to make it easier to make better food and
physical activity choices (Hill, 2006). A similar systematic review that included all the
non-drug controlled interventions in Mainland, China which used anthropometric
Texas Tech University, Janani R. Thapa, August 2014
136
outcome measures was done by (Gao, 2008). This review concluded that comprehensive
interventions with at least physical activity, dietary intervention and health education may
be effective in reducing obesity in China. Another similar search was done in MEDLINE
in 2002 (Fogelholm and Lahti-Koski, 2002). This research concluded that interventions
may need a stronger emphasis on changes in the local physical and social environment to
achieve improvements in energy expenditure due to physical activity
4.3 Conceptual framework
The energy balance as defined by the National, Heart Lung and Blood Institute is
the balance of calories consumed through eating and drinking compared to calories
burned through physical activity (NIH-NHLBI, 2013). What we eat and drink is energy-
in and what we burn is energy-out. Three conditions may occur:
1. Energy-in equal energy-out
2. More energy-in than energy-out
3. More energy-out that energy-in
Obesity is a function of an individual’s energy balance over a number of time
periods. Energy balance in a given time period is the difference between calories
consumed and expended in that period. In addition to this cumulative energy balance,
age, gender, race, ethnicity and genetic factors unique to an individual help determine
weight outcomes by influencing the process by which energy balances are translated into
changes in body mass. According to Stralen et al. (2011) psychosocial variables affect an
individual’s energy balance and are called the energy balance-related behavior (EBRBs).
The EBRBs are the set of variables that cause obesity by affecting the choice decision on
how much energy to consume, how much and by what means to expend the energy.
Texas Tech University, Janani R. Thapa, August 2014
137
There are research conducted in different discipline regarding the effect of EBRBs on
obesity such as in nutrition and physical activity (Dionne et al., 1997; Seale and
Rumpler, 1997; Murgatroyd et al., 1999 and Mobasheri et al., 2005) and psychology
(Williamson and Stewart, 2005 and Sluis et al., 2010 ). However, a single cause is not
independent of other causes, obesity is an outcome of complex set of behavior and
environment (Fiseković, 2005).
This systematic review is based on an analytic framework where obesity is the
cumulative function of individual’s energy imbalance over time. This analytical
framework was developed by Chou et al. (2004). The framework is explained here.
Obesity is a function of an individual’s energy balance over a number of time
periods (which are ages). If is the energy balance in period j, then it can be defined as,
(1)
where, are calories consumed in period j and are calories expended in all activities
(basal metabolic rate and physical activity) in period j.
Weight status (O) is a cumulative function of the individual’s energy balance over
a number of periods:
( ) (2)
where, is a vector of variables that is specific to an individual and is related to his or
her genetic predisposition towards obesity. Chou et al. (2004) have listed age, gender,
race, and ethnicity along with the genetic predisposition to be included in the vector .
Equation 2 thus suggests that according to the energy imbalance model of obesity it is
necessary to explain the determinants of calories consumed and calories expended.
Obesity is not an outcome variable by itself. Obesity is an outcome of decisions made to
Texas Tech University, Janani R. Thapa, August 2014
138
fulfill other goals. However, Chou et al. (2004) does not include other factors that
influence food choice decisions and underlying behaviors that causes obesity. In this line,
the following factors, in addition to the elements listed for vector can be considered as
factors that influences energy intake and energy expenditure:
Energy intake
Food demand (price, expenditure/income, advertising, demo, etc.)
Advertising
Food environment (including access)
Genes/Demography
Nutrition knowledge
Psychological or emotional reasons for intake (e.g. stress)
Quality of Life (vacations, aspiration-professional/cultural/educational,
Energy expenditure
Exercise
Work (nature of employment)
Lifestyle
Neighborhood environment (including access)
Genes/Demo
Nutrition knowledge
Psychological or emotional reasons for metabolism or burning (e.g. stress)
Texas Tech University, Janani R. Thapa, August 2014
139
An extension to equation 2, also developed by Chou et al. (2004) is as follows:
( ) (3)
where, is calorie consumed, is active leisure (either the time allocated to this activity
or the total energy expended by an average person in performing each of a variety of
leisure activities), is a similar concept pertaining to household chores, is energy
expended by an average person in the occupation performed by the individual at issue,
is cigarette smoking, is age, is gender, and summarizes racial and ethnic
background.
In equation 3, are independent variables. The other variables:
are the EBRBs in general or the variables affected by other EBRBs
(namely health attitude, cancer knowledge and food behaviors). The later set of EBRBs
(health attitude, cancer knowledge and food behaviors) also affects obesity directly or
indirectly by affecting the variables included in equation 3. Thus, equation 3 might not be
complete. However, the goal of this systematic review is not to develop an extension to
equation 3 or to develop a new obesity model. The objective of this review is to
assimilate the empirical evidence on the effect of community-based behavior change
intervention designed to affect obesity by changing the EBRBs. The goal is to synthesize
such research from different discipline towards recommending a translational
community-based behavior change approach in obesity prevention program.
Texas Tech University, Janani R. Thapa, August 2014
140
4.4 Methods
The systematic review technique is fairly new in applied economics. Google
scholar search for meta-analysis and systematic review in the title of articles published in
the American Journal of Applied Economics had zero hits. The same search in the
American Journal of Public Health resulted in 19 hits for search term “systematic review”
and four hits for search term “meta-analysis” in the topic. Hence the method followed in
this analysis follows from methods applied in public health. The methods have been cited
from relevant publications in obesity and public health related journals like the Obesity
Reviews and the American Journal of Public Health.
4.4.1 Search strategy
A systematic review of published literature from earliest available to December of
2013 was done from January to March of 2014. The articles were planned to be identified
through searches in the following databases: MEDLINE (through PubMed), CENTRAL
(The Cochrane Library), Web of Science, EMBASE, and Google Scholar. However, a
similar systematic review was done in MEDLINE in 2002 (Fogelholm and Lahti-Koski,
2002), hence the search in MEDLINE was dropped. The search was limited to Google
Scholar and Web of Science due to time limitations.
A preliminary search with BMI and obesity in the topic and anywhere in the
article resulted in 467,000 hits in Google Scholar and 29,734 articles in the Web of
Science alone. Hence, in lieu of the search terms anywhere in the article the search was
objectively narrowed to include search terms ‘BMI’ ‘obesity’ ‘community’ and
‘behavior’ in the title and anywhere in the article. The objective was to review the effect
of community-based behavior change intervention on obesity behavior (EBRBs) and
Texas Tech University, Janani R. Thapa, August 2014
141
BMI however the search was done for ‘community’ instead of ‘community-based’ to
avoid missing articles that refer to their intervention as a community intervention instead
of community-based intervention. To ensure inclusion of all community-based
interventions, a second search in the data bases was done for the search terms ‘BMI’
‘obesity’ ‘community-based’ and ‘behavior.’ The search strategy was adapted according
to the requirements of individual databases.
4.4.2 Language and data restrictions
No language restrictions were applied during the search. The publication year
restrictions include publications from earliest records available to the December of 2013.
However, most of the articles on the research topic were found to be published after the
year 2000 (Appendix 4.1).
4.4.3 Selection criteria
Inclusion criteria were developed to address the problem of heterogeneity in
intervention type and outcome measures as suggested by Mulrow et al. (1997). The
research question was specific. The inclusion criteria ensure that the research was
conducted in a consistent manner and the methods applied was on a community-based
intervention and the outcome variable was behavior related to obesity and BMI. The
applied selection criteria for the study were: 1) it must have a predetermined focus on
changing variables related to obesity behavior –the EBRBs, 2) it must be carried out in a
community setting, or report findings of research being carried out in community setting,
and 3) the intervention must be a community-based intervention. The set of EBRBs
developed during the systematic review were self-efficacy, nutrition knowledge, stress,
Texas Tech University, Janani R. Thapa, August 2014
142
and lifestyle behavior. Articles reporting clinical trials with or without animals will not be
included. All types of publication available in the databases were included.
The exclusion criteria are behavior change through clinical trials, research
conducted outside a community setting, research that modifies food options and/or food
prices. The researchers are specifically interested in the success of translational
community-based research for behavior change towards obesity prevention without
confounding effect from other factors such as change in food price. Observational study
that did not have an intervention component such as Peracchio et al. (2012) were not
included. This review only includes intervention evaluation studies because the
researchers are specifically interested in the success of translational community-based
research for behavior change towards obesity prevention.
4.4.4 Data extraction and synthesis
Following the systematic search, a screening of titles and abstracts was done to
identify their potential inclusion in the review. The following data were extracted into
tables in excel. In the first screening authors, publication year, research objectives,
methodology and major results were extracted in a excel file. The extracted data file was
checked for completeness and accuracy and a final data file was made. A method of
narrative synthesis adapted and developed by Popay et al. (2007) was followed for this
review. This method has been followed by several researchers in the context of behavior
change including McMahon and Fleury (2012), Everson-Hock et al. (2013), Chisholm et
al. (2012), Gordon et al. (2011), and Skov et al. (2013). The findings that made it through
the final selection were grouped and narrative synthesis was applied to each group. This
systematic review process is shown in Figure 4.1.
Texas Tech University, Janani R. Thapa, August 2014
143
4.4.5 Validation of systematic review
An additional separate search with the search terms ‘community-based’
‘systematic review’ and ‘behavior change’ in the title was done in Google Scholar and
Web of Science to validate this search.
Texas Tech University, Janani R. Thapa, August 2014
144
Figure 4.1 The Systematic Review Process Used
Texas Tech University, Janani R. Thapa, August 2014
145
4.4.6 Meta-analysis
The selected papers were checked for data on number of participants, change in
weight status and BMI to conduct meta-analysis. The comprehensive meta-analysis
software following Gourlan et al. (2011) was used to run both the fixed and random
effects model. Under the fixed effect model it was assumed that there was one true effect
size which underlies all the studies in the analysis and the differences in observed effects
are due to sampling error. Under the random effects model the variance in true effect was
allowed to vary from one study to another (Borenstein et al., 2009). For example the
effect size might be higher or lower in studies where the participants are older or more
educated etc. A forest plot of standard difference in means was also computed. The
methods are closely followed from (Gonzalez-Suarez, 2009), a meta-analysis of school
based interventions in childhood obesity prevention.
4.5 Results
The search in Google Scholar and Web of Science resulted in 21,000 and 312 hits
respectively. From these hits, a careful selection with following the inclusion and
exclusion criteria resulted in 46 hits. Of these 46 hits, the analysis was done for 24
articles. The other 22 articles were not included for the reasons shown in Figure 4.1. A
summary of the selected articles is shown in Appendix 4.1.
In the initial screening of articles based on the inclusion and exclusion criteria
school based interventions were also included. However, the search revealed numerous
systematic reviews already been done for school-based intervention evaluation research
(e.g. Peterson and Fox, 2007 and Bluford et al., 2007). A systematic review on economic
incentives and nutritional behavior of children in the school setting has been done by
Texas Tech University, Janani R. Thapa, August 2014
146
Jensen et al. (2011). A review by Gittelsohn and Kumar (2007) concluded based on a
review of school-based obesity prevention interventions that the school-based obesity
prevention researchers needs improvement in its theoretical framework and the attention
should be changed towards larger randomized clinical trials. Further, environmental- and
community-based approaches should be more promising. Other reviews in childhood
obesity prevention through school based interventions are: DeMattia and Lee Denney
(2008), Kropski et al. (2008), Wofford (2008), Cook-Cottone et al. (2009), Hesketh and
Campbell (2010), Safron et al. (2011), Bleich et al. (2013), Showell et al. (2013), and
Thapa and Lyford (2014). For the reasons of abundant literature available on learning
from school-based obesity prevention program, school-based interventions were later not
included in this systematic review but could be included in the future for comparison of
school-based interventions with community-based interventions.
Some school based interventions that were included in the initial search based on
the inclusion and exclusion criteria but were later not discussed in this review are:
Gortmaker et al. (1999), Caballero et al. (2003), Neumark-Sztainer et al. (2003), Kain et
al. (2004), Wang et al. (2006), Taylor et al. (2007), Foster et al. (2008), Hughes et al.
(2008), Sanigorski et al. (2008), Gentile et al. (2009), Hoelscher et al. (2010), Chomitz et
al. (2010), Hollar et al. (2010), Crespo et al. (2012), and Moore et al. (2013). Further,
articles explaining the design of community-based research such as Broussard et al.
(1995) and Beckman et al. (2006) and study protocols of community-based research such
as de Silva-Sanigorski et al. (2010), Eisenmann et al. (2008), Shrewsbury et al. (2009),
and Taylor et al. (2013) are not included in the narrative synthesis and meta-analysis.
Texas Tech University, Janani R. Thapa, August 2014
147
4.5.1 Trend in community-based intervention evaluation research
Based on the findings of this systematic review, the trend in community-based
behavior change intervention evaluation research reports follows an increasing trend as
shown in Figure 4.2. The period of 2005 to 2009 saw an increase in the number of
community-based research including in the US. Most of the articles on the research topic
were found to be published after the year 2000.
Figure 4.2 Trend in Community-based Evaluation Research
Further validation search of Google Scholar and Web of Science with
‘community-based’ ‘systematic review’ and “behavior change” in the title resulted in
zero hits.
0
2
4
6
8
10
12
1990 to
1994
1995 to
1999
2000 to
2004
2005 to
2009
2010 to
2013
No. of articles in and
outside US
No. of articles in US
Texas Tech University, Janani R. Thapa, August 2014
148
4.5.2 Narrative synthesis
This narrative synthesis reports the effectiveness of community-based behavior
change interventions on weight status and energy balance related behaviors. The
reporting is done under two groups: 1) Small scale relatively more intense interventions,
2) Large scale relatively less intense interventions.
4.5.2.1 Small scale relatively more intense interventions
Promotoras (community health workers) program
This review included three promotoras based studies: Balcazar et al. (2009a),
Shaibi et al. (2012), and Schwartz et al. (2013). Promotoras based programs have
emerged as promising models to address reduction of risk factors in Mexican Americans
as cited in Balcazar et al. (2009a). Both the promotoras based community-based
interventions evaluation report that the effect was significant. Balcazar et al. (2009)
reported statistically significant difference between perceived benefits and two heart
healthy behaviors (salt and sodium, and cholesterol and fat), and Schwartz et al. (2013)
reported significant decrease in weight status from pre- to post- treatment. Similarly,
Shaibi et al. (2012) reported significant decrease in BMI z-score, BMI percentile, and
waist circumference. They also reported increase in cardiorespiratory fitness and
decreases in physical inactivity and dietary fat consumption. Balcazar et al. (2009) and
Shaibi et al. (2012) were evaluation of a pilot program and both of these including
Schwartz et al. (2013) were targeted for Latino adults.
Texas Tech University, Janani R. Thapa, August 2014
149
Family-based
Four articles included in this analysis were family-based intervention in a
community-based design: Robertson et al. (2008), Sacher et al. (2010), Smith et al.
(2010), and Bean et al. (2012). Statistically significant decrease in BMI is reported in
Sacher et al. (2010) and Smith et al. (2010). No statistically significant decrease in BMI
is reported in Robertson et al. (2008). However, statistically significant improvements
were observed in children’s quality of life and lifestyle, child parent relationship and
parents’ mental health without significant increase in fruit and vegetable consumption,
participation in moderate/vigorous exercise, and children’s self-esteem. Further Sacher et
al. (2010) also observed significant between group differences in cardiovascular fitness,
physical activity, sedentary behaviors and self-esteem. Similar observation is reported by
Smith et al. (2013). The target population in these interventions was children. Another
similar evaluation of a parent intervention for overweight children is Bean et al. (2012).
This study with (N= 41 families) found no significant treatment effect with few
improvements in dietary habit.
The analysis this far reveals that with focused population in small group
participating in an intensive community-based intervention with relatively less number of
direct beneficiaries (Appendix 4.1) have reported significant decrease in weight or BMI.
Number of participants (N) was 116 in Sacher et al. (2010), N was 440 in Smith et al.
(2010), and N was 15 in Shaibi et al. (2012). Significant decrease in weight or BMI was
not reported in Robertson et al. (2008), N was 27. Though significant decrease in BMI or
weight status was not observed across all studies synthesized this far, significant effect on
behavior change was observed in all reports synthesized this far.
Texas Tech University, Janani R. Thapa, August 2014
150
Faith- and culture-based
Five articles included in this analysis were faith- and culture-based interventions:
Heath et al. (1991), McNabb et al. (1997), Yancey et al. (2006), Goldfinger et al. (2008),
and Kim et al. (2008). The numbers of intervened participants were 30, 15, 366, 26, and
36 adults respectively. All reported significant decrease in weight and BMI except
Yancey et al. (2006). The populations were African American in McNabb et al. (1997),
Yancey et al. (2006), Goldfinger et al. (2008), and Kim et al. (2008), and Native
American in Heath et al. (1991). The results suggest that peer-led weight loss and
behavior change interventions in small groups in a community-based setting can show
impact in hard to reach populations. African Americans bear the highest burden of
obesity and yet are less likely to benefit from weight loss programs compared to Whites
(Kumanyika, 2002). However the effect appears to not be significant when the number of
beneficiaries was high (N=366) in Yancey et al. (2006).
Other small-scale, relatively intense interventions research reports that have
reported significant weight loss and were included in this analysis are Graffagnino et al.
(2006) and Folta et al. (2009). A modest yet significant reduction in BMI is also reported
in Nguyen et al. (2012). However, the research findings from the same study as reported
by Nguyen et al. (2012) had reported non-significant effect in BMI or BMI score
(O’connor et al., 2008). A significant treatment effect was also reported by Rejeski et al.
(2011) based on a lifestyle intervention in a community setting for elderly cardiovascular
disease survivor.
Texas Tech University, Janani R. Thapa, August 2014
151
4.5.2.2 Large scale relatively less intense interventions
Seven articles included in this analysis were less intense large scale community-
based behavior change interventions: (Jeffery, 1995), Crawford et al. (2000), McGuire et
al. (2001), Giachello et al. (2003), Jenum et al. (2006), Mohatt et al. (2007), Fotu et al.
(2011), and Rogers et al. (2013). The direct beneficiaries of this study ranged from larger
population sample group to communities. The number of respondents ranged from 394
respondents in Giachello et al. (2003) to 12 communities in Rogers et al. (2013) and 22
villages in Fotu et al. (2011). The longest project duration reported was by Jeffery (1995).
The reports of the findings from the Minnesota Heart Health Program (MHHP), a
13 year research and demonstration project was reported by Jeffery (1995) in this no
significant changes in weight were observed in the MHHP program. Similar non-
significant change in weight status was reported in Crawford et al. (2000). Both of these
study reports that compared to more intensive interventions weight changes were modest
but participation was large in these less intense large scale interventions. Similarly, Fotu
et al. (2011) concluded that community-based interventions in high obesity prevalence
area may require more intense or longer interventions.
Positive relationship between the degree of dietary restraint and lower weight was
shown by McGuire et al. (2001). Further this study showed methods developed to
increase behavioral and cognitive strategies have potential to control weight. Significant
increase in physical activity in the intervention group compared to control group was
reported by Jenum et al. (2006). This study concluded that theory driven low cost
population based intervention program could increase physical activity and reduce weight
gain. Similarly, Mohatt et al. (2007) also concluded that primary and secondary
Texas Tech University, Janani R. Thapa, August 2014
152
prevention efforts that focus on increasing the intrinsic strengths of indigenous
knowledge can bring positive public health transformations based on a community-based
participatory research in Alaska. Significant increase in fruits and vegetable consumption
and significant sugary drink consumption limitation by children was reported by Rogers
et al. (2013). This study suggests that a multi-setting, community-based intervention with
a consistent message can positively impact behaviors that lead to childhood obesity.
4.5.3 Meta-analysis
The descriptive data was available from 15 articles as shown in Table 4.1, the
length of treatment ranged from 8 week to 1 year. Maximum length of follow-up was
three years. However, only ten articles from the 25 articles selected for narrative
synthesis including the 15 articles with descriptive data had the complete information
needed for a meta-analysis (Figure 4.3). Further, McGuire et al. (2000) had three
independent treatment groups. Hence they were treated separately during the meta-
analysis.
Texas Tech University, Janani R. Thapa, August 2014
153
Table 4.1 Descriptive data of community-based behavior change evaluation reports
Study
Total
(N) Sex Age
Follow-
up (N)
Length of
treatment
Length of
follow-up
Heath et al. (1991) 406 M/F 54±15 406 37 wks 50 wks
Jeffery (1995) 280 M/ F 47 169 8 wks 1yr
McNabb et al. (1997) 39 F 56 39 14 wks NA
McGuire et al. (2000) 1226 M/ F 32.3±6.8 1044 NA NA
Crawford et al. (2000) 1226 M/ F 20 to 45 854 1 yr 3 yrs
Graffagnino et al. (2005) 418 M/ F 51.8±11.1 198 6 mnths 3 yrs
Yancey et al. (2006) 366 F NA 187 1 yr NA
Goldfiner et al. (2008) 26 M/ F 68 21 10 wks 1 yr
Folta et al. (2009) 110 F 57 55 12 wks NA
Sacher et al. (2010) 116 M/ F 10 37 12 wks 1 yr
Fotu et al. (2011) 1712 M/ F 14.5 815 NA 2.4 yrs
Nguyen et al. (2012) 151 M/ F 13 to 16 60 1 yr NA
Shaibi et al. (2012) 15 M/ F 15±0.9 15 12 wks NA
Schwartz et al. (2013) 477 M/ F 40 450 12 wks NA
Smith et al. (2013) 440 M/ F 6.1 (0.8) 274 10 wks NA
± CI, (SD), NA-not available, NS-not significant, M-male, F-female
Texas Tech University, Janani R. Thapa, August 2014
154
Table 4.2 shows the reported baseline and post-intervention body weight with
confidence interval (CI) or standard deviation (SD), the change in body weight, and
respective p values. The minimum change in body weight was .001lbs (insignificant) in
Shaibi et al. (2012) and the maximum change in body weight was a decrease by 10 lbs
(significant) in McNabb et al. (1997). Body weight increased post-intervention only in
Nguyen et al. (2012).
Texas Tech University, Janani R. Thapa, August 2014
155
Table 4.2 Body weight comparison from selected interventions
Study Unit Baseline Post-test Change p value
Heath et al. (1991) kg 80.00±13.00 NA -4.09±4.9 <0.05
Jeffery (1995)
Eat smart for your heart-2 kg 76.9 NA -0.5 NA
Eat smart for your heart-2 kg 79.7 NA -0.9 NA
BYFF kg 79.7 NA -0.9 NA
Wise weighs kg 75.5 NA -2.5 NA
McNabb et al. (1997) lbs 199.00±29.06 189±27.69 -10 ±10.28 <0.0001
McGuire et al. (2000)
-Restraint lbs 167.12±38.44 NA -0.74 0.001
-Flexible restraint lbs 167.12±38.45 NA -1.37 0.001
-Rigid restraint lbs 167.12±38.46 NA -1.15 0.001
Crawford et al. (2000) NA NA NA NA NS
Graffagnino et al. (2005) lbs 106 (28.8) 100.1 (27) -6 (8) <0.0001
Yancey et al. (2006) lbs 81.47 83.38 1.31 0.002
Goldfinger et al. (2008) lbs 194.3 185 -9.3 0.001
Folta et al. (2009) lbs 89.5 (18.8) 87.7 (19.6) -1.7(2.4) 0.010
Sacher et al. (2010) NA NA NA NA NA
Fotu et al. (2011) NA NA NA NA NA
Nguyen et al. (2012)
Loozit only kg 82.4 (12.4) 85.9 (13.4) 1.9 ± 4.7 NS
Loozit and ATC kg 84.2 (16.3) 88.1 (17.7) 1.9 ± 4.7 NS
Shaibi et al. (2012) kg 90.60 ± 6.80 89.9 ± 7.2 -0.001 0.440
Schwartz et al. (2013) lbs 179.76 176.86 -2.9 <.0001
Smith et al. (2013) NA NA NA NA NA
± CI and (SD) when available, NA-not available, NS-not significant
Texas Tech University, Janani R. Thapa, August 2014
156
Table 4.3 shows the reported baseline and post-intervention BMI with CI or SD,
the change in BMI, and p value. The minimum change in BMI was 0.1, insignificant
decrease in Sacher et al. (2010) and insignificant increase in Nguyen et al. (2012). The
maximum change in BMI was a decrease by 1.5 (significant) in Heath et al. (1991). The
BMI increased post-intervention only in Yancey et al. (2006), Fotu et al. (2009) and
Nguyen et al. (2012).
Texas Tech University, Janani R. Thapa, August 2014
157
Table 4.3 BMI comparison in selected interventions
Study Baseline Post-test Change p value
Heath et al. (1991) 31 29.50 -1.50±2.20 <0.05
McNabb et al. (1997) 33.90±5.41 32.50±5.41 -1.40±1.61 <0.02
McGuire et al. (2000)
Restraint 27.02±5.74 NA -0.13 0.001
Flexible restraint 27.02±5.75 NA -0.26 0.001
Rigid restraint 27.02±5.76 NA -0.21 0.001
Crawford et al. (2000) 27.70 (4.70) NA NA NS
Graffagnino et al. (2005) 37.50±8.80 NA NA NA
Yancey et al. (2006) 29.72 (6.36) 30.48 0.76 <0.0001
Goldfiner et al. (2008) 32.70 30.90 -1.80 0.001
Folta et al. (2009) 33.20 (5.70) 32.60 (6) -0.60(0.90) 0.01
Sacher et al. (2010) NA NA -0.10 0.70
Fotu et al. (2011) 22.90 (4.10) 25.20 (4.20) 2.30 NA
Nguyen et al. (2012)
Loozit only 30.80 (3.50) 30.80 (3.80) 0.10 (-0.30 to 0.40) NA
Loozit and ATC 30.80 (4.20) 31.40 (4.80) 0.10 (-0.30 to 0.40) NA
Shaibi et al. (2012) 32.50 ±1.60 32.00±1.70 -1.50 0.06
Schwartz et al. (2013) 31.71 31.19 -0.52 <0.0001
Smith et al. (2013) 22.50 (3.60) 22.10 (3.70) -0.50 (-0.60 to -0.40) <0.0001
± CI, (SD), NA-not available, NS-not significant
Texas Tech University, Janani R. Thapa, August 2014
158
All the literature that reported the standard difference in mean, sample size and
paired p value were selected. The total number of individuals covered in this analysis was
5673. The follow-up data for intervention group was available for 4616 individuals. The
data used for the meta-analysis is shown in Table 4.4 and the output is shown in Figure
4.3.
Texas Tech University, Janani R. Thapa, August 2014
159
Table 4.4 Data for meta-analysis
Study Diff. in means Sample size
Paired groups
p value Tail Effect direction
Heath et al. (1991) 1.500 406 0.050 2 Negative
McNabb et al. (1997) 1.400 39 0.020 2 Negative
McGuire et al. (2000a) 0.130 1044 0.001 2 Negative
McGuire et al. (2000b) 0.260 1044 0.001 2 Negative
McGuire et al. (2000c) 0.211 1044 0.001 2 Negative
Yancey et al. (2006) 0.760 187 0.001 2 Positive
Goldfiner et al. (2008) 1.800 21 0.001 2 Negative
Folta et al. (2009) 0.600 55 0.010 2 Negative
Sacher et al. (2010) 0.100 37 0.700 2 Negative
Shaibi et al. (2012) 1.500 15 0.060 2 Negative
Schwartz et al. (2013) 0.520 450 0.001 2 Negative
Smith et al. (2013) 0.500 274 0.001 2 Negative
Figure 4.3 shows the forest plot of the standard difference in mean, the dots and
95% CI. The meta-analysis shows that the intervention had modest yet significant
decrease in BMI in both fixed and random effects model. The fixed effect model standard
difference in mean BMI is -0.109, 95% CI = -0.139, -0.080 and random effects model
standard difference in mean BMI is -0.134, 95% CI = -0.216, -0.052. This meta-analysis
provided evidence that community-based behavior change interventions are favorable for
obesity prevention. Similar meta-analysis for school-based obesity prevention
intervention by Gonzalez-suarez et al. (2009) found convincing evidence that school-
Texas Tech University, Janani R. Thapa, August 2014
160
based interventions are effective. The odds of participants being overweight or obese in
the school based intervention programs compared with the control were significantly
protective. Gonzalez-suarez et al. (2009) further found that interventions conducted for
more than one year had a higher odds ratio of decreasing the prevalence of obesity.
Texas Tech University, Janani R. Thapa, August 2014
161
Model Study
Statistics for each study Std. diff. in means and 95% CI
Std.
diff in
means
Std.
error Variance
Lower
limit
Upper
limit
Z
value p value -1.00 -0.50 0.00 0.50 1.00
Heath et al. -0.098 0.050 0.002 -0.195 -0.000 -1.961 0.050
McNabb et al. -0.389 0.166 0.028 -0.714 -0.063 -2.342 0.019
McGurie et al. -0.102 0.031 0.001 -0.163 -0.041 -3.291 0.001
McGurie et al. -0.102 0.031 0.001 -0.163 -0.041 -3.291 0.001
McGurie et al. -0.102 0.031 0.001 -0.163 -0.041 -3.291 0.001
Yancey et al. 0.332 0.075 0.006 0.185 0.479 4.422 0.000
Goldfinger et al. -0.840 0.254 0.064 -1.337 -0.343 -3.310 0.001
Folta et al. -0.360 0.139 0.019 -0.633 -0.087 -2.587 0.010
Sacher et al. -0.064 0.165 0.027 -0.386 0.259 -0.388 0.698
Shaibi et al. -0.528 0.276 0.076 -1.069 0.012 -1.917 0.055
Schwartz et al. -0.185 0.048 0.002 -0.278 -0.092 -3.893 0.000
Smith et al. -0.239 0.061 0.004 -0.359 -0.118 -3.894 0.000
Fixed -0.109 0.015 0.000 -0.139 -0.080 -7.394 0.000
Random -0.134 0.042 0.002 -0.216 -0.052 -3.216 0.001
Figure 4.3 Forest Plot of the Standard Difference in Mean BMI
Texas Tech University, Janani R. Thapa, August 2014
162
4.5.4 Large-scale, less-intense or small-scale, more-intense interventions?
Impacts of small scale relatively intense intervention suggest that significant
weight loss can be achieved. The relapse and follow-up effect of such interventions
included in this synthesis and meta-analysis have not been reported from small scale
interventions. According to Sacher et al. (2010) the limitation of these types of studies
was a short follow-up time which leaves the long term intervention impact unclear. The
significant impact in the initial months after intervention may relapse (Yancey et al.
(2006). The impacts of large scale less intense community-based interventions suggest
that a consistent decrease in BMI from pre- to post- intervention was not observed.
However, there were positive changes in attitudes and behavior. For eg. Rogers et al.
(2013) reported significant increase in fruits and vegetable consumption. Additionally, a
review published by Jeffery (1993) concluded that multiple health behavior intervention
is effective in behavior change.
A community originated community-based intervention that was not included due
to misfit in the narrative synthesis is the Whetstone et al. (2011) study of North Carolina.
This study assesses the change in children’s health behaviors and weight status after
participation in a community originated intervention. This study was unique and needs
consideration because the intervention was community originated. The conclusion of this
research was that the community originated interventions to raise awareness about food
choices and to change policies and environments may improve BMI z-scores. Other
similar hits that were not included are Samuels et al. (2010) and Cheadle et al. (2010)
which describes the approaches used to measure the extent and impact of environmental
changes in three community level obesity prevention initiatives in California. The effort
Texas Tech University, Janani R. Thapa, August 2014
163
is to provide empirical evidence from intervention studies. These reports conclude that
the Healthy Eating, Active Communities (HEAC) programs successfully changed
children’s food and physical activity environments.
The current systematic review and meta-analysis suggest that change in BMI and
wegiht status through community-based behavior change intervention has promise.
However, as shown in Chapter one of this dissertation cost-effectiveness analysis should
be considered in deciding between low per unit cost long term trials or focused short term
interventions. This research has potential to bring a needed discussion in the area of
obesity prevention through community-based intervention in designing large scale less
intense intervention or small scale more intense interventions.
4.6 Conclusions
A systematic review and meta-analysis of community-based intervention was
done with an objective to synthesize research on community-based behavior change
interventions implemented for obesity prevention. The search was done in Google
Scholar and Web of Science. The analysis accomplished both a narrative analysis and
meta-analysis. Narrative analysis was done for all the articles that fulfilled the selection
criteria including the articles included in the meta-analysis. A meta-analysis was done for
articles that had the needed information to conduct the meta-analysis. The narrative
synthesis and meta-analysis reveal that community-based behavior change interventions
have promise to change behavior towards obesity prevention and to change BMI. The
overall fixed and random effect of such intervention show a modest yet significant
decrease in BMI. Based on the findings of this research, suffcient research and focus
whould be given by designers of such intervention. This research is suggestive that
Texas Tech University, Janani R. Thapa, August 2014
164
designers should carefully consider among large scale less intense and small scale more
intense interventions. The current research provides a significant contribution to obesity
prevention literature by synthesizing the impact of community-based obesity prevention
research from different disciplines and creating a need in this area to focus on cost-
effectiveness of behavior change interventions.
This review could not include many research findings ( e.g. Epstein et al., 1984;
Johnson et al., 1997; Delahanty et al., 2002 and Klesges et al., 2008) though it was very
close to community-based intervention due to lack of identification as a community-
based intervention. Similarly other such un-included research are Pratt et al. (2007)
because it is a review report of design characteristics of worksite environmental
interventions. A future research in the area to be observed is the Growing Right Onto
Wellness (GROW) program. Po’e et al. (2013) has reported the concept of GROW
community-based research. Another important literature on family, community and
clinic collaboration to treat overweight and obese children to see is Robinson et al.
(2013). In future, a more comprehensive research that will conduct review, synthesis and
meta-analysis for effectiveness of community-based intervention by the socioeconomic
status of participants would also be beneficial to the literature as well.
Texas Tech University, Janani R. Thapa, August 2014
165
Appendix
Appendix 4.1
A Randomized Community Intervention to Improve Hypertension Control among
Mexican Americans: Using the Promotoras de Salud Community Outreach Model
(Balcazar et al., 2009)
Setting, location Mexican Americans of the El Paso, Texas
Participants, activities and
duration
Diagnosed with hypertension (N: I=58, C=40)
Mean age 54.9
9 week promotora intervention
Design Randomized feasibility pilot intervention
Theory/model Promotoras de salud community outreach model
Main EBRBs measured
Clinical measures of blood pressure, BMI, Waist
circumference
Self-reported behaviors associated with high blood
pressure
Attitudes and beliefs around blood pressure
Effectiveness
Statistically significant difference between perceived
benefits and two heart healthy behaviors (salt and sodium,
and cholesterol and fat)
Dietary intake in a randomized-controlled pilot of NOURISH: A parent intervention
for overweight children (Bean et al., 2012) *
Setting, location Virginia
Participants, activities and
duration
Parents of overweight and obese children (N: I=46,
C=50)
2008 to 2009
Design Randomized controlled pilot
Theory Social cognitive theory
Main EBRBs measured Dietary assessment of parents and children conducted at
baseline, post-test, and 6-month follow-up
Effectiveness
Among parents who self-select into a childhood obesity
program, minimal intervention can elicit short-term dietary
changes comparable to those of a structured intervention.
Can anyone successfully control their weight? Findings of a three year community-based
study
of men and women (Crawford et al., 2000)
Setting, location Theory
Participants, activities and
duration
Aged 20-45 (N=854 of 1226 enrolled), non-pregnant
3 yrs
Design Paired group design, randomized control
Theory NA
Main EBRBs measured Body weight, weight control behavior, dietary intake, fast
food consumption, physical activity, television viewing
Effectiveness Without much greater efforts to promote and support
weight control, most people will be unable to avoid weight
Texas Tech University, Janani R. Thapa, August 2014
166
gain and very few will manage to lose weight, and as a
consequence the prevalence of obesity will continue to
rise.
The StrongWomen–Healthy Hearts Program: Reducing Cardiovascular Disease Risk
Factors in Rural Sedentary, Overweight, and Obese Midlife and Older Women (Folta et
al., 2009)
Setting, location 8 counties in Arkansas and Kansas
Participants, activities and
duration
Sedentary midlife and older women
12 week twice-weekly heart health program
55 respondents followed for 12 weeks
Design Randomized (by community) controlled trial
Theory Social cognitive theory
Main EBRBs measured Weight, waist circumference, diet, physical activity and
self-efficacy
Effectiveness
Our results suggest that a community-based program can
improve self-efficacy, increase physical activity, and
decrease energy intake, resulting in decreased waist
circumference and body weight among at-risk women.
Outcome results for the Ma’alahi Youth Project, a Tongan community-based obesity
prevention programme for adolescents (Fotu et al., 2011)
Setting, location Tonga
Participants, activities and
duration
Three year
22 villages in three districts
Design Quasi-experimental
Theory NA
Main EBRBs measured
Social marketing and capacity building
Education activities that promote physical activity and
local fruits and vegetables
Effectiveness
No impact on the large increase in prevalence of
overweight and obesity among Tongan adolescents.
Community-based interventions in such populations with
high obesity prevalence may require more intensive or
longer interventions, as well as specific strategies targeting
the substantial socio-cultural barriers to achieving a
healthy weight.
Project HEAL: Peer Education Leads to Weight Loss in Harlem (Goldfinger et al., 2008)
Setting, location Minority Harlem community, New York city
Participants, activities and
duration
April to June 2006
Two trained peer leaders led the 8 sessions at the
church.
Ten weeks after enrollment, at the eighth and final
session of the course, and at 22 and 32 weeks and 1 year
after enrollment, trained research assistants, who were
blinded to patients’ baseline weights, surveyed and
weighed participants.
Texas Tech University, Janani R. Thapa, August 2014
167
Design NA
Theory NA
Main EBRBs measured Pre and post course weights, self-reported behaviors and
quality of life.
Effectiveness
A peer-led, community-based course can lead to weight
loss and behavior change. Significant decrease in weight
and BMI. The minority communities most affected by
obesity and diabetes may benefit from this low-cost,
culturally appropriate intervention.
Effect of a Community-Based Weight Management Program on Weight Loss and
Cardiovascular Disease Risk Factors (Graffagnino et al., 2006)
Setting, location Community medical wellness center
Participants, activities and
duration
2001 to 2004
N= 198 (142 women, 56 men)
Design Retrospective outcomes analysis
Theory NA
Main EBRBs measured Anthropometric measures, program participation
Effectiveness
There was a significant correlation between percentage
weight loss and number of weekly counseling sessions
attended and number of visits to the wellness center for
exercise.
The relatively high drop-out rate associated with this
program suggests the need to identify strategies and
techniques to enhance adherence and completion of
programs.
Community-based exercise and weight control: diabetes risk reduction and glycemic
control in Zuni Indians (Heath et al., 1991)
Setting, location Southwest New Mexico
Participants, activities and
duration
Zuni Indians, 2 years
Two 1-hr aerobics exercise session per week. It has grown
to
> 48 sessions 5 days/week several times daily
Design NA
Theory NA
Main EBRBs measured Weight, BMI, fasting blood glucose level
Effectiveness
Demonstrates the effectiveness of a community-based
exercise program in facilitating weight loss and improved
metabolic control in a Native American Indian group.
Community Programs for Obesity Prevention: The Minnesota Heart Health Program
(Jeffery, 1995)
Setting, location Minnesota
Participants, activities and
duration
Adult education classes
Worksite weight control program
Weight loss by home correspondence
Weight gain prevention program
Texas Tech University, Janani R. Thapa, August 2014
168
6 communities
Design NA
Theory NA
Main EBRBs measured Weight
Effectiveness
No significantly positive effect on weight status, ceiling
effect. Too many things at once must have distracted from
attaining a single goal. Clearly informing people is not
sufficient.
Promoting Physical Activity in a Low-Income Multiethnic District: Effects of a
Community Intervention Study to Reduce Risk Factors for Type 2 Diabetes and
Cardiovascular Disease (Jenum et al., 2006)
Setting, location Oslo, Norway
Participants, activities and
duration
N: Baseline= 2950 30-65 years old, Follow-up= 1776)
Promotion of physical activity
Design Pseudo experimental cohort design
Theory Social cognitive theory and social ecological models
Main EBRBs measured BMI, Non fasting blood sample
Effectiveness
A net increase in physical activity of 9%, corresponding to
25% relative reduction in the proportion of inactive people
in the intervention district compared with a 5% reduction
in the control district.
The WORD (Wholeness, Oneness, Righteousness, Deliverance): A Faith-Based Weight-
Loss
Program Utilizing a Community-Based Participatory Research Approach (Kim et al.,
2008)
Setting, location North Carolina
Participants, activities and
duration
N=73, I= 36, C=37
Weekly small groups led by trained community members
Healthy nutrition, physical activity, and faith’s connection
with health
Design Two-group, quasi –experimental delayed intervention
design
Theory Social support theory and stages of change trans-
theoretical model
Main EBRBs measured Anthropometric measurement, health-behavior variables
Effectiveness
The mean weight loss of the treatment group was 3.60 ±
0.64 lbs. compared to the 0.59 ± 0.59-lb loss of the control
group
The relationship between restraint and weight and weight-related behaviors among
individuals in a community weight gain prevention trial (McGuire, 2001)
Setting, location
Minneapolis/St Paul, Minnesota metropolitan area through
newspaper advertisements, mailings to employees at the
University of Minnesota, and direct telephone recruitment
Participants, activities and
duration
N=1226
3 year weight gain prevention study
Texas Tech University, Janani R. Thapa, August 2014
169
Participants of Pound of Prevention (POP) study
Low-intensity intervention for preventing age-related
weight gain.
Design NA
Theory NA
Main EBRBs measured Weight, weight contributing behaviors, physical activity
Effectiveness
Developing methods to increase behavioral and cognitive
strategies to control weight may help to prevent weight
gain in clinical and community samples
The PATHWAYS Church-Based Weight Loss Program for Urban African-American
Women at Risk for Diabetes (McNabb et al., 1997)
Setting, location NA
Participants, activities and
duration
Urban Africa-American women
N= 39, I=15, C=18
Design NA
Theory NA
Main EBRBs measured Weight, behaviors
Effectiveness
Apart from reaching a wider audience, administering the
program in a community-based setting did not lead to any
observable differences in outcomes compared with those
obtained in the clinic administration.
The Center for Alaska Native Health Research Study: A Community-based Participatory
Research Study of Obesity and Chronic Disease-related Protective and Risk Factors
(Mohatt et al., 2007)
Setting, location Southwestern Alaska
Participants, activities and
duration
N=753 of 922 in baseline Alaska Native
Design NA
Theory NA
Main EBRBs measured Anthropometric measurements
Effectiveness
The results strongly indicate that solution focused research
utilizing primary and secondary prevention strategies may
provide evidence to prevent further incidence of chronic
disease.
Loozit study
Twelve-Month Outcomes of the Loozit Randomized Controlled Trial. A Community-
Based Healthy Lifestyle Program for Overweight and Obese Adolescents (B. Nguyen et
al., 2012)
Setting, location Community health center and children’s hospital in
Sydney, Australia.
Participants, activities and
duration
A total of 151 overweight or obese 13- to 16-year-olds.
Design 2-arm randomized controlled trial
Theory NA
Main EBRBs measured BMI, waist to height ratio
Texas Tech University, Janani R. Thapa, August 2014
170
Effectiveness
The Loozit randomized controlled trial produced a
significant but modest reduction in body mass index z
score and improved psychosocial outcomes at 12 months.
Evaluation of a community-based weight management program for overweight and obese
adolescents: The Loozit study (O’Connor et al., 2008)
Setting, location NA
Participants, activities and
duration
Group-based intervention over five months
Design Short term effectiveness trial
Theory Cognitive behavioral approach
Main EBRBs measured
Self-efficacy, motivation, perseverance
Anthropometry, metabolic profile, self-report
questionnaire
Effectiveness
The Loozit study demonstrates for the first time that it is
feasible to perform measurement assessments and retain
overweight and obese adolescents to a weight management
program based in a community setting.
Translating Weight Loss and Physical Activity Programs Into the Community to Preserve
Mobility in Older, Obese Adults in Poor Cardiovascular Health (Rejeski et al., 2011)
Setting, location NA
Participants, activities and
duration
Ambulatory, overweight or obese, community-dwelling
older adults who either had CVD or cardio-metabolic
dysfunction and evidence of self-reported limitations in
mobility
Design Randomized controlled trial
Theory Social cognitive theory
Main EBRBs measured Time to complete a 400-m walk in seconds
Effectiveness
Existing community infrastructures can be effective in
delivering lifestyle interventions to enhance mobility in
older adults in poor cardiovascular health with deficits in
mobility
Pilot of "Families for Health'': community-based family intervention for obesity
(Robertson et al., 2008)
Setting, location Coventry, England
Participants, activities and
duration
27 overweight or obese children aged 7-13 years (18 girls,
9 boys) and their parents, from 21 families.
“Families for Health” is a 12-week program with parallel
groups for parents and children, addressing parenting,
lifestyle change and social and emotional development.
Design NA
Theory NA
Main EBRBs measured BMI z-score
Effectiveness
BMI z score was reduced by 20.18 (95% CI-20.30 to-0.05)
at 3 months and-20.21 (-0.35 to-0.07) at 9 months.
Statistically significant improvements were observed in
children's quality of life and lifestyle (reduced sedentary
Texas Tech University, Janani R. Thapa, August 2014
171
behavior, increased steps and reduced exposure to
unhealthy foods), child-parent relationships and parents'
mental health. Fruit and vegetable consumption,
participation in moderate/ vigorous exercise and children's
self-esteem did not change significantly. Topics on
parenting skills, activity and food were rated as helpful and
used with confidence by most parents.
Impact of Let’s Go! 5-2-1-0: A Community-Based, Multi-setting Childhood Obesity
Prevention Program (Rogers et al., 2013)
Setting, location Maine
Participants, activities and
duration
12 communities
Repeated random telephone
Surveys with 800 parents of children to measure
awareness of messages and child behaviors. Surveys
were conducted in schools, child care programs, and
afterschool programs to track changes in policies and
environments.
Design NA
Theory Social ecological theory
Main EBRBs measured Awareness knowledge and behavior change
Effectiveness
A multi-setting, community-based intervention with a
consistent message can positively impact behaviors that
lead to childhood obesity
MEND program
Randomized Controlled Trial of the MEND Program: A Family-based Community
Intervention for Childhood Obesity (Sacher et al., 2010)
Setting, location NA
Participants, activities and
duration
N=11 obese children (BMI ≥ 98th percentile
Randomly assigned to intervention or waiting list control
(6-month delayed intervention).
Parents and children attended eighteen 2-h group
educational and physical activity sessions held twice
weekly in sports centers and schools, followed by a 12-
week free family swimming pass.
Design Randomized control design
Theory NA
Main EBRBs measured
Waist circumference, BMI, body composition, physical
activity level, sedentary activities, cardiovascular fitness,
and self-esteem were assessed at baseline and at 6 months
Effectiveness
Participants in the intervention group had a reduced waist
circumference z-score (−0.37; P < 0.0001) and BMI z-
score (−0.2 ; P < 0.0001) at months when compared to
the controls. Significant between-group differences were
also observed in cardiovascular fitness, physical activity,
sedentary behaviors, and self-esteem.
Assessing the short-term outcomes of a community-based intervention for overweight
Texas Tech University, Janani R. Thapa, August 2014
172
and obese children:The MEND 5-7 program (Smith et al., 2013)
Setting, location Community venues at 37 locations across the UK.
Participants, activities and
duration
440 overweight or obese children (42% boys; mean age 6.1
years; body mass index (BMI) z-score 2.86) and their
parents/carers participated in the intervention.
MEND 5-7 is a 10-week, family-based, child weight-
management intervention consisting of weekly group
sessions. It includes positive parenting, active play,
nutrition education and behavior change strategies.
Design Repeated measures.
Theory NA
Main EBRBs measured NA
Effectiveness
Participation in the MEND 5-7 program was associated
with beneficial changes in physical, behavioral and
psychological outcomes for children with complete sets of
measurement data, when implemented in UK community
settings under service level conditions.
Family-Based Risk Reduction of Obesity and Metabolic Syndrome: An Overview and
Outcomes of the Idaho Partnership for Hispanic Health (Schwartz et al., 2013)
Setting, location Rural Idaho
Participants, activities and
duration
Compañeros en Salud (CeS) is a promotora- led wellness
program
Improve nutrition and physical activity behaviors as well
as increase community support and infrastructure for
healthy living.
Design NA
Theory NA
Main EBRBs measured weight, BMI, metabolic syndrome risk, A1c, glucose,
blood pressure, and cholesterol
Effectiveness
CeS model as a promising best practice for effecting
individual and family- level physiologic and behavioral
outcomes for obesity prevention.
Effects of a Culturally Grounded Community-Based Diabetes Prevention Program for
Obese Latino Adolescents (Shaibi et al., 2012)
Setting, location NA
Participants, activities and
duration
Fifteen obese Latino adolescents (body mass index
percentile = 96.3 ± 1.1, age = 15.0 ± 0.9 years)
completed a 12-week intervention that included weekly
lifestyle education classes delivered by bilingual/
bicultural promotoras and three, 60-minute physical
activity sessions per week.
Design NA
Theory Social cognitive theory
Main EBRBs measured
Anthropometrics (height, weight, BMI, and WC)
Cardiorespiratory fitness, physical activity/ inactivity,
nutrition behaviors, and insulin sensitivity and glucose
Texas Tech University, Janani R. Thapa, August 2014
173
tolerance by a 2-hour oral glucose tolerance test.
Effectiveness
The intervention resulted in significant decreases in BMI z
score, BMI percentile, and waist circumference; increases
in cardiorespiratory fitness; and decreases in physical
inactivity and dietary fat consumption.
Challenges in Improving Fitness: Results of a Community-Based, Randomized,
Controlled Lifestyle Change Intervention (Yancey et al., 2006)
Setting, location NA
Participants, activities and
duration
366 healthy, obese African American women
8-week culturally targeted nutrition and physical activity
intervention on body composition.
Data were collected at baseline, 2, 6, and 12 months
Design randomized, attention-controlled, two-group trial
Theory NA
Main EBRBs measured NA
Effectiveness
The intervention produced modest short-term
improvements in body composition, but the economic
incentive of a free 1-year gym membership provided to all
participants was a more potent intervention than the
education and social support intervention tested. However,
longer-term fitness enhancement remains elusive and
demands research and policy attention.
Texas Tech University, Janani R. Thapa, August 2014
174
CHAPTER V
SUMMARY AND CONCLUSIONS
Obesity is pervasive and has become a major public health concern. It was
traditionally addressed as an individual’s concern. However, with the increasing rates of
obesity there is a strong effort to address obesity through community-based approaches to
achieve effects on a societal level. Community-based interventions for obesity prevention
commonly have targeted reduction in body weight usually measured by an individual’s
body mass index (BMI). However, changing body weight without relapse has been a
challenge in obesity prevention. With this, the need to identify successful strategies to
design effective and efficient community-based interventions is increasing.
This dissertation develops a conceptual framework for addressing obesity as a
market failure of unhealthy foods in the first chapter. Obesity is largely an outcome of
increased demand and supply of unhealthy foods. However, the current market price of
unhealthy foods does not include the societal cost of unhealthy food i.e. the cost of
obesity. Hence obesity is conceptualized as a market failure of unhealthy foods. A
negative externality in the production and consumption of unhealthy foods those are high
in fat, sugar, and salt. The framework looks at applying behavioral tools to decrease the
demand of unhealthy foods that will lead to decrease in the social cost of unhealthy foods
and will in the long run decrease the externality effect i.e. the cost of obesity. Further, this
framework develops the argument for applying low cost large scale methods such as
community-based behavior change interventions to increase the demand of healthy foods
while decreasing the demand of unhealthy foods. This market failure framework for the
obesity problem leads to the potential of behavioral interventions for obesity prevention.
Texas Tech University, Janani R. Thapa, August 2014
175
Behavioral interventions have a potential to complement the ongoing efforts in
changing the supply and demand of products that are of concern to public health, for
example taxation to change the demand of addictive products such as cigarettes and
alcohols or perhaps even unhealthy foods. Obesity prevention can be achieved through
behavior change by complementary incorporation of behavioral economics with
traditional economics. This could lead to better addressing the obesity epidemic by
designing an integrated obesity prevention effort that incorporates all the aspects of
obesity.
This dissertation further looks at the impact of a community-based multi-tiered
behavior change intervention on cancer risk awareness and BMI, and the effect of healthy
eating signs in supermarket with community reinforcement on supermarket sales in the
second and third chapters respectively. The concluding research (in the fourth chapter) of
this dissertation is a systematic review and meta-analysis to look at the impact of the
available community-based behavior change interventions on behavior change towards
obesity prevention and weight loss (decrease in BMI).
The second chapter evaluates the project outcomes of a community-based multi-
tiered behavior change intervention by comparing food behavior and attitude before and
after the project and by analyzing the intervention effect on cancer knowledge and BMI.
The methods of analysis were adapted for this purpose from common health promotion
impact evaluation methods. The analysis used data available from five questionnaire
surveys, three in the intervention community and two in the control community.
The results from second chapter suggest that the intervention was successful in
significantly and sustainably changing health attitude and cancer knowledge. The results
Texas Tech University, Janani R. Thapa, August 2014
176
also suggested that long term interventions are more promising than short term
interventions in changing BMI. While attitude and knowledge could be changed in short
term sustainable weight change requires longer interventions. Community-based
interventions such as these have strong potential in bringing long term change in obesity
through targeting the behaviors associated with obesity rather than by targeting decrease
in body weight and will have sustainable impact.
Community-based models that incorporate supermarket in program activities are
rare despite its potential effectiveness. The third chapter reports the findings of a
community-based program specifically the project effect on food purchase behavior. The
analysis is based on monthly sales (in units sold per month) data available from the
participating supermarket. The data were available separately for the produce items and
healthier items identified with shelf talkers. The method was based on the concept of
modifying both rapid and prescriptive food choice decision, the former targeted to be
affected by shelf talkers and the later targeted to be affected by the overall promotion of
healthful eating for obesity prevention.
The overall analysis in the third chapter suggests that the project had a positive
impact on purchase of fresh produce, mainly on green leafy vegetables. The effect was
also significant in the sales of whole grain pasta and sauce. The findings suggest that
supermarket interventions have potential in changing food behavior. However, the lack of
positive change across all categories also suggests the challenges associated with
changing food behavior.
The fourth chapter reports the findings of a systematic review and meta-analysis
of community-based interventions. This paper is a synthesis of research on community-
Texas Tech University, Janani R. Thapa, August 2014
177
based behavior change interventions implemented for obesity prevention. A method of
systematic review followed by narrative synthesis and meta-analysis was done. Both the
narrative synthesis and meta-analysis reveal that community-based behavior change
interventions have promise to change behavior towards obesity prevention and to some
extent to change (decrease) BMI. Additionally, the difference in project impact between
large-scale, less-intense and small-scale more-intense interventions suggests that
designers of community-based obesity prevention interventions should carefully consider
between large-scale, less-intense and small-scale more-intense interventions. There
appears to be trade off in outcomes and cost-effectiveness between large-scale, less-
intense and small-scale more-intense interventions.
Overall, addressing obesity as a market failure to design behavioral intervention
effort to change (decrease) demand of unhelathy food provides a useful framework to
study obesity prevention in the realm of market structure. Further, behavior change
interventions through community-based interventions showed positive results with a
supermarket component. Supermarkets as the center of such intervention has promise in
changing food choice decisions in communities.
This dissertation however had some limitations. The cost effectiveness of the
intervention was not analysed though the necessity to develop cost effective methods
have been discussed in the first chapter. Further, in analyzing the effect of intervention in
supermarket sales the effect of shelf talker signs could not be isolated from the overall
community and supermarket based healthy promotion messages.
Texas Tech University, Janani R. Thapa, August 2014
178
BIBLIOGRAPHY
AAFP. (2009). AIM-HI Fitness Inventory. [Retrieved May 05, 2014, from
http://www.aafp.org/dam/AAFP/documents/patient_care/fitness/AIMFitnessInv-
2.pdf]
ACS. (2012). Cancer Facts and Figures 2011. Atlanta.
AICR. (2007). American Institute for Cancer Research, World Cancer Research Fund.
Food, nutrition, physical activity and the prevention of cancer. Washington, D.C.
Allen, M. L., Hurtado, G. A., Yon, K. J., Okuyemi, K. S., Davey, Cynthia S., Marczak,
M. S., Stoppa, P., and Svetaz, V. M. (2013). Feasibility of a Parenting Program to
Prevent Substance Use Among Latino Youth: A Community Based Participatory
Research Study. American Journal of Health Promotion, 27(4), 240–244.
Anand, P., and Gray, A. (2009). Obesity as market failure: could a “deliberate economy”
overcome the problems of paternalism. Kyklos International Review for Social
Sciences, 62(2), 182–190.
Anderson, N. B. (Ed.). (2004). Encyclopedia of Health and Behavior. American
Psychological Association, Thousand Oaks, CA.
Andreyeva, T., and Luedicke, J. (2013). Federal Food Package Revisions Effects on
Purchases of Whole-Grain Products. American Journal of Preventive Medicine,
45(4), 422–429.
Apple, R. D. (2003). Food Politics: How the Food Industry Influences Nutrition and
Health (review). Technology and Culture, 44(4), 839–841.
Ard, J. D., Fitzpatrick, S., Desmond, R. A., Sutton, B. S., Pisu, M., Allison, D. B., …
Baskin, M. L. (2007). The Impact of Cost on the Availability of Fruits and
Vegetables in the Homes of Schoolchildren in Birmingham, Alabama. American
Journal of Public Health, 97(2), 367–372.
Avery, K. N. L., Donovan, J. L., Horwood, J., and Lane, J. A. (2012). Behavior theory for
dietary interventions for cancer prevention: a systematic review of utilization and
effectiveness in creating behavior change. Cancer Causes and Control, 24(3),
409–420.
Baker, E. A., and Brownson, C. A. (1998). Defining characteristics of community-based
health promotion programs. Journal of Public Health Management and Practice
JPHMP, 4(2), 1–9.
Texas Tech University, Janani R. Thapa, August 2014
179
Balcazar, H. G., Byrd, T. L., Ortiz, M., Tondapu, S. R., and Chavez, M. (2009). A
randomized community intervention to improve hypertension control among
Mexican Americans: using the promotoras de salud community outreach model.
Journal of Health Care for the Poor and Underserved, 20(4), 1079–94.
Baranowski, T., Allen, D. D., Mâsse, L. C., and Wilson, M. (2006). Does participation in
an intervention affect responses on self-report questionnaires? Health Education
Research, 21 Suppl 1(October), i98–i109.
Baranowski, T., Anderson, C., and Carmack, C. (1998). Mediating variable framework in
physical activity interventions How are we doing? how might we do better?
American Journal of Preventive Medicine, 15(4), 266–297.
Baranowski, T., Cerin, E., and Baranowski, J. (2009). Steps in the design, development
and formative evaluation of obesity prevention-related behavior change trials. The
International Journal of Behavioral Nutrition and Physical Activity, 6, 6.
Baranowski, T., Cullen, K. W., Nicklas, T., Thompson, D., and Baranowski, J. (2003).
Are current health behavioral change models helpful in guiding prevention of
weight gain efforts? Obesity Research, 11(10S), 23S–43S.
Baranowski, T., Lin, L. S., Wetter, D. W., Resnicow, K., and Hearn, M. D. (1997).
Theory as mediating variables: Why aren’t community interventions working as
desired? Annals of Epidemiology, 7(7), S89–S95.
Bean, M. K., Wilson, D. B., Thornton, L. M., Kelly, N., and Mazzeo, S. E. (2012).
Dietary intake in a randomized-controlled pilot of NOURISH: a parent
intervention for overweight children. Preventive Medicine, 55(3), 224–7.
Beckman, H., Hawley, S., and Bishop, T. (2006). Application of Theory-Based Health
Behavior Change Techniques to the Prevention of Obesity in Children. Journal of
Pediatric Nursing, 21(4), 266–275.
Belasco, E., Chidmi, B., Lyford, C., and Funtanilla, M. (2012). Using Quantile
Regression to Measure the Differential Impact of Economic and Demographic
Variables on Obesity. Health Behavior and Public Health, 2(2), 35–45.
Bender, M., and Derby, B. (1992). Prevalence of reading nutrition and ingredient
information on food labels among adult Americans. Journal of Nutrition
Education, 24(6), 292–297.
Bianchini, F., Kaaks, R., and Vainio, H. (2002). Review Overweight , obesity , and
cancer risk. The Lancet Oncology, 3(9), 565–574.
Blankenau, J. (2009). Nutrition , Physical Activity , and Obesity in Rural America
Nutrition , Physical Activity , and Obesity in Rural America (pp. 1–6). Lyons,NE.
Texas Tech University, Janani R. Thapa, August 2014
180
Bleich, S. N., Segal, J., Wu, Y., Wilson, R., and Wang, Y. (2013). Systematic review of
community-based childhood obesity prevention studies. Pediatrics, 132(1), e201–
10.
Bluford, D. A. A., Sherry, B., and Scanlon, K. S. (2007). Interventions to prevent or treat
obesity in preschool children: a review of evaluated programs. Obesity, 15(6),
1356–1372.
Borenstein, M., Hedges, L. V., Higgins, J. P. T., and Rothstein, H. R. (2009).
Introduction to Meta-Analysis (1st ed., pp. 61–75). Hoboken, NJ,USA: Wiley.
Broussard, B. A., Sugarman, J. R., Bachman-Carter, K., Booth, K., Stephenson, L.,
Strauss, K., and Gohdes, D. (1995). Toward comprehensive obesity prevention
programs in Native American communities. Obesity Research, 3(Suppl 2), 289s–
297s.
Brown, P. (2009). Waist circumference in primary care. Primary Care Diabetes, 3(4),
259–261.
Caan, B., Ballard-Barbash, R., Slattery, M. L., Pinsky, J. L., Iber, F. L., Mateski, D. J., …
Lanza, E. (2004). Low energy reporting may increase in intervention participants
enrolled in dietary intervention trials. Journal of the American Dietetic
Association, 104(3), 357–366.
Caballero, B., Clay, T., Davis, S. M., Ethelbah, B., Rock, B. H., Lohman, T., … Stevens,
J. (2003). Pathways: a school-based, randomized controlled trial for the
prevention of obesity in American Indian schoolchildren. The American Journal
of Clinical Nutrition, 78, 1030–1038.
Calle, E. E., Rodriguez, C., Walker-Thurmond, K., and Thun, M. J. (2003). Overweight,
Obesity, and Mortality from Cancer in a Prospectively Studied Cohort of U.S.
Adults. The New England Journal of Medicine, 348(17), 1625–1638.
Calle, E. E., Thun, M. J., Petrelli, J. M., Rodriguez, C., and Heath, C. W. (1999). Body-
mass index and mortality in a prospective cohort of U.S. adults. The New England
Journal of Medicine, 341(15), 1097–1105.
Camerer, C., Issacharoff, S., Loewenstein, G., O’Donoghue, T., and Rabin, M. (2003).
Regulation for Conservatives: Behavioral Economics and the Case for
“Asymmetric Paternalism.” University of Pennsylvania Law Review, 151, 1211–
1254.
Carpenter, V., and Colwell, B. (1995). Cancer knowledge, self-efficacy, and cancer
screening behaviors among Mexican-American women. Journal of Cancer
Education, 10(4), 217–222.
Texas Tech University, Janani R. Thapa, August 2014
181
Cawley, J. (2004). An economic framework for understanding physical activity and
eating behaviors. American Journal of Preventive Medicine, 27(3 suppl), 117–
125.
Cawley, J. (2010). The Economics of Childhood Obesity. Health Affairs At the
Intersection of Health, Health Care and Policy, 29(3), 364–371.
CDC. (2012). Health, United States, 2012: With Special Feature on Emergency Care.
Cheadle, A., Samuels, S. E., Rauzon, S., Yoshida, S. C., Schwartz, P. M., Boyle, M., …
Solomon, L. (2010). Approaches to measuring the extent and impact of
environmental change in three California community-level obesity prevention
initiatives. American Journal of Public Health, 100(11), 2129–2136.
Chen, Y., and Huffman, W. (2009). An Economic Analysis of the Impact of Food Prices
and Other Factors on Physical Activity and Energy Balance of Adults. Ames,IA.
Chomitz, V. R., McGowan, R. J., Wendel, J. M., Williams, S. A., Cabral, H. J., King, S.
E., … Hacker, K. A. (2010). Healthy Living Cambridge Kids: a community-based
participatory effort to promote healthy weight and fitness. Obesity, 18(Suppl 1),
S45–S53.
Chou, C. P., Montgomery, S., Pentz, M. A., Rohrbach, L. A., Johnson, C. A., Flay, B. R.,
and MacKinnon, D. P. (1998). Effects of a community-based prevention program
on decreasing drug use in high-risk adolescents. American Journal of Public
Health, 88(6), 944–948.
Chou, S. Y., Grossman, M., and Saffer, H. (2004). An economic analysis of adult obesity:
results from the Behavioral Risk Factor Surveillance System. Journal of Health
Economics, 23(3), 565–587.
Chou, S.-Y., Grossman, M., and Saffer, H. (2004). An economic analysis of adult
obesity: results from the Behavioral Risk Factor Surveillance System. Journal of
Health Economics, 23(3), 565–587.
Ciao, A. C., and Latner, J. D. (2011). Reducing obesity stigma: the effectiveness of
cognitive dissonance and social consensus interventions. Obesity, 19(9), 1768–
1774.
Cohen, D. A., and Babey, S. H. (2012). Contextual influences on eating behaviours:
heuristic processing and dietary choices. Obesity Reviews, 13(9), 766–779.
Cook-Cottone, C., Casey, C. M., Feeley, T. H., and Baran, J. (2009). A meta-analytic
review of obesity prevention in the schools: 1997-2008. Psychology in the
Schools, 46(8), 695–719.
Texas Tech University, Janani R. Thapa, August 2014
182
Corda, K. W., Quick, V., Schefske, S., DeCandia, J., and Byrd-Bredbenner, C. (2010).
Toward a Polytheoretical Framework for Health Behavior Change. American
Journal of Health Studies, 25(4), 211–230.
Crawford, D., Jeffery, R. W., and French, S. A. (2000). Can anyone successfully control
their weight? Findings of a three year community-based study of men and
women. International Journal of Obesity and Related Metabolic Disorders, 24(9),
1107–1110.
Crespo, N., Elder, J., Ayala, G., Slymen, D., Campbell, N., Sallis, J., … Arredondo, E.
(2012). Results of a Multi-level Intervention to Prevent and Control Childhood
Obesity among Latino Children: The Aventuras Para Nios Study. Annals of
Behavioral Medicine, 43(1), 84–100.
Cummins, S., Petticrew, M., Higgins, C., Findlay, A., and Sparks, L. (2005). Large scale
food retailing as an intervention for diet and health: quasi-experimental evaluation
of a natural experiment. Journal of Epidemiology and Community Health, 59(12),
1035–1040.
De Silva-Sanigorski, A. M., Bolton, K., Haby, M., Kremer, P., Gibbs, L., Waters, E., and
Swinburn, B. (2010). Scaling up community-based obesity prevention in
Australia: background and evaluation design of the Health Promoting
Communities: Being Active Eating Well initiative. BMC Public Health, 10, 65.
Delahanty, L. M., Meigs, J. B., Hayden, D., Williamson, D. A., and Nathan, D. M.
(2002). Psychological and behavioral correlates of baseline BMI in the diabetes
prevention program (DPP). Diabetes Care, 25(11), 1992–1998.
DeMattia, L., and Lee Denney, S. (2008). Childhood Obesity Prevention: Successful
Community-Based Efforts. The Annals of the American Academy of Political and
Social Science, 615(1), 83–99.
Dionne, I., Johnson, M., White, M. D., St-Pierre, S., and Tremblay, A. (1997). Acute
effect of exercise and low-fat diet on energy balance in heavy men. International
Journal of Obesity and Related Metabolic Disorders, 21(5), 413–416.
Dooley, D. A., Novotny, R., and Britten, P. (1998). Integrating research into the
undergraduate nutrition curriculum: Improving shoppers’ awareness and
understanding of nutrition facts labels. Journal of Nutrition Education, 30(4),
225–231.
Doyle, E., and Ward, S. (2001). The Process of Community Health Education and
Promotion (pp. 198–219). Mayfield Publishing Company.
Drewnowski, A. (2003). Fat and sugar: an economic analysis. The Journal of Nutrition,
133(3), 838S–840S.
Texas Tech University, Janani R. Thapa, August 2014
183
Drewnowski, A., and Darmon, N. (2005). The economics of obesity: dietary energy
density and energy cost. The American Journal of Clinical Nutrition, 82(1 Suppl),
265S–273S.
Drewnowski, A., and Specter, S. E. (2004). Poverty and obesity: the role of energy
density and energy costs. The American Journal of Clinical Nutrition, 79(1), 6–
16.
Economos, C. D., Folta, S. C., Goldberg, J., Hudson, D., Collins, J., Baker, Z., … Nelson,
M. (2009). A community-based restaurant initiative to increase availability of
healthy menu options in Somerville, Massachusetts: Shape Up Somerville.
Preventing Chronic Disease, 6(3), A102.
Economos, C. D., Hyatt, R. R., Goldberg, J. P., Must, A., Naumova, E. N., Collins, J. J.,
and Nelson, M. E. (2007). A community intervention reduces BMI z-score in
children: Shape Up Somerville first year results. Obesity, 15(5), 1325–1336.
Economos, C. D., and Irish-Hauser, S. (2007). Community interventions: a brief
overview and their application to the obesity epidemic. The Journal of Law,
Medicine and Ethics, 35(1), 131–137.
Edwards, R. D. (2011). Commentary: Soda taxes, obesity, and the shifty behavior of
consumers. Preventive Medicine, 52(6), 417–418.
Eisenmann, J. C., Gentile, D. a, Welk, G. J., Callahan, R., Strickland, S., Walsh, M., and
Walsh, D. a. (2008). SWITCH: rationale, design, and implementation of a
community, school, and family-based intervention to modify behaviors related to
childhood obesity. BMC Public Health, 8, 223.
Epstein, L. H., Valoski, A., Wing, R., and McCurley, J. (1984). Ten year outcome of
behavioral family based treatment for childhood obesity. Health Psychology,
13(5), 373–383.
Ernst, N. D., Wu, M., Frommer, P., Katz, E., Matthews, O., Moskowitz, J., … Sondik, E.
(1986). Nutrition education at the point of purchase: the foods for health project
evaluated. Preventive Medicine, 15(1), 60–73.
Escaron, A. L., Meinen, A. M., Nitzke, S. A., and Martinez-Donate, A. P. (2013).
Supermarket and Grocery Store–Based Interventions to Promote Healthful Food
Choices and Eating Practices: A Systematic Review. Preventing Chronic Disease,
10(2), E50.
Eysenck, M. W., and Keane, M. T. (2005). Cogn t ve Psy hology: A Student’s Hand ook.
Cognitive Psychology (Vol. 5, p. 979). Psychology Press.
Festinger, L. (1962). Cognitive dissonance. Scientific American, 207, 93–102.
Texas Tech University, Janani R. Thapa, August 2014
184
Finkelstein, E. A., Ruhm, C. J., and Kosa, K. M. (2005). Economic causes and
consequences of obesity. Annual Review of Public Health, 26, 239–257.
Finkelstein, E. A., Trogdon, J., Cohen, J., and Dietz, W. (2009). Annual medical
spending attributable to obesity: payer- and service specific estimates. Health
Affairs, 5, w822–31.
Fiseković, S. (2005). Obesity biological and psychological aspect. Bosnian Journal of
Basic Medical Sciences Udruzenje Basicnih Mediciniskih Znanosti Association of
Basic Medical Sciences, 5(4), 30–34.
Flegal, K. M., Carroll, M. D., Kit, B. K., and Ogden, C. L. (2012). Prevalence of obesity
and trends in the distribution of body mass index among US adults, 1999-2010.
Jama The Journal Of The American Medical Association, 307(5), 491–7.
Fletcher, J., Frisvold, D., and Tefft, N. (2010). Can soft drink taxes reduce population
weight? Contemporary Economic Policy, 28(1), 23–35.
Fogelholm, M., and Lahti-Koski, M. (2002). Community health promotion interventions
with physical activity: does this approach prevent obesity? Food and Nutrition
Research, 46(4), 173–177.
Folta, S. C., Lichtenstein, A. H., Seguin, R. A., Goldberg, J. P., Kuder, J. F., and Nelson,
M. E. (2009). The StrongWomen-Healthy Hearts program: reducing
cardiovascular disease risk factors in rural sedentary, overweight, and obese
midlife and older women. American Journal of Public Health, 99(7), 1271–1277.
Foster, G. D., Sherman, S., Borradaile, K. E., Grundy, K. M., Vander Veur, S. S.,
Nachmani, J., … Shults, J. (2008). A policy-based school intervention to prevent
overweight and obesity. Pediatrics, 121(4), e794–e802.
Fotu, K. F., Millar, L., Mavoa, H., Kremer, P., Moodie, M., Snowdon, W., … Swinburn,
B. a. (2011). Outcome results for the Ma’alahi Youth Project, a Tongan
community-based obesity prevention programme for adolescents. Obesity
Reviews, 12(Suppl 2), 41–50.
Framson, C., Kristal, A. R., Schenk, J. M., Littman, A. J., Zeliadt, S., and Benitez, D.
(2009). Development and validation of the mindful eating questionnaire. Journal
of the American Dietetic Association, 109(8), 1439–44.
French, S. A., Story, M., and Jeffery, R. W. (2001). Environmental influences on eating
and physical activity. Annual Review of Public Health, 22(1), 309–335.
Gao, Y., Griffiths, S., and Chan, E. Y. Y. (2008). Community-based interventions to
reduce overweight and obesity in China: a systematic review of the Chinese and
English literature. Journal of Public Health (Oxford, England), 30(4), 436–448.
Texas Tech University, Janani R. Thapa, August 2014
185
Gao, Y., Ran, X.-W., Xie, X.-H., Lu, H.-L., Chen, T., Ren, Y., … Tian, H.-M. (2011).
Prevalence of overweight and obesity among Chinese Yi nationality: a cross-
sectional study. BMC Public Health, 11(1), 919.
Gentile, D. A., Welk, G., Eisenmann, J. C., Reimer, R. A., Walsh, D. A., Russell, D. W.,
… Fritz, K. (2009). Evaluation of a multiple ecological level child obesity
prevention program: Switch what you Do, View, and Chew. BMC Medicine, 7,
49.
Giang, T., Karpyn, A., Laurison, H. B., Hillier, A., and Perry, R. D. (2008). Closing the
grocery gap in underserved communities: the creation of the Pennsylvania Fresh
Food Financing Initiative. Journal of Public Health Management and Practice
JPHMP, 14(3), 272–279.
Gittelsohn, J., and Kumar, M. B. (2007). Preventing childhood obesity and diabetes: is it
time to move out of the school? Pediatric Diabetes, 8(Suppl 9), 55–69.
Gittelsohn, J., and Lee, K. (2013). Integrating Educational, Environmental, and
Behavioral Economic Strategies May Improve the Effectiveness of Obesity
Interventions. Applied Economic Perspective and Policy, 35(1), 52–68.
Glanz, K., and Mullis, R. M. (1988). Environmental interventions to promote healthy
eating: a review of models, programs, and evidence. Health Education Quarterly,
15(4), 395–415.
Glanz, K., Patterson, R. E., Kristal, A. R., Feng, Z., Linnan, L., Heimendinger, J., and
Hebert, J. R. (1998). Impact of work site health promotion on stages of dietary
change: the Working Well Trial. Health Education and Behavior, 25(4), 448–463.
Glasgow, R. E., Vogt, T. M., and Boles, S. M. (1999). Evaluating the public health
impact of health promotion interventions: the RE-AIM framework. American
Journal of Public Health, 89(9), 1322–1327.
Goldfinger, J. Z., Arniella, G., Wylie-Rosett, J., and Horowitz, C. R. (2008). Project
HEAL: peer education leads to weight loss in Harlem. Journal of Health Care for
the Poor and Underserved, 19(1), 180–192.
Gong, G., Belasco, E., Hargrave, K., Lyford, C., and Philips, B. (2012). Determinants of
Delayed Detection of Cancers in the Texas Counties in the United States of
America. International Journal of Equity in Health, 11(1), 29.
Gonzalez-Suarez, C., Worley, A., Grimmer-Somers, K., and Dones, V. (2009). School-
Based Interventions on Childhood Obesity. A Meta-Analysis. American Journal
of Preventive Medicine, 37(5), 418–427.
Texas Tech University, Janani R. Thapa, August 2014
186
Gortmaker, S. L., Peterson, K., Wiecha, J., Sobol, A. M., Dixit, S., Fox, M. K., and Laird,
N. (1999). Reducing obesity via a school-based interdisciplinary intervention
among youth: Planet Health. Archives of Pediatrics and Adolescent Medicine,
153(4), 409–418.
Gosschalk, A., and Carozza, S. (2010). Rural Healthy People 2010: A companion
document to Healthy People 2010. Science, I, 275.
Gotay, C. C. (2004). Behavior and Cancer Prevention. Journal of Clinical Oncology,
23(2), 301–310.
Gourlan, M. J., Trouilloud, D. O., and Sarrazin, P. G. (2011). Interventions promoting
physical activity among obese populations: a meta-analysis considering global
effect, long-term maintenance, physical activity indicators and dose
characteristics. Obesity Reviews, 12(7), e633–e645.
Graffagnino, C. L., Falko, J. M., La Londe, M., Schaumburg, J., Hyek, M. F., Shaffer, L.
E., … Caulin-Glaser, T. (2006). Effect of a community-based weight management
program on weight loss and cardiovascular disease risk factors. Obesity, 14(2),
280–288.
Greene, W. H. (2003). Econometric Analysis. (P. Education, Ed.)Journal of the American
Statistical Association (Vol. 97, p. 1026). Prentice Hall.
Guh, D. P., Zhang, W., Bansback, N., Amarsi, Z., Birmingham, C. L., and Anis, A. H.
(2009). The incidence of co-morbidities related to obesity and overweight: A
systematic review and meta-analysis. BMC Public Health, 9(1), 88.
Hansen, W. B., and McNeal, R. B. (1996). The law of maximum expected potential
effect:constraints placed on program effectiveness by mediator relationships.
Health Education Research, 11(4), 501–507.
Harnack, L., Stang, J., and Story, M. (1999). Soft drink consumption among US children
and adolescents: nutritional consequences. Journal of the American Dietetic
Association, 99(4), 436–441.
Heath, G. W., Wilson, R. H., Smith, J., and Leonard, B. E. (1991). Community-based
exercise and weight control: diabetes risk reduction and glycemic control in Zuni
Indians. The American Journal of Clinical Nutrition, 53(6 Suppl), 1642S–1646S.
Heath, R. (2006). Emotional persuasion. Advertising Research, (5), 47–48.
Hernandez, L. M., and Blazer Dan G. (2006). Genes, Behavior and the Social
Environment (pp. 68–108). Institute of Medicine of the National Academics, The
National Academy Press, Washington, D.C.
Texas Tech University, Janani R. Thapa, August 2014
187
Hesketh, K. D., and Campbell, K. J. (2010). Interventions to prevent obesity in 0-5 year
olds: an updated systematic review of the literature. Obesity (Silver Spring, Md.),
18(Suppl 1), S27–S35.
Hill, J. O. (2006). Understanding and addressing the epidemic of obesity: an energy
balance perspective. Endocrine Reviews, 27(7), 750–761.
Hoelscher, D. M., Springer, A. E., Ranjit, N., Perry, C. L., Evans, A. E., Stigler, M., and
Kelder, S. H. (2010). Reductions in child obesity among disadvantaged school
children with community involvement: the Travis County CATCH Trial. Obesity,
18(Suppl 1), S36–44.
Hollar, D., Lombardo, M., Lopez-Mitnik, G., Hollar, T. L., Almon, M., Agatston, A. S.,
and Messiah, S. E. (2010). Effective multi-level, multi-sector, school-based
obesity prevention programming improves weight, blood pressure, and academic
performance, especially among low-income, minority children. Journal of Health
Care for the Poor and Underserved, 21(2 Suppl), 93–108.
Hughes, A. R., Stewart, L., Chapple, J., McColl, J. H., Donaldson, M. D. C., Kelnar, C. J.
H., … Reilly, J. J. (2008). Randomized, controlled trial of a best-practice
individualized behavioral program for treatment of childhood overweight:
Scottish Childhood Overweight Treatment Trial (SCOTT). Pediatrics, 121(3),
e539–e546.
Janssen, I., Katzmarzyk, P. T., and Ross, R. (2004). Waist circumference and not body
mass index explains obesity-related health risk. The American Journal of Clinical
Nutrition, 79(3), 379–84.
Jeffery, R. W. (1993). Minnesota studies on community-based approaches to weight loss
and control. Annals of Internal Medicine, 119(7 Pt 2), 719–721.
Jeffery, R. W. (1995). Community programs for obesity prevention: the Minnesota Heart
Health Program. Obesity Research, 3(Suppl 2), 283s–288s.
Jeffery, R. W., and French, S. A. (1998). Epidemic obesity in the United States: are fast
foods and television viewing contributing? American Journal of Public Health,
88(2), 277–280.
Jenum, A. K., Anderssen, S. a, Birkeland, K. I., Holme, I., Graff-Iversen, S., Lorentzen,
C., … Bahr, R. (200 ). Promoting physical activity in a low-income multiethnic
district: effects of a community intervention study to reduce risk factors for type 2
diabetes and cardiovascular disease: a community intervention reducing
inactivity. Diabetes Care, 29(7), 1605–12.
Johnson, S. S., Paiva, A. L., Cummins, C. O., Johnson, J. L., Dyment, S. J., Wright, J. A.,
… Sherman, K. (2008). Transtheoretical Model-based multiple behavior
Texas Tech University, Janani R. Thapa, August 2014
188
intervention for weight management: Effectiveness on a population basis.
Preventive Medicine, 46(3), 238–246.
Johnson, W. G., Hinkle, L. K., Carr, R. E., Anderson, D. A., Lemmon, C. R., Engler, L.
B., and Bergeron, K. C. (1997). Dietary and exercise interventions for juvenile
obesity: long-term effect of behavioral and public health models. Obesity
Research, 5(3), 257–261.
Jp, F. (2011). Issues and misconceptions about obesity. Obesity, 19(4), 676–86.
Just, D. R. (2014). Introduction to Behavioral Economics: non-economic factors that
shape economics decisions (1st ed.). Wiley.
Just, D. R., and Payne, C. R. (2009). Obesity: can behavioral economics help? Annals of
Behavioral Medicine, 38(Suppl 1), S47–S55.
Kahneman, D. (2003). Maps of Bounded Rationality: Psychology for Behavioral
Economics. The American Economic Review, 93(5), 1449–1475.
Kain, J., Uauy, R., Albala, Vio, F., Cerda, R., and Leyton, B. (2004). School-based
obesity prevention in Chilean primary school children: methodology and
evaluation of a controlled study. International Journal of Obesity and Related
Metabolic Disorders, 28(4), 483–493.
Kamycheva, E., Joakimsen, R. M., and Jorde, R. (2003). Intakes of calcium and vitamin d
predict body mass index in the population of Northern Norway. The Journal of
Nutrition, 133(1), 102–106.
Kant, A. K., and Schatzkin, A. (1994). Consumption of energy-dense, nutrient-poor foods
by the US population: effect on nutrient profiles. Journal of the American College
of Nutrition, 13(3), 285–291.
Kim, K. H., Linnan, L., Campbell, M. K., Brooks, C., Koenig, H. G., and Wiesen, C.
(2008). The WORD (wholeness, oneness, righteousness, deliverance): a faith-
based weight-loss program utilizing a community-based participatory research
approach. Health Education and Behavior, 35(5), 634–50.
Kimokoti, R. W., Newby, P., Gona, P., Zhu, L., McKeon-O’Malley, C., Guzman, J. P.,
… Millen, B. E. (2012). Patterns of weight change and progression to overweight
and obesity differ in men and women: implications for research and interventions.
Public Health Nutrition, 16(8), 1463–1475.
Klein, W. M. P., Bloch, M., Hesse, B. W., McDonald, P. G., Nebeling, L., O’Connell, M.
E., … Tesauro, G. (201 ). Behavioral research in cancer prevention and control:
A look to the future. American Journal of Preventive Medicine, 46(3), 303–311.
Texas Tech University, Janani R. Thapa, August 2014
189
Klesges, R. C., Obarzanek, E., Klesges, L. M., Stockton, M. B., Beech, B. M., Murray, D.
M., … Sherrill-Mittleman, D. A. (2008). Memphis Girls health Enrichment Multi-
site Studies (GEMS). Phase 2: Design and baseline. Contemporary Clinical
Trials, 29(1), 42–55.
Kremers, S. P., De Bruijn, G.-J., Visscher, T. L., Van Mechelen, W., De Vries, N. K., and
Brug, J. (2006). Environmental influences on energy balance-related behaviors: A
dual-process view. The International Journal of Behavioral Nutrition and
Physical Activity, 3(1), 9.
Kristal, A. R., Curry, S. J., Shattuck, A. L., Feng, Z., and Li, S. (2000). A randomized
trial of a tailored, self-help dietary intervention: the Puget Sound Eating Patterns
study. Preventive Medicine, 31(4), 380–389.
Kristal, A. R., and Ollberding, N. J. (2012). Evaluation of Nutrition Interventions. In A.
M. Coulston, C. J. Boushey, and M. G. Ferruzzi (Eds.), Nutrition in the
Prevention and Treatment of Disease (3rd ed., pp. 191–208). Elsevier.
Kropski, J. A., Keckley, P. H., and Jensen, G. L. (2008). School-based obesity prevention
programs: an evidence-based review. Obesity, 16(5), 1009–1018.
Kumanyika, S. K. (2002). The minority factor in the obesity epidemic. Ethnicity and
Disease, 12(3), 316–319.
Kumanyika, S. K., Obarzanek, E., Stettler, N., Bell, R., Field, A. E., Fortmann, S. P., …
Hong, Y. (2008). Population-Based Prevention of Obesity. American Heart
Association Circulation, 118(4), 428 –464.
Lawrence, M. (2009). Invited commentary: Do food regulatory systems protect public
health? Public Health Nutrition, 12(11), 2247–2249.
Leung, M. W., Yen, I. H., and Minkler, M. (2004). Community-based participatory
research: a promising approach for increasing epidemiology’s relevance in the
21st century. International Journal of Epidemiology, 33(3), 499–506.
Light, L., Tenney, J., Portnoy, B., Kessler, L., Rodgers, A. B., Patterson, B., …
Tuckermanty, E. (1989). Eat for health: a nutrition and cancer control
supermarket intervention. Public Health Reports Washington DC 1974, 104(5),
443–450.
Liu, P.-H., Wang, J.-D., and Keating, N. L. (2013). Expected years of life lost for six
potentially preventable cancers in the United States. Preventive Medicine, 56(5),
309–13.
Texas Tech University, Janani R. Thapa, August 2014
190
Lockwood, C. M., DeFrancesco, C. A., Elliot, D. L., Beresford, S. A. A., and Toobert, D.
J. (2010). Mediation Analyses: Applications in Nutrition Research and Reading
the Literature. Journal of the American Dietetic Association, 110(5), 753–762.
Loewenstein, G. (2000). Emotions in economic theory and economic behavior. The
American Economic Review, 90(2), 426–432.
Ludwig, D. S., Peterson, K. E., and Gortmaker, S. L. (2001). Relation between
consumption of sugar-sweetened drinks and childhood obesity: a prospective,
observational analysis. The Lancet, 357(9255), 505–508.
Lynch, B. M. (2010). Sedentary behavior and cancer: a systematic review of the literature
and proposed biological mechanisms. Cancer Epidemiology, Biomarkers and
Prevention, 19(11), 2691–2709.
MacKinnon, D. P., and Dwyer, J. H. (1993). Estimating mediated effect in prevention
studies. Evaluation Review, 17(2), 144–148.
Maillot, M., Darmon, N., Darmon, M., Lafay, L., and Drewnowski, A. (2007). Nutrient-
dense food groups have high energy costs: an econometric approach to nutrient
profiling. The Journal of Nutrition, 137(7), 1815–1820.
Mankiw, G. (2010, June 4). Can a Soda Tax Save Us from Ourselves? The New York
Times. [Retrieved from http://relooney.fatcow.com/0_New_7338.pdf]
Martinez, J. A. (2000). Body-weight regulation: causes of obesity. The Proceedings of
the Nutrition Society, 59(3), 337–345.
Martínez, J. A., Moreno, M. J., Marques-Lopes, I., and Martí, A. (2002). Causes of
obesity. Anales Del Sistema Sanitario de Navarra, 25(Suppl 1), 17–27.
McCool, B., Lyford, C. P., Hensarling, N., Pence, B., McCool, A., Thapa, J., … Carte, T.
M. (2013). Reducing Cancer Risk in Rural Communities Through Supermarket
Interventions. Journal of Cancer Education, 28, 597–600.
Mccormick, B., Stone, I., and Team, C. A. (2007). Economic costs of obesity and the
case for. Obesity Reviews, 8(Suppl. 1), 161–164.
McGuire, M. T., Jeffery, R. W., French, S. A., and Hannan, P. J. (2001). The relationship
between restraint and weight and weight-related behaviors among individuals in a
community weight gain prevention trial. International Journal of Obesity and
Related Metabolic Disorders, 25(4), 574–580.
McNabb, W., Quinn, M., Kerver, J., Cook, S., and Karrison, T. (1997). The
PATHWAYS church-based weight loss program for urban African-American
women at risk for diabetes. Diabetes Care, 20(10), 1518–1523.
Texas Tech University, Janani R. Thapa, August 2014
191
Merzel, C., and D’Afflitti, J. (2003). Reconsidering Community-Based Health
Promotion: Promise, Performance, and Potential. American Journal of Public
Health, 93(4), 557–574.
Miljkovic, D., Nganje, W., and De Chastenet, H. (2008). Economic factors affecting the
increase in obesity in the United States: Differential response to price. Food
Policy, 33(1), 48–60.
Mobasheri, M., Van Teijlingen, E., and Smith, C. (2005). A systematic review of the
effectiveness of physical activity and other lifestyle interventions in prevention of
obesity in adults. The Cerebellum, 9999(9999).
Mohatt, G., Plartke, R., Klejka, J., Luick, B., Lardon, C., Bersamin, A., … CANHR
research team. (2007). The Center for Alaska Native Health Research Study: A
Community Based Participatory Research Study of Obesity and Chronic Disease-
related Protective and Risk Factors. International Journal of Circumpolar Health,
66(1).
Mokdad, A. H., Marks, J. S., Stroup, D. F., and Gerberding, J. L. (2004). Actual causes of
death in the United States, 2000. Jama The Journal Of The American Medical
Association, 291(10), 1238–1245.
Moodie, R., Swinburn, B., Richardson, J., and Somaini, B. (2006). Childhood obesity--a
sign of commercial success, but a market failure. International Journal of
Pediatric Obesity, 1(3), 133–138.
Moore, S. M., Borawski, E. a, Cuttler, L., Ievers-Landis, C. E., and Love, T. E. (2013).
IMPACT: a multi-level family and school intervention targeting obesity in urban
youth. Contemporary Clinical Trials, 36(2), 574–86.
Murgatroyd, P. R., Goldberg, G. R., Leahy, F. E., Gilsenan, M. B., and Prentice, A. M.
(1999). Effects of inactivity and diet composition on human energy balance.
International Journal of Obesity and Related Metabolic Disorders, 23(12), 1269–
1275.
Naresh, M. K. (2007). Marketing Research an Applied Orientation. Prentice Hall.
NCI. (2013). Obesity and Cancer Risk. National Cancer Institute Fact Sheet.
[Retrieved September 20, 2013, from
http://www.cancer.gov/cancertopics/factsheet/Risk/obesity]
Nebelling, L. (2012). A transdisciplinary approach to obesity and cancer: from cells to
society. BMC Proceedings, 6(Suppl 3).
Nelson, P. (1970). Information and Consumer Behavior. Journal of Political Economy,
78(2), 311.
Texas Tech University, Janani R. Thapa, August 2014
192
Neumark-Sztainer, D., Story, M., Hannan, P. J., and Rex, J. (2003). New Moves: A
school-based obesity prevention program for adolescent girls. Preventive
Medicine, 37(1), 41–51.
Nguyen, B., Shrewsbury, V. A., Connor, J. O., Steinbeck, K. S., Lee, A., and Hill, A. J.
(2012). Twelve-Month Outcomes of the Loozit Randomized Controlled Trial: A
Community-Based Healthy Lifestyle Program for Overweight and Obese
Adolescents. Archives of Pediatrics and Adolescent Medicine, 166(2), 170–177.
Nicklas, T. A., Bao, W., Webber, L. S., and Berenson, G. S. (1993). Breakfast
consumption affects adequacy of total daily intake in children. Journal of the
American Dietetic Association, 93(8), 886–891.
NIH-NHLBI. (2013). Balance Food and Activity. Healthy Weight Basics.
[Retrieved September 20, 2013, from
http://www.nhlbi.nih.gov/health/public/heart/obesity/wecan/healthy-weight-
basics/balance.htm]
O’Connor, J., Steinbeck, K., Hill, A., Booth, M., Kohn, M., Shah, S., and Baur, L.
(2008). Evaluation of a community-based weight management program for
overweight and obese adolescents: The Loozit study. Nutrition and Dietetics,
65(2), 121–127.
O’Loughlin, J. L., Paradis, G., Gray-Donald, K., and Renaud, L. (1999). The impact of a
community-based heart disease prevention program in a low-income, inner-city
neighborhood. American Journal of Public Health, 89(12), 1819–1826.
Ogden, C. L., Carroll, M. D., Kit, B. K., and Flegal, K. M. (2012). Prevalence of obesity
in the United States, 2009-2010. NCHS Data Brief, (82), 1–8.
Ogden, C. L., Carroll, M. D., Kit, B. K., and Flegal, K. M. (2014). Prevalence of
childhood and adult obesity in the United States, 2011-2012. JAMA : The Journal
of the American Medical Association, 311(8), 806–14.
Orbell, S., and Verplanken, B. (2010). The automatic component of habit in health
behavior: habit as cue-contingent automaticity. Health Psychology, 29(4), 374–
383.
Pagliassotti, M. J., Gayles, E. C., and Hill, J. O. (1997). Fat and energy balance. Annals
Of The New York Academy Of Sciences, 827, 431–448.
Patterson, B. H., Kessler, L. G., Wax, Y., Bernstein, A., Light, L., Midthune, D. N., …
Tuskermanty, E. (1992). Evaluation of a Supermarket Intervention: The NCI-
Giant Food Eat for Health Study. Evaluation Review, 16(5), 464.
Texas Tech University, Janani R. Thapa, August 2014
193
Peracchio, H. L., Henebery, K. E., Sharafi, M., Hayes, J. E., and Duffy, V. B. (2012).
Otitis media exposure associates with dietary preference and adiposity: a
community-based observational study of at-risk preschoolers. Physiology and
Behavior, 106(2), 264–71.
Peterson, K. E., and Fox, M. K. (2007). Addressing the epidemic of childhood obesity
through school-based interventions: what has been done and where do we go from
here? The Journal of Law Medicine and Ethics, 35(1), 113–130.
Philips, B., Gong, G., Hargrave, K., Belasco, E., and Lyford, C. (2011). Correlation of
the ratio of metastatic to nonmetastatic cancer cases with the degree of
socioeconomic deprivation among Texas counties. International Journal of
Health Geographics, 10, 10–12.
Po’e, E. K., Heerman, W. J., Mistry, R. S., and Barkin, S. L. (2013). Growing Right Onto
Wellness (GROW): a family-centered, community-based obesity prevention
randomized controlled trial for preschool child-parent pairs. Contemporary
Clinical Trials, 36(2), 436–49.
Pratt, C. A., Lemon, S. C., Fernandez, I. D., Goetzel, R., Beresford, S. A., French, S. A.,
… Webber, L. S. (2007). Design characteristics of worksite environmental
interventions for obesity prevention. Obesity, 15(9), 2171–2180.
Rejeski, W. J., Brubaker, P. H., Goff, D. C., Bearon, L. B., McClelland, J. W., Perri, M.
G., and Ambrosius, W. T. (2011). Translating weight loss and physical activity
programs into the community to preserve mobility in older, obese adults in poor
cardiovascular health. Archives of Internal Medicine, 171(10), 880–6.
Robertson, W., Friede, T., Blissett, J., Rudolf, M. C. J., Wallis, M., and Stewart-Brown,
S. (2008). Pilot of "Families for Health’': community-based family intervention
for obesity. Archives of Disease in Childhood, 93(11), 921–926.
Robinson, T. N., Matheson, D., Desai, M., Wilson, D. M., Weintraub, D. L., Haskell, W.
L., … Killen, J. D. (2013). Family, community and clinic collaboration to treat
overweight and obese children: Stanford GOALS-A randomized controlled trial
of a three-year, multi-component, multi-level, multi-setting intervention.
Contemporary Clinical Trials, 36(2), 421–35.
Rodgers, A. B., Kessler, L. G., Portnoy, B., Potosky, A. L., Patterson, B., Tenney, J., …
Mathews, O. (199 ). “Eat for Health”: a supermarket intervention for nutrition
and cancer risk reduction. American Journal of Public Health, 84(1), 72–76.
Rogers, V. W., Hart, P. H., Motyka, E., Rines, E. N., Vine, J., and Deatrick, D. a. (2013).
Impact of let’s go! 5-2-1-0: a community-based, multisetting childhood obesity
prevention program. Journal of Pediatric Psychology, 38(9), 1010–20.
Texas Tech University, Janani R. Thapa, August 2014
194
RWJF. (2011). Harnessing the Power of Supermarkets to Help Reverse Childhood
Obesity. The Food Trust and The Robert Wood Johnson Foundation.
Sacher, P. M., Kolotourou, M., Chadwick, P. M., Cole, T. J., Lawson, M. S., Lucas, A.,
and Singhal, A. (2010). Randomized controlled trial of the MEND program: a
family-based community intervention for childhood obesity. Obesity, 18(Suppl 1),
S62–8.
Sacks, G., Tikellis, K., Millar, L., and Swinburn, B. (2011). Impact of “traffic-light”
nutrition information on online food purchases in Australia. Australian and New
Zealand Journal of Public Health, 35(2), 122–126.
Safron, M., Cislak, A., Gaspar, T., and Luszczynska, A. (2011). Effects of school-based
interventions targeting obesity-related behaviors and body weight change: a
systematic umbrella review. Behavioral Medicine, 37(1), 15–25.
Samuels, S. E., Craypo, L., Boyle, M., Crawford, P. B., Yancey, A., and Flores, G.
(2010). The California Endowment’s Healthy Eating, Active Communities
program: a midpoint review. American Journal of Public Health, 100(11), 2114–
2123.
Sanigorski, A. M., Bell, A. C., Kremer, P. J., Cuttler, R., and Swinburn, B. A. (2008).
Reducing unhealthy weight gain in children through community capacity-
building: results of a quasi-experimental intervention program, Be Active Eat
Well. International Journal of Obesity, 32(7), 1060–1067.
Satterfield, D. W., Volansky, M., Caspersen, C. J., Engelgau, M. M., Bowman, B. a,
Gregg, E. W., … Vinicor, F. (2003). Community-based lifestyle interventions to
prevent type 2 diabetes. Diabetes Care, 26(9), 2643–52.
Schmitz, M. K., and Jeffery, R. W. (2000). Public health interventions for the prevention
and treatment of obesity. The Medical Clinics of North America, 84(2), 491–512.
Schwartz, R., Powell, L., and Keifer, M. (2013). Family-based risk reduction of obesity
and metabolic syndrome: an overview and outcomes of the Idaho Partnership for
Hispanic Health. Journal of Health Care for the Poor and Underserved, 24(2
Suppl), 129–44.
Scott, J. A., Begley, A. M., Miller, M. R., and Binns, C. W. (1991). Nutrition education
in supermarkets: the Lifestyle 2000 experience. Australian Journal of Public
Health, 15(1), 49–55.
Seale, J. L., and Rumpler, W. V. (1997). Comparison of energy expenditure
measurements by diet records, energy intake balance, doubly labeled water and
room calorimetry. European Journal of Clinical Nutrition, 51(12), 856.
Texas Tech University, Janani R. Thapa, August 2014
195
Segarra, E., Dicks, R. M., and Abernathy, R. J. (n.d.). Optimal Input Use in Agriculture:
The welfare Implications of Natural Resource Use Policies. Texas Tech
University College of Agricultural Sciences Publication, T 340.
Seidell, J. C. (2010). Waist circumference and waist/hip ratio in relation to all-cause
mortality, cancer and sleep apnea. European Journal of Clinical Nutrition, 64(1),
35–41.
Seo, D.-C., and Sa, J. (2008). A meta-analysis of psycho-behavioral obesity interventions
among US multiethnic and minority adults. Preventive Medicine, 47(6), 573–82.
Shaibi, G. Q., Konopken, Y., Hoppin, E., Keller, C. S., Ortega, R., and Castro, F. G.
(2012). Effects of a culturally grounded community-based diabetes prevention
program for obese Latino adolescents. The Diabetes Educator, 38(4), 504–12.
Shang, X. W., Liu, A. L., Zhang, Q., Hu, X. Q., Du, S. M., Ma, J., … Ma, G. S. (2012).
Report on childhood obesity in China: sugar-sweetened beverages consumption
and obesity. Biomedical and Environmental Sciences, 25(2), 125–32.
Showell, N. N., Fawole, O., Segal, J., Wilson, R. F., Cheskin, L. J., Bleich, S. N., …
Wang, Y. (2013). A systematic review of home-based childhood obesity
prevention studies. Pediatrics, 132(1), e193–200.
Shrewsbury, V., O’Connor, J., Steinbeck, K. S., Stevenson, K., Lee, A., Hill, A. J., …
Baur, L. (2009). A randomised controlled trial of a community-based healthy
lifestyle program for overweight and obese adolescents: the Loozit study protocol.
BMC Public Health, 9, 119.
Shumaker, S. A., Ockene, J. K., and Riekert, K. A. (2009). The Handbook of Health
Behavior Change. New York (Vol. 31, p. 827). Springer Publishing Company.
Sinha, K. (2010, November 10). Indian in Grip of Obesity Epidemic. The Times of India.
[Retrieved from http://articles.timesofindia.indiatimes.com/2010-11-
12/india/28245306_1_obesity-india-and-china-overweight-rates]
Smith, L. R., Chadwick, P., Radley, D., Kolotourou, M., Gammon, C. S., Rosborough, J.,
and Sacher, P. M. (2013). Assessing the short-term outcomes of a community-
based intervention for overweight and obese children: The MEND 5-7
programme. BMJ Open, 3(5), 1–8.
Song, H.-J., Gittelsohn, J., Kim, M., Suratkar, S., Sharma, S., and Anliker, J. (2009). A
corner store intervention in a low-income urban community is associated with
increased availability and sales of some healthy foods. Public Health Nutrition,
12(11), 2060–2067.
Texas Tech University, Janani R. Thapa, August 2014
196
Spence, J. C., Cutumisu, N., Edwards, J., Raine, K. D., and Smoyer-Tomic, K. (2009).
Relation between local food environments and obesity among adults. BMC Public
Health, 9, 192.
Stark, C. S., Gittelsohn, J., Zonderman, A. B., Evans, M. K., and Gary-Webb, T. L.
(2010). Association of Walkability With Obesity in Baltimore City, Maryland.
American Journal of Public Health, 101(1), S318–S324.
Stellefson, M., Zhongmiao, W., and Klein, W. (2006). Effects of cognitive dissonance on
intentions to change diet and physical activity among college students. American
Journal of Health Studies, 21, 219–227.
Stewart, H. (2011). Food Away from Home. (J. Lusk, J. Roosen, and J. F. Shogren, Eds.)
(pp. 647–666).
Sturm, R. (2002). The Effects Of Obesity, Smoking, And Drinking On Medical Problems
And Costs. Health Affairs, 21(2), 245–253.
Sturm, R., Powell, L. M., Chriqui, J. F., and Chaloupka, F. J. (2010). Soda taxes, soft
drink consumption, and children’s body mass index. Health Affairs (Project
Hope), 29(5), 1052–1058.
Svetkey, L. P., Ard, J. D., Stevens, V. J., Loria, C. M., Young, D. Y., Hollis, J. F., …
Vollmer, W. M. (2012). Predictors of Long-Term Weight Loss in Adults With
Modest Initial Weight Loss, by Sex and Race. Obesity, 20(9), 1820–1828.
Swinburn, B. (2009). Obesity prevention in children and adolescents. Child and
Adolescent Psychiatric Clinics of North America, 18(1), 209–223.
Swinburn, B. A., Caterson, I., Seidell, J. C., and James, W. P. T. (2004). Diet, nutrition
and the prevention of excess weight gain and obesity. Public Health Nutrition,
7(1A), 123–146.
Swinburn, B. A., Sacks, G., Hall, K. D., McPherson, K., Finegood, D. T., Moodie, M. L.,
and Gortmaker, S. L. (2011). The global obesity pandemic: shaped by global
drivers and local environments. The Lancet, 378(9793), 804–814.
Taylor, N. J., Sahota, P., Sargent, J., Barber, S., Loach, J., Louch, G., and Wright, J.
(2013). Using intervention mapping to develop a culturally appropriate
intervention to prevent childhood obesity: the HAPPY (Healthy and Active
Parenting Programme for Early Years) study. The International Journal of
Behavioral Nutrition and Physical Activity, 10, 142.
Taylor, R. W., McAuley, K. A., Barbezat, W., Strong, A., Williams, S. M., and Mann, J.
I. (2007). APPLE Project: 2-y findings of a community-based obesity prevention
Texas Tech University, Janani R. Thapa, August 2014
197
program in primary school age children. The American Journal of Clinical
Nutrition, 86(3), 735–742.
Taylor, W. C., Hepworth, J. T., Lees, E., Cassells, A., Gousse, Y., Sweeney, M. M., …
Tobin, J. N. (2004). Readiness to change physical activity and dietary practices
and willingness to consult healthcare providers. Health Research Policy and
Systems, 2(1), 2.
Ter Bogt, T. F. M., van Dorsselaer, S. A. F. M., Monshouwer, K., Verdurmen, J. E. E.,
Engels, R. C. M. E., and Vollebergh, W. A. M. (2006). Body mass index and
body weight perception as risk factors for internalizing and externalizing problem
behavior among adolescents. The Journal of Adolescent Health, 39(1), 27–34.
Thomson, C. A., and Ravia, J. (2011). A systematic review of behavioral interventions to
promote intake of fruit and vegetables. Journal of the American Dietetic
Association, 111(10), 1523–35.
Tin, S. T., Mhurchu, C. N., and Bullen, C. (2007). Supermarket sales data: feasibility and
applicability in population food and nutrition monitoring. Nutrition Reviews,
65(1), 20–30.
Van Der Sluis, M. E., Lien, N., Twisk, J. W. R., Steenhuis, I. H. M., Bere, E., Klepp, K.-
I., and Wind, M. (2010). Longitudinal associations of energy balance-related
behaviours and cross-sectional associations of clusters and body mass index in
Norwegian adolescents. Public Health Nutrition, 13(10A), 1716–1721.
Van Stralen, M. M., Yildirim, M., Te Velde, S. J., Brug, J., Van, M. W., Chinapaw, M. J.,
… Chinapaw. (2011). What works in school-based energy balance behaviour
interventions and what does not? A systematic review of mediating mechanisms.
International Journal of Obesity, 35(10), 1251–1265.
Wang, Y., and Beydoun, M. (2007). The obesity rpidemic in the United States-gender,
age, socioeconomic, racial/ethnic, and geographic characteristics: A systematic
review and meta-regression analysis. Epidemiologic Reviews, 29(1), 6–28.
Wang, Y., Tussing, L., Odoms-Young, A., Braunschweig, C., Flay, B., Hedeker, D., and
Hellison, D. (2006). Obesity prevention in low socioeconomic status urban
African-american adolescents: study design and preliminary findings of the
HEALTH-KIDS Study. European Journal of Clinical Nutrition, 60(1), 92–103.
Wannamethee, S. G., Papacosta, O., Whincup, P. H., Carson, C., Thomas, M. C., Lawlor,
D. a, … Sattar, N. (2010). Assessing prediction of diabetes in older adults using
different adiposity measures: a 7 year prospective study in 6,923 older men and
women. Diabetologia, 53(5), 890–8.
Texas Tech University, Janani R. Thapa, August 2014
198
Ward, R. W., and Davis, J. E. (1978). A pooled cross-section time series model of
coupon promotions. American Journal of Agricultural Economics, 60(3), 393–
401.
Watts, J. J., and Segal, L. (2009). Market failure, policy failure and other distortions in
chronic disease markets. BMC Health Services Research, 9, 102.
Weinrich, S. P., Weinrich, M. C., Boyd, M. D., and Atkinson, C. (1998). The impact of
prostate cancer knowledge on cancer screening. Oncology Nursing Forum, 25(3),
527–534.
Whetstone, L. M., Kolasa, K., and Collier, D. (2012). Participation in Community-
Originated Interventions Is Associated With Positive Changes in Weight Status
and Health Behaviors in Youth. Journal of Health Promotion, 27(1), 10–16.
Williamson, D. A., and Stewart, T. M. (2005). Behavior and lifestyle: approaches to
treatment of obesity. The Journal of the Louisiana State Medical Society Official
Organ of the Louisiana State Medical Society, 157(Spec N), S50–S55.
Wing, R. R., Goldstein, M. G., Acton, K. J., Birch, L. L., Jakicic, J. M., Sallis, J. F., …
Surwit, R. S. (2001). Behavioral Science Research in Diabetes Lifestyle changes
related to obesity, eating behavior, and physical activity. Diabetes Care, 24(1),
117–123.
Wofford, L. G. (2008). Systematic Review of Childhood Obesity Prevention. Journal of
Pediatric Nursing, 23(1), 5–19.
Wright, R. W., Brand, R. A., Dunn, W., and Spindler, K. P. (2007). How to write a
systematic review. Clinical Orthopaedics and Related Research, 455(455), 23–
29.
Yancey, A. K., McCarthy, W. J., Harrison, G. G., Wong, W. K., Siegel, J. M., and Leslie,
J. (2006). Challenges in improving fitness: results of a community-based,
randomized, controlled lifestyle change intervention. Journal o Women’s Health,
15(4), 412–429.
Yoon, P. W., Bastian, B., Anderson, R. N., Collins, J. L., and Jaffe, H. W. (2014).
Potentially Preventable Deaths from the Five Leading Causes of Death - United
States, 2008-2010. MMWR. Morbidity and Mortality Weekly Report, 63(17), 369–
374.
Zimmerman, G. L., Olsen, C. G., and Bosworth, M. F. (2000). A “stages of change”
approach to helping patients change behavior. American Family Physician, 61(5),
1409–1416.
Texas Tech University, Janani R. Thapa, August 2014
199
Zizza, C., Siega-Riz, A. M., and Popkin, B. M. (2001). Significant increase in young
adults’ snacking between 1977-1978 and 1994-1996 represents a cause for
concern! Preventive Medicine, 32(4), 303–310.