/

University of Southern California Dissertations and Theses

/

Influences of specific environmental domains on childhood obesity and related behaviors

(USC Thesis Other)

Influences of specific environmental domains on childhood obesity and related behaviors

PDF

Download

Open document

Flip pages

Contact Us

Copy asset link

Request this asset

Transcript (if available)

Content Influences of specific environmental domains on childhood obesity and related behaviors A Dissertation Presented to Faculty of the USC Graduate School In Partial Fulfillment for the Degree of Doctor of Philosophy in Health Behavior Research University of Southern California Keck School of Medicine Department of Preventive Medicine Institute for Health Promotion and Disease Prevention Research Lauren Cook Martinez May 2016 2 TABLE OF CONTENTS Acknowledgements…………………………………………………………………………………..…………………….4 List of Tables and Figures…………..……………………………………………………………..…………………….6 Abstract……………………………………………………………………………………………………..…………………….7 Chapter 1: Background………………………………………….……………………………..……………………...….9 Introduction…………………………………………………………………………………..……………………….9 The problem: Childhood obesity……………………………………………………………..……...……….10 Behavioral determinants of obesity…………………………………………………..………………...…..12 Environmental impact on obesity-‐related behaviors…………………………………….….….…..16 Social environment ……………………………………………………….…….………………………17 Structured learning environment………………………………………..………………….…….21 Physical environment ………………………………………………………………..….…………….25 Aims and Hypotheses……………………………………………………………………………………………..31 Summary of Contribution………………………………………………………..……………………………...32 Chapter 2: Similarities in consumption of fruit, vegetables and soda among middle school friends………………………………………………………………………………………………………………...……………33 Introduction…………………………………………………………………………………..……………………....33 Methods………………………………………………….……………………………………………..…………...….35 Results…………………………………………………..…………………………………………...………………….39 Discussion……………………………………………………………………………………..…………….….….….43 Chapter 3: Co-‐occurring change in dietary determinants and fruit and vegetable intake in Latino elementary school youth……………………..………...………………………..……………………………...51 Introduction…………………………………………………………………………………..………………………51 Methods………………………………………………….……………………………………………..…………..….53 Results…………………………………………………..…………………………………………...………………....57 Discussion……………………………………………………………………………………..…………….….….….58 Chapter 4: Effect of environmental greenness and open recreational space use on youth momentary perceived stress…………………………………….………………………..……………………………..67 Introduction…………………………………………………………………………………..………………………67 Methods………………………………………………….……………………………………………..………………69 Results…………………………………………………..…………………………………………...………………….73 Discussion……………………………………………………………………………………..…………….….….….74 3 Chapter 5: Discussion………………………………………………………………………………..……………………80 Summary of Findings……………………………………………………………………..……………………….80 Interaction of Environmental Levels…………………………..………...………………..…………….….82 Future Research…………………….………………………………………………………..……………………..84 Limitations………………………………………………………………………………………………….….….…..85 Recommendations……………………………………………………………………………………….….….….86 Conclusion.………………………………………………………………………………………………….….….….87 References ……………….………………………………………………………………………………………...…..…….. 88 4 Acknowledgements I would like to extend my thanks and sincere gratitude to the members of my dissertation committee. My mentor, Dr. Donna Spruijt-‐Metz, has provided me with invaluable guidance and support throughout my graduate training. I have learned so much from her about transdisciplinary research, psychology and psychosocial metrics, mobile health, and pediatric obesity. She has been an amazing advocate for me, has been patient and understanding during challenging times, and has helped me to grow as an individual. Also, her intelligent sense of humor delights me. I would also like to thank Dr. Chih-‐Ping Chou, who has taught me so much about modeling and mathematics, and who I admire as a statistician, independent thinker, and kind-‐hearted individual. Dr. Jaimie Davis has taught me so much about dietary research, and to always think critically and dive deeper. It has truly been a joy to work with her. Dr. Tom Valente has provided me with incredible instruction on social network theory and methodology. He has a wonderful spirit and positive outlook that have made him a pleasure to work with. Dr. Jennifer Unger has provided me with insightful, level-‐ headed guidance through all my years in graduate school, and Dr. Genevieve Dunton, has taught me so much about mobile health, environmental research, and career development. I have loved working with both of these intelligent, inspiring women. Other faculty have also provided me with indispensable training and guidance, and I am grateful to have had the opportunity to have known and worked with them. Dr. Michael Goran was my first supervisor, and under him I learned so much about pediatric obesity, transdisciplinary research and project management. He is an incredibly thoughtful researcher and has been very generous throughout the many years I have known him. I would also like to thank Dr. Nicole Gatto, from whom I have learned epidemiology and gardening. She has an incredible attention to detail and amazes me with her dedication to her work and society. Dr. Jimi Huh has been so patient and generous with her time in teaching me about longitudinal research. Dr. Lorraine Turcotte, my undergraduate mentor, provided me with excellent training in kinesiology and introduced me to the USC Department of Preventive Medicine that has been home to me for so long. I would also like to thank my labmates. Gillian O’Reilly Gentner has been my friend, sounding board, and companion in this graduate school journey. The quality of her work is incredible and she has inspired me to be a more diligent, better researcher. Cheng (Freddy) 5 Wen inspires me with his enthusiasm, and has always provided me with a laugh or smile when I most needed it. Dr. Lauren Gyllenhammer, Dr. Tanya Alderete, and Dr. Claudia Toledo-‐ Corral have helped me to learn so much about obesity and metabolic function, and have been amazing friends and supporters for many years. I would also like to thank a special group of classmates, Stephanie Dyal, Dr. Eleanor Tate Shonkoff, Dr. Kim Miller, and Dr. Myriam Forster. They, along with Gillian Gentner, have been a wonderful group of friends who have provided perspective and many laughs, and I am so grateful to have found them. Finally, I would like to thank the most important people in my life. They have supported me during difficult times, and celebrating during good ones. My parents, Fran and Nancy, have always encouraged me in all my endeavors, and have been my inspiration in pursuing a graduate degree in Preventive Medicine. My brother, Ryan, has been my anchor and helps me to stay grounded during difficult times. My husband, Juan Manuel, is my biggest advocate and most patient and loving supporter. He has provided me with so much valuable advice and helped me grow immensely as a professional and individual. I would never be the individual I am today without him. 6 List of Tables and Figures List of Figures Figure 1-‐1: Dissertation Conceptual Model: environmental influences on childhood obesity and related behaviors ………………………………………...……………………..…………………...….…10 Figure 2-‐1: Sample classroom of middle school students partitioned into unique communities using the Walktrap Community Detection Algorithm………………………..…...………46 Figure 2-‐2: Network plots of 17 Los Angeles middle school classrooms…………………………………………………………………………………………………..…………………..47 Figure 2-‐3: Plots of communities of middle school students were significant homophily on vegetable intake was observed………………………………………………..………….……….49 Figure 2-‐4: Sample plot of vegetable intake in a community without significant homophily………………………………………………………………….………………………………..……………….….50 Figure 3-‐1: Conceptual model for intervention effects on change in FV intake as a mediator between change in FV determinants and change in BMI z-‐scor.……………….……….… 63 Figure 3-‐2: Associations between change in FV determinants and change in FV intake among LA Sprouts study participants………………………...………………………………………….……….… 65 Figure 3-‐3: Baseline-‐adjusted associations between FV determinants and FV intake among LA Sprouts study participants at follow-‐up………...……………………………………….……….… 66 List of Tables Table 2-‐1: Baseline characteristics of 617 middle school students in Los Angeles……….……...46 Table 2-‐2: Network summary statistics for 17 Los Angeles middle school classrooms……..…48 Table 2-‐3: Correlations between ego intake of vegetables, fruit and soda with named friend intake of the same food item among 617 middle school students in Los Angeles..……..48 Table 2-‐4: Summary of interactions between social network effects and dietary intake in 17 Los Angeles middle school classrooms……..………………………………………...……………………..49 Table 3-‐1: Baseline characteristics of elementary school children who participated in a garden-‐based randomized control trial. (n=363) ……..………………………………...……………….….….64 Table 3-‐2: Fit statistics at each model-‐building step……..………………………………...………….….…..64 Table 4-‐1: Baseline characteristics of 120 3 rd -‐8 th grade children in Chino, CA………………...….77 Table 4-‐2: Momentary normalized difference vegetation index (NDVI) as a predictor of stress in 3 rd -‐8 th grade children in Chino, CA……..……………………………………...………………….….78 Table 4-‐3: Momentary open recreational space (ORS) use as a predictor of stress in 3 rd -‐8 th grade children in Chino, CA……..…………………………………………………...………………….….78 Table 4-‐4: Thirty-‐minute normalized difference vegetation index (NDVI) as a predictor of stress in 3 rd -‐8 th grade children in Chino, CA……..………………………………………….….79 Table 4-‐5: Thirty-‐minute open recreational space (ORS) use as a predictor of stress in 3 rd -‐8 th grade children in Chino, CA……..…………………………………………………...………………….….79 7 ABSTRACT The high rates of childhood obesity in the US make it imperative that factors contributing to the onset of obesity are better understood. The goal of this dissertation is to better understand how the environment contributes to obesity-‐related behaviors at the social, structured learning, and physical levels. The objectives for this dissertation are the following: 1) better understand similarities among adolescents and their friends in intake of fruit, vegetables (FV) and soda, 2) evaluate the impact of a school-‐based nutrition and gardening intervention on FV intake as a mediator of dietary determinants and BMI, and 3) determine the impact of green space on perceived stress among youth in their everyday living environments. Study 1 utilized cross-‐sectional data from 617 middle school students in 17 classrooms in the Get Moving study. Social networks were constructed by asking students to name their five best friends in their classroom, and dietary frequency of FV and soda were obtained. Exponential random graph models were used to measure homophily on dietary intake among friends. Study 2 utilized data from the LA Sprouts study, a randomized, controlled, school-‐ based gardening and nutrition intervention with 363 elementary school students. At baseline and 12-‐weeks, BMI was measured and participants completed questionnaires on dietary determinants and behavior. Path modeling was used to examine co-‐occurring changes in these measures. Study 3 utilized data on 63 3 rd -‐8 th grade students who participated in the Mobile Healthy PLACES study. Children completed Ecological Momentary Assessment (EMA) questions on stress, and GPS-‐derived variables Open Recreational Space (ORS) and Normalized Difference Vegetation Index (NDVI) measured green space. Analyses to examine the effect of green space on stress were completed using multilevel modeling and partial correlations. In Study 1, there was no significant association between friends and their intake of FV, or soda (p>0.9). In Study 2, change in controlled motivation to eat FV, willingness to try FV, and vegetable preferences was associated with change in vegetable intake (0.21 ± 0.6, p<0.001; 0.17 ± 0.07, p<0.01; 0.17 ± 0.06, p<0.01; respectively), and change in vegetable preferences was associated with change in fruit intake (0.12 ± 0.06, p=0.047). There were no associations between change in FV intake and BMI, and no significant differences in effect 8 sizes between the control and intervention groups. In Study 3, there was no significant association between participant reported stress and either NDVI or ORS use. Although findings from Study 1 are null, expecting homogeneity in the association between friendships and dietary intake may not be realistic. There is some evidence that similarities in dietary intake exist among friends in small groups, and new techniques are needed to better understand how often and when these associations occur. Findings from Study 2 indicate that motivation to eat FV, willingness to try FV, and vegetable preferences are relevant targets for interventions addressing FV intake. In this study, the structured learning environment did not alter the strength of association between simultaneous change in dietary determinants and intake, but can be one mechanism through which changes in dietary intake and determinants occur. Although findings from Study 3 are null, participants reported minimal stress, so more advanced, biometric data collection techniques are needed to better capture participant momentary stress and co-‐occurring environmental contexts. Taken together, these findings provide further insight into environmental effects on obesity-‐related behaviors, yet additional research is needed on all study topics to provide a more robust understanding of environmental influences. 9 CHAPTER 1: Background Introduction Childhood obesity has emerged as an epidemic in the United States (US) and around the world, (Ebbeling, Pawlak, & Ludwig, 2002; Ogden, 2012) and this condition has far-‐ ranging chronic disease consequences. (Ebbeling et al., 2002) The direct costs of childhood obesity are estimated at $14 billion, yet the costs escalate to alarming numbers when obese children become obese adults, and an estimated $147 billion annually goes towards treating obesity-‐related illness. (Cawley, 2010) Individual behavioral choices, such as in diet and exercise, are commonly implicated in obesity development, but many conditions outside an individual’s control, such as early life factors and the environment, interact to have a significant impact. (Ebbeling et al., 2002) The obesity-‐related environment is a complex system that encompasses many domains. Firstly, child behaviors are influenced by peers and parents, such as through modeling or norms. (McClain, Chappuis, Nguyen-‐Rodriguez, Yaroch, & Spruijt-‐Metz, 2009; Salvy, de la Haye, Bowker, & Hermans, 2012) Next, the physical environment plays an important role, as children are limited to health-‐promoting options available to them at home, school and in their neighborhood. (French, Story, & Jeffery, 2001; Papas et al., 2007) Finally, while schools can present barriers to health through lack of sufficient access to physical activity and health dietary choices, schools can also play a protective role, as learning opportunities in these settings can empower children to make healthier choices. (T. Brown & Summerbell, 2009) This proposal will examine the impact of these three distinct environmental domains (social, physical and structured learning) on individual-‐level factors related to obesity (Figure 1-‐1). 10 Figure 1-‐1: Dissertation Conceptual Model: environmental influences on childhood obesity and related behaviors Green arrows indicate pathways that will be examined in this proposal. The problem: Childhood obesity Prevalence Childhood obesity is a major public health concern, as 32% of US children aged 2-‐19 are overweight (body mass index, BMI ≥ 85 th percentile for age and sex), and 17% are obese (BMI ≥ 95 th percentile for age and sex). (Ogden, 2012) Rates of childhood overweight and obesity are even higher among minority populations, and 39% of Blacks and Latinos are overweight, relative to only 28% of non-‐Hispanic Whites who are overweight. (Ogden, 2012) Some populations are at an even greater risk, including predominantly Latino elementary school children living in a large urban environment, in which over 50% of students are either overweight or obese. (J. N. Davis, Ventura, Cook, Gyllenhammer, & Gatto, 2011) Health consequences of childhood obesity The problem of childhood obesity is an important one, as overweight children are more likely to be overweight adults. (D. S. Freedman et al., 2003; Power, Lake, & Cole, 1997) Obesity is associated with many adult chronic health conditions, including cardiovascular 11 disease (CVD) and type 2 diabetes (T2D), yet more frequently, overweight children are showing clinical signs of adult chronic conditions. (Dietz, 1998) According to an international group of physicians, obesity is the leading factor contributing to childhood high blood pressure, due in part to increased blood volume and required cardiac output, and/or chronic hypoventilation found in these children. (Speiser et al., 2005) Furthermore, childhood obesity is also a strong predictor of adult hypertension. (Daniels, 2009) Other mechanisms through which obesity can put children at higher risk for CVD are higher levels of triglycerides and lower levels of high density lipoprotein (HDL), which may partly be explained by poor dietary patterns. (Daniels, 2009) As mentioned previously, T2D is another condition closely associated with obesity, (Kahn, Hull, & Utzschneider, 2006) and youth in the US are being diagnosed with T2D with increasing frequency. Prevalence of T2D has been estimated at 2.8 cases per 1000 among adolescents aged 10-‐19 years. (SEARCH for Diabetes in Youth Study Group, 2006) Central adiposity is the specific type of adiposity most related to T2D, and in youth it is associated with fasting insulin and impaired insulin action. (D. S. Freedman et al., 1987; Gutin et al., 1994) This relationship is believed to be driven by release of fatty acids, cytokines, glycerol and hormones from adipose tissue, which impact insulin uptake by body cells. (Kahn et al., 2006) Insulin resistance is also strongly implicated in the development of non-‐alcoholic fatty liver disease (NAFLD, defined as liver fat fraction > 5.5%), which can eventually progress to cirrhosis in severe cases. (Schwimmer et al., 2003) Prevalence of NAFLD is estimated at 10% in children, with the oldest children having the highest rates of this condition (17% prevalence in children aged 15-‐19 years). (Schwimmer et al., 2006) Several other non-‐ metabolic conditions are additional common consequences of childhood obesity, including sleep apnea, asthma, polycystic ovary syndrome, musculoskeletal problems and psychological issues such as anxiety, depression and poor self-‐esteem. (Bray & Bouchard, 2003) Similar to higher prevalence rates of obesity in minority groups, minority children are also at higher risk for related metabolic conditions. (Cowie et al., 2009) Among overweight Latino children, 60% had either intermittent or persistent pre-‐diabetes, and many had excess levels of visceral adiposity. (Goran, Lane, Toledo-‐Corral, & Weigensberg, 2008) Furthermore, 38% of these children also had clinical characteristics of NAFLD. (Goran et al., 2010) These children are critical targets for interventions and research on obesity-‐related behaviors and determinants, topics that are covered in this proposal. 12 Behavioral determinants of obesity Genetics play an at least moderate role in obesity development, (C. G. Bell, Walley, & Froguel, 2005) but behavioral factors are modifiable over the course of a lifetime, and will be the focus of this proposal. There are many behavioral determinants of obesity, with diet and physical activity (PA) being the most prominent given their strong relation to energy balance. (Rennie, Johnson, & Jebb, 2005; Spruijt-‐Metz, 2011) Stress also plays a very important role, both as a determinant of behavior and independent contributor to obesity. (Bray & Bouchard, 2003) All major behavioral determinants of obesity are discussed briefly below, but diet and stress are the individual-‐level factors that will be examined in this proposal and are thus discussed with more detail. Dietary intake Behaviors related to diet are unique in that food is consumed several times throughout the day, so compounded intake of deleterious foods is the primary concern relevant to chronic conditions. (Bray & Bouchard, 2003) Larger portion sizes, greater consumption of energy-‐ dense low-‐nutrient foods (for example, sugar-‐sweetened beverages, SSB), and lower consumption of nutrient-‐dense foods (such as fruits and vegetables, FV), are the primary dietary behaviors leading to childhood obesity. (Rennie et al., 2005). Because of this, the overall carbohydrate quality (i.e., proportion of foods with added sugars/refined grains, versus whole grains/FV) or, similarly, glycemic index (measured by blood glucose levels following consumption) may be relevant targets for interventions and recommendations. (Ludwig, 2002) In fact, interventions specifically targeting carbohydrate quality have been successful with Latino youth. (J. N. Davis et al., 2011; 2007) SSB and FV intake are specific behaviors that will be examined in this proposal, and are thus discussed in further detail below. Sugar-‐sweetened beverages Increased consumption of foods with added sugars, especially SSB, is a major contributor to the obesity epidemic and subsequent risk for T2D. (Bray & Bouchard, 2003) It is estimated that 16% of our caloric intake comes from added sugars, (Bray, Nielsen, & Popkin, 2004) and in less than 25 years, caloric intake from SSB rose 135%. (Nielsen & Popkin, 2004) A meta-‐analysis reveals that SSB intake is significantly associated with weight 13 gain, (Malik, Schulze, & Hu, 2006) and data from a large cohort study indicates that those who drink SSB daily have 80% greater risk for T2D, relative to those who do not drink SSB. (Schulze et al., 2004) High-‐fructose corn syrup (HFSC), a component in many SSB, is an especially deleterious ingredient because it has a higher concentration of fructose than table sugar, and fructose has different metabolic actions compared to glucose, the other component of table sugar. (Bray et al., 2004; Goran et al., 2013) Unlike glucose, fructose does not promote the same satiety signals in the brain as solid foods do, (Bray et al., 2004; Hu & Malik, 2010) so these beverages may not be recognized by the satiety system as highly caloric. Without dietary compensation made to account for the calories in SSB, greater total energy may be consumed throughout the day. (Hu & Malik, 2010) Furthermore, fructose promotes de novo lipogenesis, which can encourage weight gain and insulin resistance. (Basciano, Federico, & Adeli, 2005) Insulin resistance is also promoted by the rapid postprandial insulin response following consumption of these drinks high in sugar. (Hu & Malik, 2010) Early life consumption of SSB can have far-‐reaching consequences into adulthood, including effects on neuroendocrine and metabolic function, appetite regulation and fat deposition. (Goran et al., 2013) Fruit and vegetable intake Low FV intake is another national dietary concern, as only 1% of US adolescents meet recommendations for intake, (Kimmons, Gillespie, Seymour, Serdula, & Blanck, 2009) which is 3.5 – 5 cups/day for youth based on age and sex. (United States Department of Agriculture, 2013) Although the link between FV intake and obesity is inconclusive, (He et al., 2004; Tohill, Seymour, Serdula, Kettel-‐Khan, & Rolls, 2004) there have been several large studies linking increased FV intake to decreased risk for T2D and other chronic diseases. (Ford & Mokdad, 2001; Ford, Ajani, McGuire, & Liu, 2005; S. Liu et al., 2004; Montonen et al., 2005; Pan & Pratt, 2008) The lack of evidence supporting a link between FV intake and obesity may be due to frequent use of imprecise measurements of obesity (i.e., BMI) and dietary intake (i.e., screeners). (Tohill et al., 2004) The primary mechanism through which FV are purported to be beneficial is via fiber, (Lattimer & Haub, 2010; Weickert & Pfeiffer, 2008) and research indicates that one additional serving of fiber per day may reduce central adiposity and improve metabolic health in 14 overweight youth. (E. E. Ventura et al., 2008) Fiber slows gastric emptying and therefore impedes macronutrient absorption, (Ray et al., 1983) and also increases satiety, which limits intake of additional calories. (Bolton, Heaton, & Burroughs, 1981) Furthermore, polyphenols, which are present in FV, may alter metabolic pathways and limit inflammation (lower inflammation can also be protective against obesity). (Esmaillzadeh et al., 2006; Qureshi, Singer, & Moore, 2009; Williamson, 2012) Given the potentially important role of polyphenols, intake of nutrient rich vegetables (dark green and deep orange/yellow colored) appears to be of particular importance. (van Duyn & Pivonka, 2000) However, intake of these foods is particularly low, as the median adolescent intake is 0.06 cups/day, (Kimmons et al., 2009) and almost 40% of overweight Latino adolescents do not consume any of these beneficial vegetables. (Cook, O'Reilly, Goran, Weigensberg, Spruijt-‐Metz, & Davis, 2014b) One final note on FV intake: FV are often considered together as one food group, but determinants for fruits versus vegetables may differ, (Cook, O'Reilly, DeRosa, Rohrbach, & Spruijt-‐Metz, 2014a; Glasson, Chapman, & James, 2010) and vegetables have been shown to be more beneficial metabolically, compared to fruit. (Cooper et al., 2012) Therefore, in this proposal, F and V intake will be examined separately. Physical activity Lack of PA is another important factor, and while 42% of US children aged 6-‐11 meet recommendations for 60 min/day of PA, only 8% of adolescents aged 12-‐19 meet these recommendations. (Troiano et al., 2008) Furthermore, levels of PA are staggeringly low in some at-‐risk populations, and longitudinal data indicate that activity rates decline at an alarming rate in minority adolescent females. (Spruijt-‐Metz et al., 2013) Higher levels of PA are associated with lower levels of obesity, CVD and T2D, but greater sedentary behavior (waking time spent sitting, reclining, or lying down) is also independently associated with the aforementioned conditions. (Owen, Healy, Matthews, & Dunstan, 2010; Sisson et al., 2009; Spruijt-‐Metz, 2011) Data from the National Health and Nutrition Examination Survey (NHANES) indicate that children 6-‐11 years old spend at least 40% of their time in sedentary activities, and children 12-‐19 spend over 50% of their time sedentary. (C. E. Matthews et al., 2008) One related activity may be the large amounts of television watched by US children while sedentary, which can affect their dietary habits via distracted eating or increased exposure to unhealthy food advertising. (Rennie et al., 2005) 15 Sleep and smoking Other behaviors implicated in the development of obesity include sleep and smoking. Inadequate sleep has been implicated in risk for obesity, T2D and CVD. (Chen, Beydoun, & Wang, 2008; Gangwisch et al., 2006; Lumeng et al., 2007; Vgontzas et al., 2009) Short sleep duration may lead to hormonal changes, which influence metabolism, or lead to greater consumption of unhealthy foods. (Chen et al., 2008) Adolescent sleep need is approximately 9 hours/night, yet youth aged 11-‐18 have less sleep at all ages (with a linear decrease over time; at age 11, mean hours are 8.4; at age 17, mean hours are 6.9). (Van Cauter & Knutson, 2008) Smoking also contributes to risk for obesity, and in 2012, 7% of middle school students and 23% of high school students in the US were current smokers. (Centers for Disease Control and Prevention (CDC), 2013b) Generally, smokers tend to have lower BMI, (Bamia, Trichopoulou, Lenas, & Trichopoulos, 2004) but they also tend to have higher visceral adiposity (likely driven by elevated stress hormone activity), (Chiolero, Faeh, Paccaud, & Cornuz, 2008; Clair et al., 2011; Cryer, Haymond, Santiago, & Shah, 1976) which, as previously mentioned, is more metabolically deleterious. (D. S. Freedman et al., 1987) Smokers also have a higher risk of developing T2D, which may be driven partly by elevated visceral adiposity. (Willi, Bodenmann, Ghali, Faris, & Cornuz, 2007) Stress One more individual-‐level factor that is implicated in obesity development is stress. Stress, although not a behavior, is a modifiable risk factor, and perceived stress appears to be more important than objectively measured stress with respect to disease risk. (Spruijt-‐Metz, O'Reilly, Cook, Page, & Quinn, 2014) Interventions have been shown to be effective in reducing stress, and stress-‐reduction practices such as mindfulness may be a relevant behavioral focus for obesity prevention and treatment. (Grossman, Niemann, Schmidt, & Walach, 2004; O'Reilly, Cook, Spruijt-‐Metz, & Black, 2014) Stress acts by influencing obesity-‐related behaviors, and is also detrimental in and of itself. Stress can contribute to increased consumption of foods high in fat and sugar, by increasing the salience of these positive stimuli. (Dallman et al., 2003; Scott, Melhorn, & Sakai, 2012) Stress may also impact PA levels, and a recent review found that individuals who are regularly active are more likely to exercise as a response to stress, but more sedentary 16 individuals are less likely to exercise in stressful circumstances. (Stults-‐Kolehmainen & Sinha, 2013) Chronic stress can also lead to dysfunction of the hypothalamic-‐pituitary-‐adrenal (HPA) axis, which is responsible for the release of the stress hormone cortisol. (McEwen, 2006; Rosmond, 2003) Elevated cortisol levels can lead to increased insulin resistance (via decreased insulin secretion, and increased hepatic glucose production), and can also lead to increased visceral adiposity. (Björntorp, Holm, & Rosmond, 1999; Rosmond, 2003) Adolescence can be a particularly stressful time for individuals, and one quarter of Los Angeles high school students report being ‘very stressed’ on a daily basis. (Anda et al., 2000) Environmental stressors likely play a large role in determining an individual’s stress level (and are discussed below), but other individual-‐level factors may also be important. As youth transition towards adulthood they experience many challenging cognitive and biological changes, (Stang & Story, 2005) and may find this transition daunting as they struggle to form adult identities. (Arnett, 2006) Coping strategies, such as mindfulness-‐based practices mentioned above, can also serve to mitigate the impact of external stressors. (Anda et al., 2000) Environmental impact on obesity-‐related behaviors The contextual backdrop influencing obesity-‐related behaviors is complex, and will be discussed in detail below. One overarching factor is socio-‐economic status (SES). As previously mentioned, there are strong ethnic disparities in obesity and related disease risk in children and adults, and socio-‐economic factors are often implicated as the major cause this. (McLaren, 2007; Y. Wang, 2001) In a study of Los Angeles children, obesity prevalence was significantly associated with economic hardship of their communities, and in communities with the highest economic hardship, obesity rates were more than double that of communities with the least hardship. (Shih, Dumke, Goran, & Simon, 2013) However, there is not a strong direct theoretical relationship between obesity and either financial capacity or education (strong SES benchmarks), but rather these qualities of an individual or family affect access to health promoting resources, and influence behavioral choices. This proposal will examine the impact of environmental influences on obesity and related behaviors in three domains: the social environment, the structured learning environment, and the physical environment. 17 Study 1: Impact of the social environment on obesity-‐related behaviors There are two primary social environments that can impact obesity-‐related behaviors in children. The first is the family social environment, which is chiefly composed of interactions with parents or guardians (or other adult role models like teachers). The second is the peer environment, which is discussed in detail below. The larger community social environment also play a role in determining obesity-‐related behaviors, however the extent to which effects of this environment on obesity are mediated by the family or peer social environments is unclear. Therefore, the focus here will be on family and peer environments. Family social environment In the family social environment, parental modeling plays an important role in affecting children’s dietary intake. Parental intake is consistently associated with child intake, including with fruit and vegetables. (Rasmussen et al., 2006; van der Horst et al., 2007) Interestingly, food physical availability (i.e., what is served in meals, in this case) is not completely driving this association, but child reported perceived parental modeling is more strongly associated with child dietary intake, while parental report of modeling is not. (McClain et al., 2009) The family social environment can also impact PA; parental support and direct help from parents were associated with adolescent PA in a review of 108 studies. (Sallis, Prochaska, & Taylor, 2000) Also, parental stress can impact child obesity-‐related behaviors, not only due to altered parenting practices, but also by increasing child stress, (Tate, Wood, Liao, & Dunton, 2014) which as mentioned above, has a negative impact on diet and PA. Peer social environment The peer social environment also plays a role in youth stress, PA, and diet. Peers can provoke adolescent stress via explicit or implicit pressure to fit in, judging, and other behaviors constituting peer pressure. (Moksnes, Byrne, Mazanov, & Espnes, 2010a; Moksnes, Moljord, Espnes, & Byrne, 2010b) Peers also play a role in PA participation, most likely through social facilitation, in which access to friends increases opportunities for PA. (Salvy et al., 2012) Yet, similar to the negative effects of peer pressure, peer victimization (i.e., bullying) is negatively associated with PA levels. (Salvy et al., 2012; Storch et al., 2006) 18 Peer social environment and diet Peers can also be influential when it comes to dietary behaviors in youth. (Badaly, 2013; Contento, Williams, Michela, & Franklin, 2006; Feunekes, de Graaf, Meyboom, & van Staveren, 1998; Fitzgerald, Heary, Kelly, Nixon, & Shevlin, 2013) In as study of Australian youth, male friends were likely to share similar levels of high calorie food consumption. (de la Haye, Robins, Mohr, & Wilson, 2010; de la Haye, Robins, Mohr, & Wilson, 2013) A similar association was also seen in adolescents from the Netherlands, such that peers tend to share similar snack and soft drink intake. (Wouters, Larsen, Kremers, Dagnelie, & Geenen, 2010) Data from the Add Health study indicate that peers tend to share participation in eating at fast food restaurants, such that a 1-‐day increase in the number of days that friends (also known as ‘alters’) eat fast food corresponds to a 0.18 increase in the number of days a child (or, ‘ego’) will eat at these restaurants. (Ali, Amialchuk, & Heiland, 2011) Peer effects on eating FV and calorie-‐dense snacks were also examined in this study and found to be non-‐significant, but one limitation of this study is that FV and snack intake were measured as binary variables. (Ali et al., 2011) Data from the Project EAT study were also used to examine peer influence on adolescent food intake, and similar eating behaviors were found between friends and egos in breakfast eating, fast food restaurant visits, and intake of whole grains, dairy, SSB and vegetables (but not fruit). (Bruening et al., 2012; 2014) Furthermore, a simulation study demonstrated that high-‐school students form peer groups with students who share the same lunch-‐eating behaviors, (Dabbaghian, Mago, Wu, Fritz, & Alimadad, 2012) and dieting and disordered eating also appear to be shared behaviors among girls. (Eisenberg & Neumark-‐ Sztainer, 2010; Fletcher, Bonell, & Sorhaindo, 2011; Hutchinson & Rapee, 2007) A comprehensive review by Salvy et al identifies several social mechanisms through which peers may affect eating behaviors, which are described below. (Salvy et al., 2012) The first mechanism is via social facilitation, such individuals alter their eating behavior when in a group setting, versus eating alone. (Salvy et al., 2012) This can be due to increased duration of meals, (de Castro, 1994; Pliner, Bell, Hirsch, & Kinchla, 2006) and/or due to release of eating inhibitions, such that individuals eat more when in the presence of familiar peers compared to unfamiliar ones. (de Castro, 1994; Salvy et al., 2012) A second mechanism through which peers have an effect on eating is modeling, in that egos will strive to mimic behavior of an alter. (Bandura, 1977; Salvy et al., 2012) Ego consumption patterns consistently mirror those of alters in adults and children in controlled experiments, (Salvy et al., 2012) and overweight 19 youth may be more prone to mimicry of peers’ behaviors than normal-‐weight youth. (Bevelander, Anschütz, & Engels, 2012) Social Cognitive Theory, in which individuals learn from observing their social environment, has been well studied in a health behavior context, (Armitage & Conner, 2000) and modeling is a well-‐established predictor of dietary behavior. (McClain et al., 2009) The third mechanism is through impression management, whereby individuals seek to project a specific image of themselves, demonstrated by their eating behavior. (Leary & Kowalski, 1990; Salvy et al., 2012) This often manifests in healthier eating, (Vartanian, Herman, & Polivy, 2007) but may also result in increased consumption of American junk foods for children eager to demonstrate their assimilation to the US. (Unger et al., 2004) Finally, norms may play a role, such that individuals strive to eat more acceptable foods or limit inappropriate intake of food, so that they can gain or maintain position in a peer group. (Herman, Roth, & Polivy, 2003; Salvy et al., 2012) Study 1 description Although there is a strong theoretical foundation for the influence of peers on dietary behavior, there have been few studies that explicitly examine the association between individual and ego eating behaviors (key papers on this topic included above). Furthermore, very few studies examining peer effects on dietary behavior have utilized techniques developed specifically for social network analysis (SNA). SNA methods allow for the examination of the complex interrelationships between members of a specific community, and traditional regression approaches are not sufficient to capture this given the high inter-‐ correlations with networks. Furthermore, use of these specific methodologies is important because it allows for examination of the following: 1) relationships with an individual’s personal connections, 2) an individual’s role within the greater community, and 3) overall structure and properties of the community. (Valente, 2010) Study 1 will use SNA methodology to examine the similarity between friends in dietary intake within 17 classroom friendship networks. The specific dietary behaviors of interest are consumption of FV and soda, due to the metabolic importance of these foods, as mentioned above. Data for this study are from the Get Moving! project, a classroom-‐based media intervention aimed at improving PA in middle school students. (Nguyen-‐Michel, Unger, & Spruijt-‐Metz, 2007; Spruijt-‐Metz, Nguyen-‐Michel, Goran, Chou, & Huang, 2008) In this sample of 617 students, friendships are more likely to exist between students of the same weight 20 status, rather than between students in different weight categories. (Valente, Fujimoto, Chou, & Spruijt-‐Metz, 2009) Also, overweight students named more friends than their normal weight peers, although they also tended to be less likely to be named as a friend. (Valente et al., 2009) There are several strengths of this study. The first is in the inclusion of specific dietary behaviors, namely intake of FV and soda, which have been documented as high-‐importance foods for obesity and related disease development. To our knowledge, only two studies have specifically examined friends and either FV or soda intake, (Ali et al., 2011; Bruening et al., 2012; 2014) and neither of this projects utilized methodologies to account for network structure. Second, 17 adolescent classroom networks are included in these analyses, allowing for greater generalizability of findings. The only study thus far to use SNA methodologies to examine the effect of peers on dietary intake had access to only two classroom networks. (de la Haye et al., 2010; 2013) Finally, given the aforementioned gaps in this area of research, the third strength is methodological, and is in the use of exponential random graph models (ERGMs) for the analyses. As indicated earlier, traditional regression techniques are not appropriate for network data because the assumption of independence of observations is violated, which even multilevel models cannot account for. (Lusher, Koskinen, & Robins, 2013) A brief introduction to ERGM methodology is presented here: In regression analyses, we attempt to replicate occurrence of a specific outcome by weighting various characteristics of an individual (or whichever unit an observation corresponds to). On the other hand, ERGMs attempt to replicate occurrence of a specific network by weighting both network configurations and attributes of individuals within the network. (Lusher et al., 2013) Network configurations are similar to atoms within a molecule: they are basic relationships between individuals. These include effects such as reciprocity between two people (i.e., the tendency to reciprocate a tie, such that if Donna names Chih-‐Ping as a friend, Chih-‐Ping will also name Donna), and transitivity between three people (i.e., the tendency for friends of friends to have a tie, such that if Tom names Jennifer and Genevieve as friends, either Jennifer or Genevieve will name the other as a friend). These network configurations overlap with many others throughout the network to form the overall network structure. Because the final outcome of an ERGM is the network itself, rather than an outcome like dietary intake, we examine if there is a significant similarity between friends (i.e., if there is homophily) in their dietary intake, 21 much like in cross-‐sectional data where conclusions are limited to non-‐causal relationships as well. (Lusher et al., 2013) By modeling the underlying social processes related to youth intake of specific foods, this study fills a gap in our knowledge of the impact of the social environment on obesity-‐related behaviors. Study 2: Impact of the structured learning environment on obesity-‐related behaviors Education in a structured learning environment, such as in a traditional classroom where there is direct, active instruction, is an effective way to improve student knowledge and performance. (Doyle & Rutherford, 1984) Structured learning environments (as opposed to unstructured environments) are particularly beneficial for lower performing students, (Doyle & Rutherford, 1984) so they may be especially beneficial for students with minimal previous experience practicing healthy behaviors. Educational interventions are one of the more popular approaches to reduce childhood obesity and other health problems. These are likely popular because of their general effectiveness, (Waters et al., 2011) and a meta-‐analysis of school-‐based randomized controlled trials (RCTs) revealed these programs to be effective in reducing overweight/obesity, relative to controls (OR=0.74, 95% CI=0.60, 0.92). (Gonzalez-‐ Suarez, Worley, Grimmer-‐Somers, & Dones, 2009) Educational environments may be stressful for students if workload demands are difficult to meet, (Torsheim, Aaroe, & Wold, 2003), and may restrict physical activity opportunities and access to healthy foods. However, several structured education programs in schools have been effective in improving diet and PA, (Caballero et al., 2003; J. N. Davis et al., 2011; Gonzalez-‐Suarez et al., 2009; Lytle et al., 1996; Story, 1999; Zenzen & Kridli, 2009) and there is also some evidence that mindfulness-‐based instruction in schools can reduce stress. (Mendelson et al., 2010) It is also worthwhile to note that while the majority of children’s structured learning takes place in schools, this can also occur in the community or home. (J. N. Davis et al., 2007; Robinson-‐O'Brien, Story, & Heim, 2009) Yet, schools are a popular setting for educational interventions because a large number of students can be reached at one time. (Zenzen & Kridli, 2009) Garden-‐based interventions in schools One promising and novel type of school-‐based educational program to reduce obesity is the garden-‐based intervention. Thus far, the only studies published examining gardening 22 interventions and obesity (as opposed to effects on diet) have been pilot programs. One of these studies, a community-‐based gardening education program with a pretest-‐posttest design found that 17% of overweight/obese children improved their BMI classification following a 7-‐week program. (Castro, Samuels, & Harman, 2013) The other study with a documented link between gardening and obesity is the LA Sprouts program. (J. N. Davis et al., 2011; Gatto, Ventura, Cook, Gyllenhammer, & Davis, 2012; Gatto, Martinez, Spruijt-‐Metz, & Davis, under review; Martinez, Gatto, Spruijt-‐Metz, & Davis, 2015) LA Sprouts is a 12-‐week gardening, cooking and nutrition program in predominantly Latino elementary school students, and aims to improve obesity, and FV intake and dietary determinants. (Martinez et al., 2015) The quasi-‐experimental pilot for this program found that overweight students significantly decreased their weight and BMI relative to control students. (J. N. Davis et al., 2011) Following up on the pilot, a RCT was conducted with 320 students who were enrolled in an after-‐school program (during which the intervention took place) in four Los Angeles elementary schools. As a result of this intervention, participating students had significantly greater decreases in BMI z-‐score (p=0.01) and waist circumference (p<0.001), compared to controls. (Gatto et al., under review) The primary mechanism through which gardening interventions are purported to reduce obesity is via improved dietary intake. Other secondary mechanisms include decreased sedentary activity, which has not been examined to date and is an important area for future research; and decreased stress, which is discussed in the physical environment section below. (van den Berg & Custers, 2011) Gardening may be more valuable than other traditional nutrition education programs because it increases children’s exposure to and familiarity with FV. (Heim, Stang, & Ireland, 2009; Robinson-‐O'Brien et al., 2009) Increased exposure is associated with greater preferences, (Anzman-‐Frasca, Savage, Marini, Fisher, & Birch, 2012; Cooke, 2007) and preferences are a well-‐documented predictor of intake. (McClain et al., 2009) The importance of exposure and familiarity is supported by baseline data from the LA Sprouts RCT, in which willingness to try FV (a construct influenced by exposure (Birch & Marlin, 1982; Birch, McPhee, Shoba, Pirok, & Steinberg, 1987)) was found to be associated with intake. (Martinez et al., under review) Secondly, it is hypothesized that by teaching children about the origins of food and giving them personal experience with growing their own foods, they will be more likely to make healthy choices. (Blair, 2009) Finally, gardening may be an especially relevant approach in low-‐income communities where 23 there is limited access to fresh produce. (Cummins, 2005) Gardening provides access to high-‐ quality FV at little monetary cost, (Carney et al., 2011) and therefore may affect dietary behavior by changing the physical neighborhood, school, or home environment, independent of learning activities related to gardening. To support these theoretical hypotheses, many studies show that gardening programs are effective in improving FV intake. (A. Evans et al., 2012; Hermann et al., 2006; Lautenschlager & Smith, 2007; Robinson-‐O'Brien et al., 2009; Somerset & Markwell, 2008) As with studies examining the impact of gardening on obesity, many of these studies are exploratory in nature and have methodological limitations, such as non-‐randomized designs. One methodologically sound study in sixth grade students found that those who participated in a gardening and nutrition program had an approximately 2.5 serving/day increase in FV intake, which was significantly more than controls or students with nutrition education only. (McAleese & Rankin, 2007) Additionally, a study of elementary and middle schools with initiatives to improve school food (via gardening programs and other mechanisms), found that schools with a higher program intensity had improvements in FV intake of about 1 serving/day, whereas students at schools with low program intensity decreased their intake. (M. C. Wang et al., 2010) In the LA Sprouts program, the pilot intervention showed an improvement in dietary fiber in program participants, relative to controls, (J. N. Davis et al., 2011) as did the RCT. (Gatto et al., under review) Also, results from the LA Sprouts RCT indicate that intervention students had a lesser decrease in vegetable intake than controls. (Gatto et al., under review) The finding that both groups decreased in their intake is surprising, but may be an artifact of the tool used to measure intake (which has limitations with its use in this population), (Garcia-‐Dominic et al., 2012) or seasonal effects (i.e., dietary habits are likely worse right before winter break, when posttest data were collected for over half the participants). There are also several studies demonstrating a significant impact of school garden-‐ based intervention programs on dietary determinants. (Robinson-‐O'Brien et al., 2009) These include, but are not limited to the following determinants: attitudes towards eating FV, FV preferences, willingness to taste FV, nutritional knowledge, motivation to eat FV, and self-‐ efficacy to garden, cook and eat FV. (A. Evans et al., 2012; Lineberger & Zajicek, 2000; McAleese & Rankin, 2007; Morgan et al., 2010; Parmer, Salisbury-‐Glennon, Shannon, & Struempler, 2009; Ratcliffe, Merrigan, Rogers, & Goldberg, 2010; Robinson-‐O'Brien et al., 24 2009; Somerset & Markwell, 2008; M. C. Wang et al., 2010; W. Wright & Rowell, 2010) In the LA Sprouts pilot study, preferences for FV improved, as did attitudes about garden-‐grown vegetables. (Gatto et al., 2012) The results of the RCT indicate that LA Sprouts participants increased in their ability to identify of vegetables, in their self-‐efficacy to eat FV and related behaviors, and were more likely to report gardening at home. (J. N. Davis, Martinez, Spruijt-‐ Metz, & Gatto, in press) Study 2 description Many of the previously mentioned dietary determinants have been demonstrated to be significant correlates of FV intake in various settings. (McClain et al., 2009) However, it is not clear which of these determinants have an impact on FV intake or the ultimate target, obesity, as a result of a school-‐based gardening intervention. Furthermore, although most, if not all, garden-‐based intervention programs used theory to guide their development, none have fully examined the soundness of these theoretical assumptions. Meditational analysis examining the foundational theory of obesity interventions is an area generally lacking in research. (Lubans, Foster, & Biddle, 2008; Spruijt-‐Metz, 2011) Study 2 will utilize a path model to examine the following: 1) change in FV intake as a mediator between change in FV determinants and change in BMI over the 12-‐week intervention period, 2) the LA Sprouts intervention as a moderator of the aforementioned relationships, and 3) strength of associations between change in FV determinants and change in FV intake. Effects of the intervention on FV determinants, FV intake and BMI have already been examined and were described in the previous section. However, differences between the intervention groups in the impact of changing dietary determinants on change in obesity, with change in FV intake as a mediator, have not yet been examined. There are several strengths of this study. The first is the use of a path model, which has not been previously employed to examine effects of garden-‐based interventions. This technique will allow for investigation of the interventional theory, specifically the impact of determinants, and mediating effects of diet, on obesity, resultant from an educational program. Secondly, this study is one of the few garden-‐based programs to measure obesity as an outcome. The intervention has already been determined to be effective in reducing obesity, but the question remains as to which direct targets of the intervention (motivation, self-‐ efficacy, willingness to try FV, knowledge, preferences, attitudes) contributed to this change 25 among program participants. Finally, a strength of the LA Sprouts study is the RCT design, which reduces threats to validity, and few other studies have used. Study 3: Impact of the physical environment on obesity-‐related behaviors The physical environment plays a major role in obesity-‐related behaviors. (K. M. Booth, Pinkston, & Poston, 2005; Davison & Lawson, 2006; Ding & Gebel, 2012; Dunton, Kaplan, Wolch, Jerrett, & Reynolds, 2009a; Papas et al., 2007) For children, the physical environment generally encompasses three major areas: home, school and community. The impact of these areas on diet and PA has been well-‐studied and is described briefly below. Physical environment and diet The neighborhood environment plays a significant role in child dietary intake, and a review of 54 articles found that people living in neighborhoods with poorer access to grocery stores, and greater numbers of convenience stores and fast food outlets, had less healthy diets. (Larson, Story, & Nelson, 2009) The school food environment can also impact diet via the National School Lunch Program and the School Breakfast Program. Participation in these programs is associated with greater childhood obesity, (Millimet, Tchernis, & husain, 2010; Schanzenbach, 2009) but one caveat is that students have non-‐random participation in such programs. (Millimet et al., 2010) Studies examining diets of lunch program participants found that these children do have higher intakes of several micronutrients (including sodium), but also have diets higher in fat, compared to non-‐participants. (Gleason & Suitor, 2003; Gordon, Devaney, & Burghardt, 1995) Finally, the physical home environment can also impact dietary intake through the availability and accessibility of specific foods. (Jago, Baranowski, & Baranowski, 2007; Pearson, Biddle, & Gorely, 2008) Yet this relationship may not be as straightforward as it appears: for example, only child perception of vegetable availability, rather than actual availability, in the home environment is associated with vegetable intake. (Cook, O'Reilly, DeRosa, Rohrbach, & Spruijt-‐Metz, 2014a) Physical environment and physical activity With respect to PA, a recent review of 103 articles identified several neighborhood characteristics to be associated with youth PA, including neighborhood walkability, traffic speed and volume, access to parks and other recreational facilities, residential density, and 26 land-‐use mix, among others. (Ding, Sallis, Kerr, Lee, & Rosenberg, 2011) Schools also provide opportunities for PA in addition to structured physical education classes, such as during recess or lunch periods. (Haug, Torsheim, & Samdal, 2008; Haug, Torsheim, Sallis, & Samdal, 2010) A study of youth in 130 Norwegian schools revealed that students with greater access to outdoor PA facilities, such as a soccer field or playground, were more likely to be physically active. (Haug et al., 2010) A positive association between areas for PA in the schoolyard and youth PA are also seen in Canadian and US youth. (Cradock, Melly, Allen, Morris, & Gortmaker, 2007; Nichol, Pickett, & Janssen, 2009) Finally, the home physical environment may have a limited influence on PA in some populations, (Davison & Lawson, 2006) but may be more important in terms of sedentary time, specifically with respect to the presence of television sets. (Dennison, Erb, & Jenkins, 2002; Roemmich, Epstein, Raja, & Yin, 2007; Saelens et al., 2002) Physical environment and stress An area of research on the obesogenic role of the physical environment in youth that has not been as extensively researched (relative to a large body of research on diet and PA) is the environmental impact on stress. Characteristics of home, school and community settings are tend to overlap (for example, housing variables like high-‐rise versus low-‐rise homes are relevant to both the home and community environment) (G. W. Evans, 2003; G. W. Evans, Wells, & Moch, 2003), and environmental variables associated with stress, like overcrowding, noise, and substandard facilities, can occur at each of these levels. (G. W. Evans, 2003; G. W. Evans et al., 2003; G. W. Evans & English, 2002; Sandel & Wright, 2006) By using state-‐of-‐the art technologies to assess specific geographic context, as will be done in this proposal, the impact of particular environmental factors can be examined across environmental settings (i.e., home, school and community). The physical environment plays an important role in stress and self-‐rated health, whereby negative built environment characteristics strengthened the association between greater stress and poor health in over 4,000 residents is Philadelphia. (S. A. Matthews & Yang, 2010) Yet, there are some elements of the physical environment that have shown a protective effect against stress. One of these is green space, and this will be the focus of the proposed study on this topic. Evidence of this association and the theoretical justification for this relationship is described in detail below. Green space was chosen as a focus because 27 gardening, the focus of study 2 and one type of activity conducted in a green environment, has some evidence to be preventive against stress. In one intervention, participants who experienced stress took part in a stress-‐reduction gardening program, and said that the enjoyment of gardening created an environment for relaxation, and promoted participation in other stress-‐reducing activities. (Eriksson, Westerberg, & Jonsson, 2011) Also, in an experimental design in which adult participants were required to perform a stressful task, cortisol levels and positive mood were significantly better in participants after they subsequently completed an outdoor gardening activity, compared to indoor reading. (van den Berg & Custers, 2011) One way in which green space can improve stress is by providing opportunities for PA, (Dyment & Bell, 2007; A. C. K. Lee & Maheswaran, 2011) which can decrease stress, (Fox, 1999) but there is evidence that PA does not fully account for the relationship between green space and stress. (Maas, Verheij, Spreeuwenberg, & Groenewegen, 2008) Only a limited number of studies have examined the association between stress and green space, and even fewer had a strong study design (or large sample size). In one study of over 11,000 adults in Denmark, it was found that individuals living farther from green space reported higher levels of stress than those that lived closer, and that individuals who did not report stress were more likely to visit green spaces (analyses were adjusted for SES). (Stigsdotter et al., 2010) Another study of 112 young adults found that those who had just completed a drive in a car or other attentionally-‐demanding task had greater decreases in diastolic blood pressure when sitting in a room with a nature view versus sitting in a room with no window. (Hartig, Evans, Jamner, Davis, & Gärling, 2003) This same study found that walks in a nature reserve produced greater decreases in blood pressure relative to walks in an urban setting. (Hartig et al., 2003) Also, a pilot study of 25 adults found significant inverse relationships between green space quantity, measured via census data, and both self-‐reported stress and salivary cortisol (although it is not clear if controlled for SES). (Thompson et al., 2012) Furthermore, in a study of Japanese adults, ‘forest bathing’ (taking in the forest atmosphere) was associated with decreased salivary cortisol, heart rate, and reported tension and anxiety, relative to being in an urban setting. (J. Lee et al., 2011; Park et al., 2011) There is also some evidence that access to and activities in green space have positive mental health outcomes. (A. C. K. Lee & Maheswaran, 2011; R. Mitchell, 2013; Sugiyama, Leslie, Giles-‐Corti, & Owen, 2008; van Dillen, de Vries, Groenewegen, & Spreeuwenberg, 2012) An analysis of 10 studies found a 28 positive link between green space activity and both self-‐esteem and mood (which was especially strong with the presence of water) (Barton & Pretty, 2010), and a meta-‐analysis of 25 studies indicates that anxiety is also improved following activity in a natural environment (Bowler, Buyung-‐Ali, Knight, & Pullin, 2010). Although there are few other studies with good validity linking neighborhood green space to lower stress levels (to our knowledge), there is a strong theoretical underpinning for this association. One major theory guiding this area of research is Ulrich’s Stress Reduction Theory, in which natural landscapes are pleasing and therefore increase positive affect, hold attention, and subsequently decrease incidence of negative or stressful thoughts. (Bratman, Hamilton, & Daily, 2012; Ulrich, 1986; Ulrich et al., 2014) A second theory, Kaplan’s Attention Restoration Theory, posits that since stress can result from taxing tasks that require directed focus, natural environments can be restorative because they do not require directed attention and can allow the mind to rest. (Berto, 2005; Bratman et al., 2012; Herzog, Black, Fountaine, & Knotts, 1997; S. Kaplan, 1995; van den Berg, Hartig, & Staats, 2007) Additionally, there is high compatibility in nature, meaning the environment is restorative in that it meets the needs for desired activities, and there is a respite from environments that require directed attention. (van den Berg et al., 2007) Although these theories have some overlap, the first focuses on affect and emotional state, whereas the second centers on attention. (Bratman et al., 2012) These two theories have driven the majority of research in this area, but other hypotheses exist as to how green space can reduce stress. One is that when humans feel a connectedness with nature, we feel we belong to something greater than ourselves, which has a positive effect on our well-‐being. (Bratman et al., 2012; Cervinka, Roderer, & Hefler, 2012; Howell, Dopko, Passmore, & Buro, 2011) Another theory uses an evolutionary perspective, and hypothesizes that humans feel a preference for natural land because of the resources our ancestors found there, (Verheij, Maas, & Groenewegen, 2008) which likely impacts affect, such as in Ulrich’s theory. There is also evidence that landscape plants, such as trees and small hedges, have the ability to reduce the noise level of the environment, (F. Yang, Bao, & Zhu, 2011) which, as previously mentioned, can induce stress. One important caveat in this area of research is that some green environments may be perceived as unsafe, for example urban parks, and would not be expected to decrease stress. 29 Study 3 description Although there is strong theoretical support and some empirical evidence for an association between green space and stress in adults, very few well-‐executed studies exist that examine this association in children. Children are an important population to study on this topic, not only because of the potential obesity and metabolic benefits, but greater childhood visits to nature are associated with individuals being more likely to visit nature as adults, meaning these behaviors can have long-‐term effects. (Thompson, Aspinall, & Montarzino, 2007) One study found a significant inverse association between percent neighborhood park area and perceived stress in adolescents, after controlling for both PA and SES. (Feda et al., 2014) Another study found neighborhood greenness to be inversely associated with child BMI after 2 years, but did not examine stress as a possible mechanism for this relationship; the authors merely theorized about its role. (J. F. Bell, Wilson, & Liu, 2008) Yet, none of these studies have examined the relationships between stress and environment in real-‐time, using ecologically valid designs that are now available through the use of new technologies including mobile phones and sophisticated location tracking. Study 3 will examine the effect of a green space environment on self-‐reported stress in children, using ecological momentary assessment (EMA) and global positioning system (GPS)-‐derived data from the Mobile Healthy PLACES study. (Dunton et al., 2014; Dunton, Kawabata, Intille, Wolch, & Pentz, 2012; Dunton, Liao, Intille, Spruijt-‐Metz, & Pentz, 2009b; Dunton, Liao, Intille, Wolch, & Pentz, 2011) The purpose of the Mobile Healthy PLACES is to examine the context of physical and sedentary activity in children; including mood, location, social company and enjoyment; using real-‐time self-‐report data. (Dunton et al., 2009b) In this sample of 121 children aged 9-‐13 years, there was lower intra-‐individual variability of negative affect (which is measured through four items: stress, mad/angry, nervous/anxious, sad) in children with more moderate-‐to-‐vigorous physical activity (MVPA). (Dunton et al., 2014) Also, negative affect was lower when children exercised with friends, compared to alone or with their family. (Dunton et al., 2011) Furthermore, more MVPA occurred outdoors, relative to at home or a friends house, and enjoyment of MVPA was also greater outdoors. (Dunton et al., 2011) When children exercised outdoors, they were most frequently in a park or on a trail, (Dunton et al., 2012) and environmental greenness was also associated with MVPA. (Almanza, Jerrett, Dunton, Seto, & Pentz, 2012) Finally, demographic differences for exercising outdoors were 30 seen, such that the following groups were more likely to exercise outdoors: older children; children from households with higher annual income; Black, Latino and biracial children; and normal-‐weight and overweight children (relative to underweight children). (Dunton et al., 2012) The use of EMA data is a major strength of this study. There are several benefits to the use of EMA: 1) decreased reporting bias, primarily since data collected are not based on recall; 2) use of repeated measures to capture dynamic processes; and 3) observations can be made in the environments that individuals regularly inhabit. (Smyth & Stone, 2003; Stone & Shiffman, 1994) These benefits generally lead to more accurate reflections within the context of space and time. (Smyth & Stone, 2003) Use of EMA is especially beneficial in this study because stress is generally contextual and will fluctuate throughout the day based on settings and experiences. (Zimring, 1982) Another methodological strength of this study is in the use of GPS-‐derived variables, which allow for environmental green space to be measured objectively, before and during the EMA prompt for stress. Not only does this data allow for the exploration of temporal relationships, but other observational studies only consider the broader neighborhood context where children live, and this study will examine actual green space use. Another strength of this study is that PA will be well-‐controlled, given that accelerometry data was also collected on study participants, so PA will not be a possible confounder. Household income will also serve as a control variable. Finally, as mentioned previously, few studies on this topic have been conducted in children, and this multi-‐ethnic sample from households with varying levels of income is rich in diversity. (Dunton et al., 2014) The use of these progressive methodologies makes this study very unique in having strong internal validity to measure the effect of green space on stress. 31 Aims and Hypotheses The specific aims of this proposal are as follows: Specific aim 1: To examine the similarity between youth friends in their intake of FV (a positive health behavior) and soda (a negative health behavior). Hypothesis 1: There will be evidence of similarity between named friends and their reported intake of FV. Hypothesis 2: There will be evidence of similarity between named friends and their reported intake of soda. Specific aim 2: To examine the impact of a school garden-‐based nutritional education program on FV intake as a mediator between FV determinants and BMI. Hypothesis 1: Change in FV intake will mediate the association between change in FV determinants (motivation to eat FV, self-‐efficacy to eat FV, identification of FV, willingness to try FV and preferences for FV) and change in BMI. Hypothesis 2: The LA Sprouts intervention will moderate the aforementioned relationships (as listed in Hypothesis 1) . Post hoc exploratory analysis: Compare strength of association between change in FV determinants (as listed in Hypothesis 1) and change in FV intake. Specific aim 3: To examine the impact of open recreational space (ORS) and vegetation density on child reported momentary stress. Hypothesis 1: Children with greater time in ORS over the entire monitoring period will have lower overall mean reported stress. Hypothesis 2: Children in areas with higher mean vegetation density over the entire monitoring period will have lower overall mean reported stress. Hypothesis 3: Children will report lower stress in an ORS, relative to not in an ORS, when they are in this environment both at the time of self-‐report and in the 30 minutes prior. Hypothesis 4: Children will report lower stress in an environment with greater vegetation density, relative to less vegetation-‐dense environments, when they are in this context both at the time of the self-‐report and in the 30 minutes prior. 32 Summary of Contribution As a whole, this dissertation proposal will contribute to our knowledge of environmental influences on obesity-‐related behavior in the following ways: • Determine which protective and obesogenic foods; specifically fruit, vegetables and soda; have shared consumption patterns among adolescent friends. • Evaluate the theoretical underpinning of a 12-‐week school garden-‐based educational program by examining the role of changing fruit and vegetable intake as a mediator between change in dietary determinants and obesity. • Document associations between green space and self-‐reported stress in children in real time, using an observational design with high internal validity. Rates of obesity remain high, especially in minority youth, (Ogden, 2012) and because genetic factors are not highly modifiable over a lifetime, it is imperative that youth are empowered to make healthy behavioral choices in order to decrease their risk for disease. While a broad literature exists on determinants of obesity-‐related behaviors, US youth continue to practice poor dietary behaviors, (Kimmons et al., 2009) participate little in physical activity, (Troiano, Briefel, Carroll, & Bialostosky, 2000) and experience high rates of stress. (Anda et al., 2000) The environment in which children are raised holds a strong influence on their behaviors, (Bronfenbrenner, 1997) and these environment-‐behavior relationships must be better understood in order to enact positive changes. Opportunities exist in various environmental domains; social, structured learning and physical; and results of these studies can help drive the policy and intervention efforts needed to curtail childhood obesity. 33 CHAPTER 2: Similarities in consumption of fruit, vegetables and soda among middle school friends Introduction Social relationships are impactful in both adult and child obesity, and individuals with overweight friends are themselves more likely to be overweight. (Christakis & Fowler, 2007; Trogdon, Nonnemaker, & Pais, 2008; Valente et al., 2009) Although shared norms for obesity (i.e., the acceptability of being overweight) are considered one of the main factors contributing to this relationship, shared behaviors may a more important driver. (Hruschka, Brewis, Wutich, & Morin, 2011) Data on physical activity in adults and children indicate that individuals are more likely to be active when they share activities with friends. (M. L. Booth, Owen, Baum, Clavisi, & Leslie, 2000; Voorhees et al., 2005) Dietary practices can also be highly influenced by peers, and are hypothesized to work through several mechanisms. These include 1) social facilitation, such that eating practices differ when eating alone versus with others; 2) modeling and attempted mimicry of others’ eating behaviors; 3) impression management, such that individuals wish to project a certain image of themselves via their dietary choices; and 4) norms, or the acceptability of eating a specific type of diet. (Salvy et al., 2012) Peer relationships are particularly important to adolescents, (B. B. Brown, 2004), and US adolescents often fail to meet dietary recommendations. (Kimmons et al., 2009) Therefore, understanding the relationship between friendships and youth dietary behaviors may be valuable for childhood obesity prevention and treatment. Several studies have demonstrated associations between dietary intake by a friend (or, alter) and intake by an individual of interest (or, ego). Alter intake has been linked to ego intake of high-‐calorie foods in Australian youth, (de la Haye et al., 2010; 2013) and adolescent friends share similar patterns of eating in fast food restaurants in the US. (Ali et al., 2011; Bruening et al., 2014) Some data also exist that link adolescent alter and ego intake of specific food types, namely sugar-‐sweetened beverages (SSB), dairy, whole grains, and vegetables. (Bruening et al., 2012; 2014; Wouters et al., 2010) Although this growing area of research is promising, a major methodological limitation exists in some of these studies. Few studies on this topic thus far have accounted for social network structure: the greater community of interrelated individuals, more than just a given 34 individual and his/her close friends (de la Haye et al., 2010; 2013). Considering a more comprehensive social network allows researchers to observe not only an ego’s interaction with people with whom he/she has direct ties (i.e., a reported relationship such as a friendship), but also an ego within the greater community, and the overall properties and organization of the community. (Valente, 2010) Furthermore, network data violates independence assumptions of regression approaches because of the complex interdependencies among individuals. (Lusher et al., 2013) Exponential random graph modeling (ERGM) is a method developed to test hypotheses while accounting for network characteristics, and is described in detail in the methods section of this manuscript. To our knowledge, a modeling approach which accounts for network interdependencies has not be utilized to examine intake of specific foods, namely soda and fruit and vegetables (FV). These examples of simple and complex carbohydrates, respectively, are of interest given the importance of this macronutrient in obesity development. (Samaha et al., 2003) SSB have been shown to lead to weight gain in youth, (Berkey, Rockett, Field, Gillman, & Colditz, 2004; Ludwig, Peterson, & Gortmaker, 2001; Malik et al., 2006) and data suggest that SSB not only contribute to a higher overall energy intake, but also replace healthier beverages, like milk, in children’s diets. (Harnack, Stang, & Story, 1999) FV intake has been linked to reduced obesity and decreased risk for type 2 diabetes (to which SSB is also strongly linked). (Ford & Mokdad, 2001; Ledoux, Hingle, & Baranowski, 2010; Malik et al., 2010; Montonen et al., 2005) Furthermore, only 1% of US adolescents consume the recommended amount of FV/day, (Kimmons et al., 2009) which indicates that understanding drivers of intake, such as peer influence, may be important for reducing childhood obesity. This study examines middle school children and the similarity between friends in their consumption of soda and FV. In this sample, it has been previously shown that overweight children are likely to have overweight friends, (Valente et al., 2009) and this study aims to determine if dietary behaviors may be partially contributing to this relationship. Existence of a shared characteristic among those with direct network ties is also known as homophily, (Valente, 2010) and we hypothesize that among youth in these classroom networks, there will be homophily between friends in soda and FV intake. Follow-‐up exploratory analyses will also estimate the proportion of communities (groups of friends) within middle school classroom networks that exhibit homophily on dietary behaviors. Friendship dynamics have not been well considered in programs for obesity prevention or treatment, but some meaningful 35 intervention approaches may be to target opinion leaders or to form peer learning and support groups. (Valente, Hoffman, Ritt-‐Olson, Lichtman, & Johnson, 2003) These findings will help clarify the role friends play in shaping obesity-‐related behaviors. Methods This study utilizes cross-‐sectional baseline data from the Get Moving! study, a classroom media-‐based intervention to reduce sedentary activity in middle school children. The recruitment strategy and methods for this study have been described in detail elsewhere (Nguyen-‐Michel et al., 2007; Spruijt-‐Metz et al., 2008). Briefly, seven schools in the Los Angeles area participated in this program, and 85% of students and parents provided informed assent and parental permission. Seventeen classrooms (individual networks) participated. Data collected in the student questionnaire included demographic information; physical activity behavior, attitudes, barriers, and social support; smoking behavior; body image; school performance (perceived) and engagement; mental health status; pubertal status; dietary behaviors; sensation-‐seeking attitudes; and classroom friendship networks. No inclusion or exclusion criteria were applied to the study sample for use in these analyses. This study was approved by the University of Southern California Institutional Review Board. Demographic measures Student age, sex, ethnicity, height and weight were collected. Ethnicity was collapsed into four variables: Black, Hispanic, Asian/Pacific Islander (consisting of those reporting as Chinese, Filipino, Korean, Pacific Islander, Vietnamese), and White. Weight was collected using a Tanita TBF 300/A (Arlington Heights, IL), and height was measured using a Seca Mobile Height Rod (Birmingham, UK). Participant BMI percentile for age and sex was calculated based on guidelines from the Centers for Disease Control and Prevention (Kuczmarski et al., 2002). Dietary intake measures Frequency of FV and soda consumption was measured. (Nguyen-‐Michel et al., 2007; Willett et al., 1985) A single item on frequency of “soda-‐ not diet (1 can or glass)”, had seven response options ranging from “never/less than 1 per month” to “more than 3 cans per day”. A single item on frequency of “fruit, fresh or canned (not containing juice)” was measured 36 with six response options, ranging from “less than once a week” to “2+ a day”. There were three items on vegetable frequency, including “green salad”; “vegetable soups, or stew with vegetables”; and “any other vegetables, including string beans, peas, corn, broccoli or any other kind”, all with the same response options as the fruit item. Items on FV juice and potatoes were not included in these analyses. For descriptive purposes, the categorical response scores were converted into an average daily frequency. For example, a response of eating fruit once per week was converted to a frequency of eating fruit 0.14 times/day (if a range was given, the mean value was used; 0.5 was used to represent “less than 1”, 2 was used to represent “2+” and 4 was used to represent “more than 3”). To consolidate vegetable items into one value, the frequency of all three items was summed. Friendship network measure Participants were asked to nominate their five best friends in class. For these analyses, relationships between friends were directed, such that friendships were not assumed to be reciprocated (i.e., data were unidirectional, not bidirectional). Analyses Network graphs and summary statistics were obtained for all 17 networks using the sna (Butts, 2008; 2010; 2014) and igraph (Csardi & Nepusz, 2006) packages in R (version 3.0.1, 2013 The R Foundation for Statistical Computing, (Ihaka & Gentleman, 1996). To visualize networks, they were plotted based on sex, given the strong gender homophily expected in middle school students. (Shrum, Cheek, & Hunter, 1988) Since many classrooms exhibited strong gender segmentation, networks were first examined with boys and girls together, then split by gender. Summary statistics were derived for all networks, and the overall mean(SD) and range was taken across networks for each measure. Intraclass correlations were determined using a multilevel model to describe the level of similarity in fruit, vegetable and soda intake among individuals within classrooms. Correlations between ego intake of fruit, vegetables and soda and intake by named friends of the same food item were also examined. To test whether or not friends shared common dietary behaviors, exponential random graph models (ERGMs) were examined (Lusher et al., 2013; Snijders, Tom A.B., Pattison, 37 Robins, & Handcock, 2006), using the statnet suite in R. (Handcock, Hunter, Butts, Goodreau, & Morris, 2003) Briefly, ERGMs attempt to replicate an observed network structure using purely structural effects (relationships among individuals) and individual-‐level attributes, or characteristics of the people within the network. ERGMs compare the observed network to other possible network configurations, and Markov chain Monte Carlo (MCMC) sampling is used to create a distribution of graphs (given the exponentially large number of potential configurations) where the observed data are central to the distribution. Parameter estimation is achieved via maximum likelihood and approximate Wald tests are used to determine statistical significance. A parameter coefficient of zero indicates that the effect of the variable is consistent with chance, a positive coefficient indicates the effect occurs more often than due to chance alone, and a negative coefficient indicates the reverse, given other effects in the model. The following a priori network covariates were used in the models, as recommended by Lusher et al.: ties/edges (number of ties, also known as degree), reciprocity (tendency for friendship nominations to be reciprocated), simple connectivity (the extent to which people who send nominations also receive them), popularity spread (distribution of nominations received), connectivity spread/expansiveness (distribution of nominations sent), triangulation (tendency for two people nominated by one person to also share a tie), cyclic closure (tendency for nominations to exist A → B → C → A), and multiple connectivity (tendency for two unconnected individuals to have two-‐degree connections through multiple individuals). (Lusher et al., 2013) The following a priori individual-‐level covariates were also included in the models: age, sex, ethnicity and BMI percentile. The dietary variables were examined in three separate models, each including one food type only. In each model, effects of diet were included for “sender” (the extent to which people with a given characteristic nominate friends, compared to others; ‘nodeocov’ term in statnet suite), “receiver” (the extent to which people with a given characteristic receive friendship nominations, compared to others; ‘nodeicov’ term in the statnet suite) and homophily (the extent to which people nominate friends with a shared characteristic; the main outcome of interest in these analyses). Homophily was measured using the ‘absdiff’ term in the statnet suite. This measures the absolute difference in an attribute among individuals who share a tie. Thus, for this parameter, a negative parameter indicates there is less differences between individuals (i.e., more similarity, or homophily), than is expected by chance alone, and a positive 38 parameter indicates there is more difference (i.e., less similarity, or more heterophily) than is expected by chance. Goodness of fit was assessed by examining diagnostic plots and the differences between observed and sample statistics. If there was a significant difference at p<0.05 between the sample statistics and the observed statistics, then backward selection was used until good fit was achieved. In some cases, individual network models were not able to achieve appropriate fit, and were excluded from these analyses. To summarize homophily, sender, and receiver effects across all networks, a meta-‐analysis was conducted using the method described by Snijders and Baerveldt. (Snijders, Tom AB & Baerveldt, 2003) For follow-‐up exploratory analyses on the prevalence of homophily within communities, groups of friends within a classroom network were identified using the Walktrap Community Detection Algorithm from the ‘igraph’ package in R. (Csardi & Nepusz, 2006; Pons & Latapy, 2006) This algorithm performs random walks on a graph, and walks are more likely to occur within communities given that there are few opportunities to walk along ties that lead to individuals outside of the community. Modularity is used to determine the number of communities within a network. Modularity is a measure used to determine the strength of a network’s division into groups; it compares the amount of ties to other members of the group, relative to what is expected randomly. (Valente, 2010) Figure 2-‐1 provides a sample diagram of a classroom split into communities. ERGM models were run on each of the communities, with the following model terms: ties, reciprocity, simple connectivity, and homophily on the food of interest (measured by absolute difference). A term for cyclic closures was added to models if needed to meet convergence criteria. Individual-‐level attributes (i.e., sex and ethnicity) were not included as covariates in these models, as it was already established that individuals in these communities were densely connected, and the reasons for their connections were not as relevant in these analyses. However, BMI was a relevant covariate, given that individuals with certain BMI’s may strive to eat or avoid certain foods, and individuals with higher BMI’s are likely consume greater quantities of food. In order to include BMI in the models, vegetable, fruit, and soda intake were regressed on BMI, resulting in adjusted measures of dietary intake. Homophily was examined on both raw and adjusted dietary intake measures. 39 Results Descriptive Characteristics Descriptive measures of the study sample are in Table 2-‐1. Six hundred and seventeen participants had at least partial data. They were 36% male, over 40% each Latino and Asian/Pacific Islander, and had a mean age of 12.8±1.0 years. Two hundred and one participants were overweight (34.0%, BMI ≥ 85 th percentile) and 96 were obese (16.2%, BMI ≥ 95 th percentile). On average, students reported eating fruit 0.8±0.6 times/day, eating vegetables 1.5±1.2 times/day, and drinking less than one can or glass of soda/day (mean: 0.8±1.1). Plots of the 17 classroom networks are found in Figure 2-‐2, and summary measures across all networks are in Table 2-‐2 (along with a brief definition of terms). Network sizes ranged from 19-‐52 students, and two classrooms consisted only of females. On average, students were connected to 3.8±0.4 friends, who either nominated them, or that they nominated (or both; reciprocity was high at a mean of 0.9±0.03). Generally networks did not have a structure highly centralized among one or multiple individuals (centralization mean: 0.12±0.04), and students appeared to often have friends with students of the same gender as expected (from graphs). Transitivity was moderate (mean: 0.44±0.10), although varied between networks (range 0.25 – 0.62). Unadjusted analyses Correlations between ego and named friend intake of vegetables, fruit and soda can be found in Table 2-‐3. There were no significant correlations between egos and named friends in their intake of fruit or vegetables, yet there was a significant correlation between ego soda intake and soda intake among all named friends (r=0.09, p=0.02). There was also a significant correlation between ego soda intake and soda intake among the fifth named friend (r=0.10, p=0.04). Furthermore, there were also two correlations significant at p<0.1: between ego soda intake and soda intake of the first named friend (r=0.07, p=0.09) and between ego soda intake and soda intake of the second named friend (r=0.08, p=0.09). Intraclass correlations within classrooms were 0.024 for vegetable intake, 0.017 for fruit intake, and 0.013 for soda intake. 40 Classroom-‐level network effects in the full sample Sixteen classroom networks were used to evaluate the relationship between network ties and dietary intake of vegetables and soda, and all seventeen classroom networks were used to evaluate the relationship between network ties and fruit. Sender effects of vegetable intake were not significant in any of the sixteen classroom networks, and receiver effects were significant in only one classroom. In classroom 17, a positive effect was seen such that students who consumed more vegetables were named as friends more frequently than students who consumed fewer vegetables (0.14 ± 0.07, p=0.047). Heterophily was observed in one network: in classroom 14, there was a significant positive absolute difference, indicating that individuals had differing levels of vegetable intake, compared to their friends (0.20 ± 0.09, p=0.03). Homophily on vegetable intake was not observed in any of the classroom networks. Similar findings were found for fruit intake: there were no classrooms where significant sender effects were exhibited, yet in one classroom there was a significant receiver effect with fruit. In network 2 there was a positive receiver effect, such that those who consumed more fruit received a greater number of friendship nominations from classmates, compared to those who ate less fruit (0.09 ± 0.04, p=0.03). Homophily on fruit intake was observed in one classroom: in classroom 4, there was a significant negative absolute difference, such that individuals and their friends consumed similar amounts of fruit (-‐0.09 ± 0.04, p=0.03). For soda intake, there were not any networks that demonstrated significant sender or receiver effects. There were two networks with significant associations in soda intake between friends: in classroom 2, there was a significant positive absolute difference, such that friends had soda intake that differed from their friends’ intake (0.12 ± 0.05, p=0.02), whereas in classroom 10, there was a significant negative absolute difference, such that individuals shared similar patterns of soda intake with their friends (-‐0.12 ± 0.05, p=0.03). Results from the meta-‐analysis of all classroom networks revealed that there were no significant homophily, sender, or receiver effects in the overall sample for vegetable, fruit or soda intake (Table 2-‐4; all effects p>0.9). Also, in networks where homophily was observed, this is no longer significant following a Bonferroni adjustment for multiple comparisons (revised alpha level: p=0.001). 41 Classroom-‐level network effects in Girls only Sixteen networks were used to evaluate the relationships between vegetable, fruit, and soda intake and female friendships in middle school classrooms. For vegetable intake, there were not any classrooms where significant sender or receiver effects of vegetable intake were observed. Homophily on vegetable intake was observed in only one classroom. In classroom 13, there was a significant negative absolute difference, indicating that there was less difference in vegetable intake between friends than expected (-‐0.34 ± 0.14, p=0.01). For fruit intake, there were no networks where significant homophily, sender or receiver effects were observed. For soda intake, there were no classrooms were significant sender effects were observed, yet there was one classroom where significant receiver effects were found. In classroom 15, there was a significant negative receiver effect, indicating that girls who consumed greater amounts of soda were nominated fewer times as friends than those who drank less soda (-‐0.18 ± 0.09, p=0.04). Homophily on soda intake was observed in one female classroom network, and heterophily on soda intake was observed in two female classroom networks. In classroom 10, there was a significant negative absolute difference in soda intake among female friends, indicating that they were more likely to have similar intake than expected (-‐0.16 ± 0.08, p=0.046). In classrooms 2 and 15 there were significant positive absolute differences in soda intake among female friends, indicating that they were more different in their intake than expected (0.18 ± 0.8, p=0.02; 0.26 ± 0.10, p<0.01; respectively). The meta-‐analysis revealed that there were not any significant homophily, sender or receiver effects for vegetable, fruit or soda intake among girls (for all effects, p>0.9). Similar to findings in the whole sample, significant homophily effects do not remain following a Bonferroni correction for multiple comparisons. Classroom-‐ level network effects in boys only Thirteen classroom networks were used to evaluate the relationship between vegetable intake and male friendships, fourteen networks were used to evaluate male friendships and fruit intake, and twelve networks were used to evaluate male friendships and soda intake. As indicated in Figure 2-‐2, some co-‐ed classrooms had a small amount of boys and thus models for these networks had more challenges converging. 42 For vegetable intake, there were not any networks where significant effects of homophily, sender or receiver were observed. For fruit intake, sender and receiver effects were seen in one classroom each. In classroom 9, there was a significant negative sender effect, such that those boys who ate more fruit nominated fewer classmates as friends (-‐0.82 ± 0.29, p<0.01). In classroom 11, there was a significant positive receiver effect of fruit, such that those boys who ate more fruit were more likely to be nominated as a friend by their classmates (0.28 ± 0.14, p=0.46). Homophily on fruit intake was observed in one network: in classroom 15, there was a significant negative absolute difference in fruit intake among friends, indicating that they were more similar in their intake of fruit than expected (-‐0.44 ± 0.21, p=0.03). For soda intake, sender and receiver effects for boys were observed in classroom 16, but not in any other classrooms. A significant negative sender effect was observed, such that boys who drank more soda nominated fewer classmates as friends (-‐0.53 ± 0.24, p=0.03), and a positive receiver effect was observed, such that boys who consumed more soda were more likely to be nominated as friends by their classmates (0.60 ± 0.26, p=0.02). Heterophily on soda intake was observed in one network of boys, and homophily on soda was not observed in any networks. In classroom 12, there was a significant positive absolute difference in soda intake among boys who were friends, indicating that they were more different than expected (0.26 ± 0.11, p=0.02). From the meta-‐analysis, there were not any significant effects of homophily, sender or receiver interactions with vegetable, fruit or soda intake among boys in this sample (for all effects, p>0.9). Again, no significant effects of homophily in individual networks are significant after a Bonferroni adjustment is made. Community-‐level network effects The total number of communities in sample was 105 (isolates are considered their own community using the Walktrap Algorithm). The average number of communities was 6.2 ± 1.5 communities/classroom (within-‐classroom range: 4-‐9 communities/classroom). For many communities with three or fewer members, ERGM models did not achieve convergence. There were 80 communities in which models on vegetable intake converged, and five communities displayed a significant negative absolute difference of vegetable intake (unadjusted). Seventy-‐four communities were used to measure fruit intake, and four had a 43 significant negative absolute difference on fruit. There were 76 communities in which soda intake was measured, and three had a significant negative absolute difference. Following a Bonferroni adjustment (233 models were run with unadjusted dietary data; revised alpha level: 0.00021), two communities had a significant negative absolute difference with vegetable intake (-‐61.6 ± 2.8, p=2.0e-‐08; and -‐10.0 ± 0.03, p=9.0e-‐51), one community had a significant negative absolute difference in fruit intake (-‐24.7 ± 4.6, p=5.8e-‐5), and one community had a significant negative absolute difference in soda intake (-‐11.8 ± 1.8, p=0.00019). Negative absolute differences indicate that members of these communities were more similar in their dietary intake of the indicated foods than was expected by chance. These communities are pictured in Figure 2-‐3. In ERGM models with BMI-‐adjusted variables, there were no significant effects of homophily in any communities. Discussion Overall, significant homophily on dietary intake of specific obesity-‐related foods (fruit, vegetables, and soda) was not found in this sample of middle school students in Los Angeles. There was a significant correlation between ego soda intake and soda intake of all named friends, but this association was not significant in models including network effects and individual-‐ level covariates. There were some individual classroom networks where friends shared similar dietary intake patterns, but significant findings were likely due to Type 1 error. (It is expected that 1 out of 20 classrooms will demonstrate significant findings with the initial alpha level set at 0.05.) Furthermore, there were not any significant sender or receiver effects of dietary intake, such that those who consumed specific obesity-‐related foods were not more or less likely to nominate friends, or be nominated themselves by classmates. Also, no significant findings were observed when these relationships were examined separately by gender. Findings from this study are not generally consistent with similar published work, although it is possible that there is publication bias on this topic. The EAT-‐2010 study is one of the few larger studies with this information collected, and researchers found similarities among friends in their soda, fruit, and vegetable intake. (Bruening et al., 2012; 2014) However, this study did not use social network-‐specific methodologies that account for overall network dynamics. The one study to utilize network methodologies to examine peer effects on dietary intake is by de la Haye et al., and it was found that middle school students 44 influence their friends’ junk food intake. (de la Haye et al., 2013) One notable difference between the study by de la Haye et al. and this study is in the type of dietary intake examined, as the study by de la Haye et al. examined broader food groups. Perhaps middle school friends do not have similar intake of specific food items, but general categories of food types (such as junk food, or healthy food) may be similar among peers. Homophily on junk food intake was not significant in this sample (data not shown), but differences in results could be attributable to different measurement techniques (i.e., this study asked about specific food items, although candy was not included, whereas the study by de la Haye et al. asked about sweet snack foods, savory snack foods, fast food, and high-‐calorie drinks). It is also possible that behaviors related to obtaining food (like buying lunch at school, or buying soda from a vending machine) are commonly shared among friends. The use of methodologies that account for the entire network structure is an important consideration when interpreting these findings: these analyses considered whether or not there was similarity in dietary intake among friends across entire classroom networks. It is perhaps not reasonable to expect that friends always share similar dietary behaviors, given the numerous factors that impact dietary behavior (such as individual-‐level factors, including preferences and knowledge, and factors in the built environment). (McClain et al., 2009; van der Horst et al., 2007) It is possible that some groups of friends are very similar in their dietary intake, while others are less similar. The magnitude of similarity could be based on factors such as the degree to which values related to appearance are shared, or the frequency of consumption of meals from the same location (which would likely have more similar nutrient composition than foods from differing sources). To examine small groups dynamics, follow-‐up analyses were conducted in order estimate the prevelance of homophily within small groups of friends, and to examine any characteristics of these communities that made them distinct from other communities where homophily was not present. Surprisingly, there were very few communities where there was significant homophily on dietary intake. However, there are some communities that appear to have similarities in dietary intake, even though statistically significant effects were not seen. For example, many individuals in the community pictured in Figure 2-‐4 have similar levels of vegetable intake. In this specific community, the measure of absolute difference was 0.06 ± 0.26, p=0.81, and this is supported by visual inspection, given that there are a large number of ties between individuals with different levels of intake. However, it is clear from this 45 community plot that there are similarities in intake among many members of this community, especially between 2 nd and 3 rd degree connections. Given that the ERGM models examine the probability of a tie, and there is already high probability of ties in these communities, there may be some other methodology that could be more meaningful to address the question of homophily within a small community of middle school friends. This is a relevant topic for future research. In addition to the measure of homophily having implications on interpretations of findings, the dietary variables used could bias findings towards null. Participants were asked one question each about the frequency of their fruit and soda consumption, and three questions on the frequency of their vegetable intake over an average week (or month, in the case of soda). Not only is there often bias in reporting of dietary intake among teens, (A. K. Ventura, Loken, Mitchell, Smiciklas-‐Wright, & Birch, 2006) but it may also be difficult for individuals to estimate their habitual intake over such a broad range of time. Additionally, only the similarity of dietary behaviors with classroom friends is examined in this study, and does not consider behaviors of friends outside the classroom, or family dietary intake. For middle school students, their dietary intake is likely largely influenced by the dietary practices of family members, particularly parents, who may prepare all or most of their meals. Another limitation is the cross-‐sectional design, which prevents causality from being examined in this study. Furthermore, ERGM models measure selection, or the presence of ties that form a network, rather than the effect that friends have on each other, as can be examined with longitudinal network data. However, one strength of this study is in the large sample used, which allows for examination of many differing and complex social dynamics within classrooms. The larger sample allows for better generalizability, and also allows for comparisons to be made between groups, as was done with the follow-‐up community analysis. Although there was no overall similarity among friends in their dietary intake of fruit, vegetables and soda, there do appear to be some groups of students where homophily on intake is practiced. Further research should be done on this topic to better determine the prevalence of homophily on dietary intake in youth, and should utilize measures relevant to this specific research question. Also, future research should examine why these behavioral similarities exist within some groups and not others, for example, whether it is due to a deliberate effort among group members, or if it is a function of external influences (such as 46 food availability). Better understanding the role of friends in influencing dietary behavior will inform determinants of intake and pave the way for network-‐based interventions to be used to improve dietary behaviors in children. Table 2-‐1: Baseline characteristics of 617 middle school students in Los Angeles Demographic characteristics n (%) Sex (male) 220 (35.7%) Ethnicity: Asian/Pacific Islander 234 (40.3%) Black 26 (4.5%) Latino 243 (41.9%) White 77 (13.3%) mean (SD) Age (years) 12.8 (1.0) BMI percentile 61.7 (31.9) Dietary characteristics Fruit (freq consumed/day) 0.8 (0.6) Vegetables (freq consumed/day) 1.5 (1.2) Soda (can freq/day) 0.8 (1.1) Figure 2-‐1: Sample classroom of middle school students partitioned into unique communities using the Walktrap Community Detection Algorithm 47 Figure 2-‐2: Network plots of 17 Los Angeles middle school classrooms a a Individuals are colored by their sex: coral nodes represent girls, teal nodes represent boys, and white nodes did not have gender specified. 48 Table 2-‐2: Network summary statistics for 17 Los Angeles middle school classrooms Measure Mean (SD) Range Term definition a Network size 38.94 (7.89) 19 -‐ 52 n students in classroom Density c 0.11 (0.03) 0.07 – 0.18 n ties in network, per total number of possible ties Average degree (ties) b 3.82 (0.42) 3.02 – 4.39 n connections a person has (either outgoing or incoming) Degree centralization c 0.12 (0.04) 0.07 – 0.23 Measure of how central (how many connections, or ties) the most central node is relative to others Average path length b 2.20 (0.53) 1.13 – 2.89 n steps along the shortest path length between any 2 given people (nodes) Reciprocity c 0.92 (0.03) 0.85 – 0.95 Extent to which friend nominations are reciprocated Transitivity c 0.44 (0.10) 0.25 – 0.62 Extent to which if A and B share a tie, and B and C share a tie, that A and C also share a tie a (Valente, 2010) b Individual-‐level variable, rather than a network-‐level variable; values therefore represent mean of network means c Possible range of values: 0-‐1 Table 2-‐3: Correlations between ego intake of vegetables, fruit and soda with named friend intake of the same food item among 617 middle school students in Los Angeles With all named friends With first named friend With second named friend With third named friend With fourth named friend With fifth named friend Vegetable intake 0.01 (0.78) 0.06 (0.16) 0.06 (0.17) -‐0.01 (0.78) 0.00 (0.90) -‐0.03 (0.52) Fruit intake 0.00 (0.94) 0.03 (0.52) 0.00 (0.93) 0.05 (0.32) -‐0.03 (0.60) -‐0.07 (0.16) Soda intake 0.09 (0.02) 0.08 (0.09) 0.08 (0.09) 0.00 (0.93) 0.03 (0.57) 0.10 (0.04) 49 Table 2-‐4: Summary of interactions between social network effects and dietary intake in 17 Los Angeles middle school classrooms a Effect Whole Sample Estimated average effect size (SE) Girls Only Estimated average effect size (SE) Boys Only Estimated average effect size (SE) Sender – vegetable intake -‐0.07 (0.08) -‐0.05 (0.10) <-‐0.01 (0.02) Receiver – vegetable intake 0.02 (0.08) <-‐0.01 (0.09) <-‐0.01 (0.02) Homophily – vegetable intake -‐0.03 (0.08) -‐0.04 (0.09) <0.01 (0.03) Sender – fruit intake -‐0.05 (0.07) -‐0.07 (0.07) -‐0.03 (0.08) Receiver – fruit intake 0.01 (0.06) -‐0.02 (0.07) <0.01 (0.04) Homophily – fruit intake -‐0.02 (0.06) <0.01 (0.07) -‐0.02 (0.06) Sender – soda intake -‐0.03 (0.06) -‐0.04 (0.07) <-‐0.01 (0.05) Receiver – soda intake -‐0.03 (0.07) -‐0.05 (0.07) <0.01 (0.06) Homophily – soda intake <0.01 (0.06) 0.03 (0.07) <0.01 (0.03) a p-‐values for all effects in this table were not significant. Figure 2-‐3: Plots of communities of middle school students were significant homophily on vegetable intake was observed a a Individuals are colored by their dietary intake of an item of interest, with lighter colors indicating higher levels of intake. Panels A and B portray communities with significant homophily on vegetable intake, panel C portrays a community with significant homophily on fruit intake, and panel D portrays a community with significant homophily on soda intake. 50 Figure 2-‐4: Sample plot of vegetable intake in a community without significant homophily a a Lighter colors indicate higher levels of intake 51 CHAPTER 3: Co-‐occurring change in dietary determinants and fruit and vegetable intake in Latino elementary school youth Introduction Childhood obesity is a complex issue with many determinants, including genetic, behavioral and environmental factors at many levels. (Bray & Bouchard, 2003) It is an escalating issue in the United States and around the world,(Lobstein, Baur, Uauy, IASO International Obesity TaskForce, 2004) and it is estimated that approximately one third of US youth are overweight. (Ogden, 2012) The Centers for Disease Control identify eating behavior as one of the two major drivers of childhood obesity, with the other being lack of physical activity. (Centers for Disease Control and Prevention (CDC), 2013a) Consumption of fruit and vegetable (FV) intake is one eating behavior targeted for obesity prevention, as greater FV intake has been linked to both decreased adiposity and prevention of associated metabolic disease. (Hung et al., 2004; Ledoux et al., 2010; Montonen et al., 2005) Low consumption of FV is a problem in US youth, as only 1% meet recommendations for FV intake established in the Dietary Guidelines for Americans. (Kimmons et al., 2009) Yet, changing diet is not an easy task, given the complexity of this behavior: it is practiced multiple times throughout the day, in various settings and circumstances. Several factors have been established as determinants of FV intake, including FV availability, attitudes, knowledge, intentions, modeling, motivation, norms, parenting practices, preferences, and self-‐efficacy, among others. (McClain et al., 2009) Several approaches have been undertaken in attempt to change child dietary practices, and one of the more popular strategies in recent years has been garden-‐based education, inspired by Alice Waters’ Edible Schoolyard Project and Michelle Obama’s Let’s Move! campaign. (Let's Move!, 2015; M. C. Wang et al., 2010) Garden-‐based education has potential to improve child intake of FV by increasing exposure to these foods in a fun activity-‐based setting, where children learn about the origins of the foods they eat. (Blair, 2009; Heim et al., 2009) Several garden-‐based nutrition interventions have been developed and evaluated, and many have shown improvements in FV intake among participants. (J. N. Davis et al., 2011; A. Evans et al., 2012; Heim et al., 2009; Hermann et al., 2006; McAleese & Rankin, 2007; Somerset & Markwell, 2008; M. C. Wang et al., 2010; W. Wright & Rowell, 2010) One program, the LA Sprouts intervention, has also demonstrated to be effective in reducing body mass index (BMI) among intervention participants in both the pilot study and subsequent 52 randomized controlled trial, (J. N. Davis et al., 2011; Gatto et al., 2012; Gatto et al., under review) whereas few other studies have reported to have examined adiposity outcomes. (Castro et al., 2013) The LA Sprouts program may have been effective in reducing BMI in part because this intervention was adapted from a nutrition curriculum aimed at changing the quality of carbohydrates consumed (i.e., fewer simple and more complex carbohydrates, such as whole grains, FV), and was previously successful in improving dietary intake and reducing adiposity. (J. N. Davis et al., 2007) LA Sprouts, like many other gardening programs, also utilizes theory to identify and target several determinants of FV intake in the instructional content and activities. Some determinants that have been shown to improve as a result of garden-‐based education include preferences, self-‐efficacy, knowledge, willingness to try FV. (J. N. Davis et al., in press; Gatto et al., 2012; Gibbs et al., 2013; Lineberger & Zajicek, 2000; Morgan et al., 2010; J. L. Morris & Zidenberg-‐Cherr, 2002; Parmer et al., 2009; Ratcliffe et al., 2010; Somerset & Markwell, 2008) Although there is compelling evidence that garden-‐based nutrition education improves dietary intake, and is promising in reducing obesity, there is minimal research exploring the relationships between determinants and intake as a result of this type of intervention. This is important because some determinants of intake may be more valuable to target in this setting, compared to others. For example, in a cross-‐sectional analysis of LA Sprouts baseline data, willingness to try FV was associated with FV intake, but surprisingly, preferences were not. (Martinez et al., under review) Discerning which determinants are most important in changing intake will help focus future interventions, especially in settings where curricula need to be shortened due to time or other logistic constraints. Furthermore, while many garden-‐based interventions are theory-‐based, none to date have fully tested the theoretical assumptions underlying their design. (Spruijt-‐Metz, 2011) This meditational analysis will explore the associations between change in dietary determinants and change in both FV intake and subsequent BMI, as a result of the LA Sprouts intervention. We hypothesize that there will be evidence of mediation, and that there will be differences between the control versus intervention group (i.e. intervention group will serve as a moderator of these effects). We also perform an exploratory analysis on the strength of associations between change in different dietary determinants and change in FV intake. 53 Methods Participants and study design These analyses used data from the LA Sprouts randomized controlled trial. Detailed descriptions of the design and methodology can be found elsewhere. (Gatto et al., under review; Martinez et al., 2015) Briefly, four Los Angeles-‐area elementary schools were invited to participate in the intervention, and randomization occurred at the school-‐level, with two schools receiving the intervention and two serving as controls with a delayed intervention. These schools were selected based on the following criteria: 1) the partner after-‐school program was offered, 2) student body was ≥75% Latino, 3) ≥75% of the student population qualified to receive free/reduce price lunch, 4) school was within 10 miles of the University of Southern California (USC), 5) administrators expressed interest in having a school garden program, and 6) staff affirmed they could make the administrative commitment. All 3 rd -‐5 th grade students in the after-‐school program at these schools were allowed to participate in the intervention, but data was only collected on those without major illness and who provided child assent and parental permission. All permission and assent materials were provided in English and Spanish, and this study was approved by the USC Institutional Review Board. Intervention design A detailed description of the intervention curriculum can also be found elsewhere, (Gatto, et al., under review; Martinez et al., 2015) and materials are available online at www.lasprouts.org. The intervention consisted of 12 weeks of once-‐weekly 90-‐minute classes conducted after school in a school garden. Forty-‐five minutes of each class were dedicated to cooking and nutrition, and 45 minutes were dedicated to gardening. In the nutrition component, students learned the importance of and strategies for consuming fewer simple carbohydrates and greater amounts of complex carbohydrates, specifically FV. Each week students prepared a healthy snack, such as vegetable quesadilla with fresh salsa, and conducted gardening activities relevant to cultivating FV. The gardening curriculum also focused on sustainability and techniques for growing FV at home. Several dietary determinants were targeted in this intervention, including FV preferences, willingness to try FV, FV identification, self-‐efficacy to eat FV, and motivation to eat FV. Preferences are a well-‐established predictor of FV intake (McClain et al., 2009), and to 54 target improving preferences, children made and consumed a child-‐friendly snack containing fresh FV each week. This activity also served to target willingness to try FV, and hesitant children were encouraged by educators to try each snack. Literature on willingness to try foods suggests that preferences are a product of increased exposure, (Birch & Marlin, 1982) but as previously mentioned, baseline data from this study sample also indicate that willingness to try FV is significantly associated with intake, independent of preferences. (Martinez et al., under review) Children were also instructed on FV identification throughout the program, in the gardening, cooking, and nutrition activities. Self-‐efficacy, or an individual’s belief that he/she can perform a given behavior, (Bandura, 1977) was targeted in discussions throughout the curriculum on how to incorporate more FV into meals and snacks, and examples were provided in the cooking component. Finally, motivation to eat FV was a focus of the nutrition component, as the benefits of eating FV for oneself and with family were emphasized. Anthropometric measures Height was measured by a free-‐standing portable stadiometer (Seca, Birmingham, UK), and weight was assessed using a Tanita TBF 300A (Arlington Heights, IL). BMI z-‐score for age and sex was calculated according to CDC criteria. (Kuczmarski et al., 2002) Questionnaire measures Dietary intake was assessed using the Block Kids Food Screener (last week version). (Garcia-‐Dominic et al., 2012) This tool provides summary measures in cup equivalents (CE/day) for major food categories, including vegetables (not potatoes) and fruit (including juice). Because determinants of fruit versus vegetables may differ, (Glasson et al., 2010) these variables were not collapsed into one FV measure, but were examined separately. In order mitigate severe recall bias, total calories consumed was regressed on weight, and data from those with a standardized residual >|3| were removed from analyses (n=8 diet data removed at baseline, n=6 diet data removed at follow-‐up). Eight items were used to assess motivation to eat FV (adapted from Reasons for healthy diet, Treatment Self-‐Regulation Questionnaire), (Ryan & Connell, 1989; G. C. Williams, Grow, Freedman, Ryan, & Deci, 1996) with questions relevant to either autonomous or controlled motivation. Factor analysis of this scale did not result in separate factors for 55 autonomous and controlled motivation (one factor with eigenvalue >1 revealed from factor analysis: eigenvalue=3.39), yet this scale was divided into these two constructs based on the theoretical principles used in the development of this scale (Cronbach’s alpha autonomous motivation=0.76, n=4 items; Cronbach’s alpha controlled motivation=0.68, n=4 items). For these analyses, mean scores for constructs were used rather than latent factors with several indicators each, in order to reduce model complexity. This holds for all other dietary determinants, as well. A 14-‐item scale measured self-‐efficacy for eating FV and related behaviors (cooking and gardening), adapted from Baranowski et al. (T. Baranowski et al., 2000) Seven of these items were related to consuming FV, and were therefore used for these analyses (Cronbach’s alpha=0.84; one factor with eigenvalue >1 revealed from factor analysis: eigenvalue=3.54). To assess willingness to try FV (a lack of reluctance to eat novel foods, or food neophobia), six items adapted from Pliner and Hobden were used to measure willingness to try vegetables, and six items were used to measure willingness to try fruit. (Pliner & Hobden, 1992) Scale psychometric properties were good: for vegetable willingness to try, one factor with eigenvalue >1 revealed from factor analysis (eigenvalue=4.01), Cronbach’s alpha= 0.90; for fruit willingness to try, one factor with eigenvalue >1 revealed from factor analysis (eigenvalue=3.03), Cronbach’s alpha= 0.80. To measure FV preferences, a scale adapted from Domel et al., was used, with 8 items for fruit and 17 items for vegetables. (Domel et al., 1993) Data from this scale was separated into two separate constructs related to FV, identification and preference. For each FV item, participants were asked if they liked it “A lot”, “A little”, “Not at all”, or “I don’t know what this is”. To measure identification, the number of items for which they reported a degree of liking (i.e., they knew what the item was) was summed per fruit or vegetable category and divided by total number of items answered in that category (division by total numbered answers was done to account for missingness, range of items answered: 11-‐17 for vegetables, 6-‐8 for fruit). To measure preferences, a mean score of liking for those they identified was used. For the revised preference scale with fruit, one factor was considered acceptable (2 factors returned with eigenvalue >1, eigenvalues = 2.98 and 1.08, decision was made by examining Scree plot; Cronbach’s alpha = 0.75), and one factor was also acceptable for the revised vegetable preference scale (3 factors returned with eigenvalue >1, eigenvalues = 6.07, 1.58 and 1.01, decision was made by examining Scree plot; Cronbach’s alpha = 0.88). 56 The following demographic characteristics were also assessed via questionnaire: sex, age, ethnicity, and whether or not English was spoken at home (a proxy for socioeconomic status; significantly different between groups at baseline: 30.2% of intervention participants and 18.5% of control participants spoke no English at home, p=0.01). (Martinez et al., 2015) Analyses Paired t-‐tests and chi-‐square tests were used to compare differences between groups at baseline. A path model was used to test the following relationships: 1) change in FV intake as a mediator between change in FV determinants and change in BMI, 2) intervention group as a moderator of the aforementioned effects, and 3) whether differences exist in strength of association between change in various FV determinants and change in FV intake. Figure 3-‐1 provides a visual reference for this hypothesized model. The determinants that were included in this model were the following: autonomous motivation to eat FV, controlled motivation to eat FV, self-‐efficacy to eat FV, vegetable identification, fruit identification, vegetable preferences, fruit preferences, willingness to try vegetables, and willingness to try fruit. Covariance terms between determinant change scores (n=36 terms) and between change in fruit intake and change in vegetable intake were also included. A priori covariates in this model included sex, school, ethnicity and English spoken at home. Change in the amount of total calories (kcal) consumed was included as a covariate in a follow-‐up model. To create change scores for each variable, the difference between baseline and follow-‐ up values was regressed on the baseline value. Data appear to be missing at random; to account for missingness, the correlation matrix was analyzed rather than the raw data. The model building process was as follows: 1) the hypothesized model was fit on the entire study sample and the Lagrange Multiplier test was examined to ensure no significant and theoretically relevant associations were unaccounted for, 2) the sample was divided into two groups, intervention and control, and the models were fully constrained such that all relationships were set to be equivalent between groups, 3) constraints were released where significant differences exist between groups in covariance terms or effects of covariates, 4) constraints were released where significant differences exist in parameters relevant to research questions. Change in kcal was later added as a covariate to confirm findings. A follow-‐up model, in which associations between baseline-‐adjusted post-‐test variables were 57 evaluated, was also examined to confirm findings. All analyses will be performed with SAS PROC CALIS, version 9.3 (SAS Institute, Inc., Cary, NC), with a significance level set to α=0.05. All reported estimates are standardized. Results Three hundred and sixty-‐three participants had at least partial data and were included in these analyses (n=197 intervention participants and n=166 control participants, Table 3-‐ 1). This sample was 88% Latino, 47% male, and mean age was 9.3±0.9 years. At baseline, there were significant differences in identification of FV, with the control group reporting knowing more FV (p=0.002 for fruit and p<0.001 for vegetables). There was also a trend for the control group to have higher preferences for fruit (p=0.07). As previously mentioned, control participants were more likely to speak English at home, but there were no other significant demographic differences between groups. On average, participants consumed 1.5±1.3 CE/d of fruit, and 1.0±1.0 CE/d of vegetables at baseline. Average BMI z-‐score was 1.0±1.0 units, and 182 students (53.4%) were overweight (BMI ≥ 85 th percentile for age and sex), and 121 (35.5%) were obese (BMI ≥ 95 th percentile for age and sex). Fit statistics for all steps described in the model building process above can be found in Table 3-‐2. Model fit was good for the final model step (step 4 above; χ 2 =137.0, degrees of freedom=141, p=0.58), although was not as strong once kcal was added as a covariate (χ 2 =204.6, degrees of freedom=162, p=0.01). In both groups, the following associations were significant between change in dietary determinants and vegetable intake (Figure 3-‐1): change in controlled motivation to eat FV had a positive association with change in vegetable intake (0.21±0.6, p<0.001; i.e., a 1 standard deviation increase in controlled motivation was associated with a 0.21 CE/d increase in vegetable intake), change in willingness to try vegetables had a positive association with change in vegetable intake (0.17±0.07, p<0.01), change in willingness to try fruit had a negative association with change in vegetable intake (-‐0.17±0.06, p<0.01), and change in vegetable preferences had a positive association with change in vegetable intake (0.17±0.06, p<0.01). Only change in vegetable preference was associated with change in fruit intake (0.12±0.06, p=0.047). Change in FV intake did not mediate the relationship between 58 change in determinants of FV intake and change in BMI, as there were no significant associations between change in either F intake or V intake and change in BMI z-‐score. There were also no direct effects of change in FV determinants on change in BMI z-‐score. Results did not differ with the addition of kcal as a covariate. In the group comparison, there was only one significant difference between the intervention and control participants. The relationship between change in identification of fruit and change in vegetable intake was different between groups (p=0.01). However, this relationship was not significant in either group (-‐0.11±0.07, p=0.09 in the intervention group; 0.09±0.07, p=0.18 in the control group). Similar to above findings, results did not change with the additional of kcal as a covariate. There were minimal differences in the strength of associations between change in FV determinants and change in intake, although with change in vegetable intake, change in controlled motivation had the greatest strength of association. However, this was not statistically significantly greater than other significant effects. Findings from the follow-‐up model of associations between baseline-‐adjusted post-‐test variables can be found in Figure 3-‐3. Fit for the partially constrained (final) model was good (χ 2 =130.0, degrees of freedom=142, p=0.76). In both groups, willingness to try vegetables, vegetable preferences, and controlled motivation were positively associated with vegetable intake at follow-‐up (0.17±0.07, 0.15±0.06, and 0.21±0.06, respectively, p<0.01). Willingness to try fruit was negatively associated with vegetable intake (-‐0.13±0.06, p<0.05). There were not any significant associations between dietary determinants and fruit intake, and no significant differences between groups in the associations of interest. Discussion In this study of co-‐occurring change over an intervention period, increases in controlled motivation to eat FV, willingness to try FV, and vegetable preferences were associated with increased vegetable intake. Change in vegetable preferences was also associated with change in fruit intake over the intervention period. However, there was no significant association between change in either fruit or vegetable intake and BMI z-‐score, and there were minimal differences between intervention groups. Findings from the path model of change scores were generally supported in a cross-‐sectional path model using baseline adjusted follow-‐up data. 59 Significant relationships between change in FV determinants and change in FV intake provide evidence of relevant targets for future interventions, and these results are consistent with a large body of research supporting the importance of personal factors in driving FV intake. (Guillaumie, Godin, & Vezina-‐Im, 2010; McClain et al., 2009; Rasmussen et al., 2006) However, there are few longitudinal studies examining these effects in children, (Guillaumie et al., 2010) and this study helps elucidate which behavioral determinants are most relevant to change in dietary behaviors. Also, because determinants of intake differed for fruit versus vegetables, this study supports the practice of examining fruit and vegetables separately. In this study, change in controlled motivation to eat FV had the greatest strength of association with change in vegetable intake, compared to other determinants, although the effect size was not significantly larger than those in other significant associations. With controlled motivation, an individual’s thoughts and actions are influenced by external factors, such as the desire to please others. (Deci & Ryan, 2008) Given that the LA Sprouts intervention was delivered in a group setting, it is possible that the desire to adhere to social norms could influence vegetable intake. Drivers of this association in the control group are not clear, but it is possible that this same dynamic occurs when meals are consumed in group settings, like during school lunch. Controlled motivation in an intervention setting may not be as sustainable as other determinants, given that this is influenced by external factors which could disappear after the intervention has ended. However, given that a significant association was also seen in control participants, it is likely that there are other factors driving this relationship in these predominantly Latino elementary school children, such as a district-‐wide campaign to eat healthier, or standard health class curriculum promoting practices for healthy weight. Further research is needed to determine key drivers of controlled motivation in this group, and to identify opportunities to foster this quality in individuals. Preferences for FV are a well-‐documented predictor of FV intake, (McClain et al., 2009; Rasmussen et al., 2006), and in this study, preference for vegetables was associated with FV intake. It was surprising that change in preference for vegetables was significantly associated with change in fruit intake, rather than preference for fruit. Since fruit are generally considered more palatable than vegetables (due to higher sugar content), greater preferences for all fruit may not be as important in predicting the quantity of fruit consumed (although may influence the variety of fruit consumed). It may be possible that preference for vegetables 60 is an important enough factor that it influences both fruit and vegetable intake (although this was not found to be significant in the model with baseline-‐adjusted follow-‐up data). It was also surprising that change in other FV determinants was not associated with change in fruit intake. This may indicate that environmental factors such as availability and accessibility of fruit, rather than personal factors, are more important in driving fruit intake. (Hearn et al., 1998) In addition to preferences, willingness to try FV was predictive of intake. Change in willingness to try vegetables was associated with change in vegetable intake, and baseline data from this study also found that willingness to try FV was strongly associated with intake, yet preferences were not. (Martinez et al., under review) The importance of willingness to try foods as a predictor of intake should be further explored in intervention contexts, as it may have similar importance to preferences (willingness to try foods is not as frequently reported in the literature). However, the inverse association between willingness to try fruit and vegetable intake is surprising. The bivariate association between these two variables is not significant (p=0.16), yet willingness to try fruit is highly correlated with willingness to try vegetables (0.53±0.03, p<0.0001). Willingness to try fruit is also significantly correlated with all other determinants of FV intake, except for with the ability to identify fruit (all others p<0.05). This is likely an example of the suppression effect, whereby a variable with little correlation with an outcome variable is predictive in a multivariate model by virtue of its high correlation with at least one other variable. (Lancaster, 1999) Given that there were significant differences in BMI z-‐score between groups following the LA Sprouts intervention, (Gatto et al., under review) it was surprising that change in neither fruit nor vegetable intake was associated with change in BMI, as these were the primary behavioral targets of the intervention. It is not clear from these analyses which alternate behavioral changes led to the observed changes in BMI, but it is possible that collective changes in diet could be the cause. Whole grains, fiber, added sugar and sugar-‐ sweetened beverages were other foods and nutrients targeted in the intervention, and changes in some or all of these factors in combination with change in FV intake could have led to changes in BMI. It is also possible that a decrease in sedentary behavior could have led to changes in BMI. Also, determinants of cooking and gardening behaviors improved as a result of the intervention, (J. N. Davis et al., in press) which may have contributed to changes in BMI 61 observed. Future studies are needed to determine which specific behavioral changes are associated with decreases in obesity following garden-‐based interventions. The lack of differences between intervention and control groups was also surprising, given that significant differences were observed between groups following the intervention in determinants, vegetable intake and obesity. (J. N. Davis et al., in press; Gatto et al., under review) The intervention was thought to intensify the relationships under examination. Yet, in the main outcomes analyses, it was found that both groups decreased in many of the desired outcomes (including in vegetable intake and many determinants), with the LA Sprouts group decreasing to a lesser extent. (It is not known why both groups changed in this manner over the intervention period, but it is possible that there was a seasonal effect: follow-‐up data was collected just before winter and summer vacations, which could have had an effect on children’s attention, opinions and behaviors. For example, during holiday and end-‐of-‐year parties, children could be exposed to more junk food than under other circumstances.) Because this study measures co-‐occurring change, a positive association can be found with two variables that both increase over time, or from two variables that both decrease over time. Therefore, significant effects from these analyses could be driven by individuals who decreased both in intake and in determinants of intake, as these analyses do not discriminate between individuals with positive versus negative changes. Unfortunately, it is not known from these analyses why individuals changed, especially those in the control group. It has been demonstrated that the LA Sprouts intervention is one way that individuals can change intake and determinants of intake in the desired directions, and there are many other factors (such as the school and home food environments, parenting practices, etc.) that could have influenced behavior and determinants of behavior in control participants in positive or negative directions. There are some additional limitations of this study that may effect the interpretation and generalizability of findings. First, data were only collected at two time-‐points, which may not be sufficient to truly measure change. (J. D. Singer & Willett, 2003) Second, there is often recall bias with the use of dietary screeners, and the screener chosen may have posed some challenges for children in this age group. (Garcia-‐Dominic et al., 2012) Third, the use of co-‐ occurring change scores prevents conclusions about causality from being drawn. Fourth, determinants of other dietary behaviors discussed in the LA Sprouts intervention (i.e., whole grain and added sugar intake) were not measured, so only FV intake (the primary targeted 62 behavior) and respective determinants were examined in these analyses. This may limit our understanding of how dietary determinants effect dietary behaviors as a whole, and how intake overall effects obesity. Finally, the significant effects observed in these analyses provide more evidence on how dietary determinants are related to behavior, but findings in this sample of predominantly Latino 3 rd -‐5 th graders in Los Angeles may not generalize to other populations. These findings indicate that changes in controlled motivation, preferences, and willingness to try foods are associated with changes in FV intake. These may be relevant areas to consider when developing future interventions targeting dietary determinants of obesity in children. Controlled motivation may be addressed by conducting interventions in group or family settings, and these contexts may also work to increase willingness to try foods via peer pressure, social norms or parental modeling. Giving children the opportunity to consume snacks may also improve willingness to try foods and preferences, especially when they are introduced in ways that are appealing and fun. Presenting the same foods in a variety of formats (for example, spinach in a salad, on a pizza, or with eggs in a taco) may also help to increase both preferences and willingness to try foods. Community-‐ or school-‐based interventions like LA Sprouts, where children get to try their own fruit and vegetable snacks that they prepare together, are one format to target these mechanisms of dietary behavior change. 63 Figure 3-‐1: Conceptual model for intervention effects on change in FV intake as a mediator between change in FV determinants and change in BMI z-‐score Note: A multiple-‐group approach was used on the above model, with group 1= intervention participants, group 2= control participants. For ease of interpretation, model covariates and covariances are not included in the above diagram. Covariates included age, sex, school, ethnicity and English spoken at home. Covariances were included between change in F intake and change in V intake, and among all FV determinants change scores. 64 Table 3-‐1: Baseline characteristics of elementary school children who participated in a garden-‐based randomized control trial. (n=363) LA Sprouts participants (n=197) Control participants (n=166) p-‐value n (%) Demographic characteristics Sex (male) 93 (47.2%) 79 (47.6%) 0.94 Ethnicity (Latino) 174 (88.3%) 142 (87.7%) 0.62 English spoken at home (yes) 134 (69.8%) 128 (81.5%) 0.01 a mean (SD) Age (years) 9.3 (0.9) 9.3 (0.9) 0.96 Dietary determinants Motivation to eat FV (range 1-‐4) 3.1 (0.7) 3.1 (0.7) 0.98 Self-‐efficacy to eat FV (range 1-‐4) 3.2 (0.8) 3.3 (0.7) 0.23 Willingness to try fruit (range 1-‐4) 3.3 (0.6) 3.3 (0.7) 0.51 Willingness to try vegetables (range 1-‐4) 3.0 (0.8) 2.9 (0.9) 0.16 Fruit identification (range 0-‐1) 0.93 (0.16) 0.97 (0.09) 0.002 a Vegetable identification (range 0-‐1) 0.79 (0.18) 0.86 (0.14) <0.0001 a Fruit preferences (range 1-‐3) 2.7 (0.4) 2.8 (0.3) 0.07 Vegetable preferences (range 1-‐3) 2.2 (0.5) 2.2 (0.5) 0.86 Outcome measures Fruit intake (CE/d) 1.5 (1.3) 1.4 (1.3) 0.37 Vegetable intake (CE/d) 0.9 (0.9) 0.9 (0.9) 0.78 BMI z-‐score 0.9 (1.0) 1.1 (1.0) 0.20 a Significant difference between intervention and control group at baseline Table 3-‐2: Fit statistics at each model-‐building step Model step χ 2 , degrees of freedom (df), p-‐value Comparative Fit Index (CFI) Root Mean Square Error of Approximation (RMSEA) 1: Whole sample χ 2 =18.0, df=9, p=0.04 0.99 0.06 2: Two-‐group, fully constrained χ 2 =181.4, df=147, p=0.03 0.97 0.04 3: Two-‐group, partially constrained (covariance terms and effects of covariates released) χ 2 =143.5, df=142, p=0.45 1.00 0.01 4: Two-‐group, partially constrained (parameters of interest released) χ 2 =137.0, df=141, p=0.58 1.00 0.00 Two-‐group, partially constrained (change in kcal added as a covariate) χ 2 =204.6, df=162, p=0.01 0.97 0.04 65 Figure 3-‐2: Associations between change in FV determinants and change in FV intake among LA Sprouts study participants Note: A multiple-‐group approach was used, with group 1= intervention participants, group 2= control participants. For ease of interpretation, model covariates and covariances are not included in the above diagram. Covariates included sex, school, ethnicity and English spoken at home. Covariances were included between change in F intake and change in V intake, and among all FV determinants change scores. Gray pathways indicate non-‐significant effects. 66 Figure 3-‐3: Baseline-‐adjusted associations between FV determinants and FV intake among LA Sprouts study participants at follow-‐up Note: A multiple-‐group approach was used, with group 1= intervention participants, group 2= control participants. For ease of interpretation, model covariates and covariances are not included in the above diagram. Covariates included sex, school, ethnicity and English spoken at home. Covariances were included between F intake and V intake, and among all FV determinants. Gray pathways indicate non-‐significant effects. 67 CHAPTER 4: Effect of environmental greenness and open recreational space use on youth momentary perceived stress Introduction Social and physical environments provide a variety of stimuli attracting attention throughout the day, from sources such as cell phones, advertisements, and the behaviors of other people in our vicinity. Although many of these stimuli enrich our lives in a variety of ways, they may also wear on our attentional capacity and ability to be resilient in difficult situations. (S. Kaplan, 1995) An inadequacy of resources (either psychological, for example patience, or tangible, for example, money) to respond to these stimuli can elicit a stress response, manifesting in a physiological change in our bodies (for example, higher blood pressure and increased respiration). (S. Kaplan, 1995) Furthermore, some stimuli, such as noise, traffic and neighborhood violence may encourage stress without having any positive benefit in our lives. (G. W. Evans, 2003) In contrast to an urbanized environment, activities in the natural world may serve to reduce stress (however, it should be noted that natural environments that are perceived to be unsafe are unlikely to reduce stress). Two prominent theories attempt to explain the effect that the natural world has on individual stress levels. Kaplan’s Attention Restoration Theory posits that time spent in nature can be beneficial to decrease mental fatigue and improve concentration because the natural environment is less complex than developed ones. (R. Kaplan & Kaplan, 1989; S. Kaplan, 1995) Another prominent theory of the environmental impact on stress is Ulrich’s Stress Reduction Theory, which posits that natural environments are inherently pleasing and promote positive emotions, therefore mitigating negative emotions and diminishing stress. (Ulrich, 1986) Experimental studies also support this relationship. For example, in a study where participants were asked to walk in a nature preserve versus an urban setting, those in nature had a greater decrease in blood pressure following the walk. (Hartig et al., 2003) In this same study, after performing a stressful task, participants had more rapid decreases in blood pressure when sitting in a room with a nature view, compared to participants who recovered in a windowless room. (Hartig et al., 2003) Other similar experiments have demonstrated additional positive physiological outcomes following experiences with nature, including decreased salivary cortisol, lower heart rate, and less reported tension and anxiety. (J. Lee et al., 2011; Park et al., 2011) 68 Elucidating the relationship between the natural environment and stress is valuable given the importance of stress in affecting our mental health and quality of life. (McEwen, 2008; Ramirez, Graham, Richards, Cull, & Gregory, 1996; Reibel, Greeson, Brainard, & Rosenzweig, 2001) Our physical health is also impacted by stress, and notably, stress increases the risk for obesity, a substantial problem among children in the United States. (Ogden, 2012) The primary metabolic mechanism through which stress contributes to obesity is in activation of the hypothalamic-‐pituitary-‐adrenal (HPA) axis, which releases the hormone cortisol and increases visceral adiposity. (McEwen, 2008; Rosmond, 2003) Additionally, stress can impact obesity-‐related behaviors, namely in consumption of foods high in fat and sugar. (Dallman et al., 2003) High sugar foods are especially deleterious for obesity-‐related chronic disease, and are possibly the most damaging food type. (Lustig, Schmidt, & Brindis, 2012) Better understanding of the determinants of stress, including those at the environmental level, may be valuable in limiting child obesity and subsequent related disease. One environmental factor that may mitigate child stress is green space in the natural environment, as mentioned above. This area of research is not well explored in children, especially the impact of green spaces that are visited as part of their everyday lives. In one study, percent of neighborhood park area was inversely related to adolescent stress, (Feda et al., 2014) and another found that neighborhood greenness was negatively related to child body mass index (BMI) two years later, with stress being a possible mechanism the authors suggest for this relationship. (J. F. Bell et al., 2008) One major issue in this area of research is that stress is highly variable and may be affected by a variety of different factors; therefore, exploring the true temporal causal relationship is difficult. (Spruijt-‐Metz et al., 2014) Also, neighborhood presence of green space may not represent child utilization of that space, and there is uncertainty in the timing and duration of use of this space. (Kwan, 2012a; 2012b) A solution to both of these problems is in the use of ecological momentary assessment, or EMA. EMA allows for data to be collected on individuals in their free-‐living everyday environment, and includes periodic prompts for individuals to report their current activities and perceptions (such as perceived stress). (Stone & Shiffman, 1994) Global positioning system (GPS) data can also be collected on individuals via the same mobile devices used for EMA, and GPS-‐derived variables can provide information on the specific physical environmental conditions individuals are in when they answer EMA prompts. This study uses EMA and GPS data to examine the impact that exposure to green space presence has on momentary 69 perceived stress in children. We hypothesize that after accounting for physical activity, when children are in greener environments, they will report less stress, compared to when they are in other environments. Methods Participants and Study Design Data used in these analyses are from the Mobile Healthy PLACES study, an observational EMA study of a subset of children participating in the Healthy PLACES study. (Dunton et al., 2009b; 2011; 2012; 2014) In this study, participants either recently moved to a Smart Growth community, or lived in a neighborhood comparison community (quasi-‐ experimental design with non-‐random groups, communities were demographically equivalent). All data used in these analyses were collected at baseline of the Health PLACES study. A detailed description of study design, recruitment, participants and measures can be found elsewhere. (Almanza et al., 2012; Dunton et al., 2009b; 2011; 2012; 2014) Briefly, criteria for inclusion in this study were the following: 1) child was in grades three through eight; 2) family lived in Chino, CA (the location of the Smart Growth community) or in the surrounding community; 3) annual family income <$165,000; and 4) child ability to answer questions in English. No additional criteria were imposed for inclusion in these analyses. This study was approved by the University of Southern California Institutional Review Board, and all children and families provided assent/parental permission to participate. Demographic and Anthropometric Measures Sex, age and ethnicity were reported via questionnaire. Height was measured using a portable stadiometer (PE-‐AIM-‐101, Perspective Enterprises, Portage, MI) and weight was measured using a Tanita WB 110A (Arlington Heights, IL). Body mass index (BMI) percentile for age and sex was calculated according to Centers for Disease Control criteria. (Kuczmarski et al., 2002) Stress Measure EMA data on perceived stress were collected over a four-‐day period, from Friday at 4pm to Monday at 8:30pm (no data were collected prior to 4pm on Monday when children were in school). All EMA data were collected via a mobile phone (HTC Shadow, T-‐Mobile USA, 70 Inc.) provided to participants with custom software installed, and participants were compensated up to $40 for their participation, depending on the number of questions answered. Stress was measured with a single item: “How STRESSED were you feeling just before the beep went off?”, and children had four response options ranging from “Not at All” to “Extremely”. A total of 20 EMA prompts were given to each participant at a random time over preprogrammed intervals over the four-‐day observation period, and stress was randomly measured at 12 of the 20 times (60%). Participants were prompted to answer questions by an audible indicator, and three reminder signals were given at 5-‐minute intervals if participants failed to answer the prompt. Green Space Measures Participants were provided with a BT-‐335 portable GPS (GlobalSat Technology Corp, Taiwan) to measure momentary location, and green space measures were derived using the geographical information system ArcGIS (version 9.3, ERSI, Redlands, CA) for the period of March-‐May 2010. (Almanza et al., 2012) Two variables were used as a measure of green space: Normalized Difference Vegetation Index (NDVI, a measure of vegetation density) and Open Recreational Space (ORS), a land use type defined by the Southern California Association of Governments. NDVI has a range of -‐1 to +1, with a greater number indicating greater vegetation density. Negative values (usually representing water) were coerced to zero, with a zero NDVI value corresponding to barren earth (such as in areas of sand or rock). All data were rescaled over the 10-‐90 th percentile range to improve interpretability. The ORS variable was dichotomized to indicate being in an ORS (which included golf courses, parks, wildlife preserves, gardens and arboreta, and beaches, among others), versus not. (Southern California Association of Governments, 2005) Physical Activity Measure Physical activity (PA) was used as a covariate in these analyses, given the association between PA and stress. (Fox, 1999) PA was measured using an Actigraph GT2M accelerometer (Actigraph LLC, Pensacola, FL). (Lopes, Vasques, Maia, & Ferreira, 2007) Total number of steps in the 30-‐minute interval prior to each EMA prompt was used as an indicator of PA. 71 Analyses EMA, GPS and accelerometer data were merged within the closest 30-‐second period. The environmental context at both the time of the EMA prompt (simultaneous) and in the 30-‐ minutes prior to each prompt was of interest, and both NVDI and ORS at each of these times were used as predictors. To create the 30-‐minute NDVI variable, the mean of the NDVI measures for 30 minutes prior to an EMA prompt were used. To create the 30-‐minute ORS variable, the number of minutes a participant was in an ORS during that time period was counted, and the proportion of minutes in an ORS was used. Also, mean NVDI score over the entire observation period and total number of minutes in an ORS was obtained. Partial correlations were examined between these summary variables for NDVI and ORS over all observation days and mean reported participant stress. Stress was treated as a continuous variable throughout these analyses. Multi-‐level models were used to examine effect of green space variables (simultaneous and 30-‐minute prior NDVI and ORS) as predictors of stress. (J. D. Singer & Willett, 2003) The simultaneous NDVI variable was partitioned into two variables, one to represent the between-‐person simultaneous NDVI (i.e., the mean NVDI score for each individual at the times when stress was measured), and the other to represent the within-‐person variance in NDVI (i.e., at any given timepoint, the difference between the individual mean and the value at that instance). (Hedeker, Mermelstein, & Demirtas, 2012) This was also be done for the 30-‐minute NDVI variable, 30-‐minute ORS variable, and PA. For momentary ORS (binary), the between-‐ person variance was set to the proportion of occasions in an ORS when stress was measured, and the within-‐person variance remained binary. (Hedeker, Mermelstein, Berbaum, & Campbell, 2009) The aforementioned between-‐subject variables, in addition to other continuous covariates, were centered around means for the entire sample (grand mean centered). Predictors were examined in univariate analyses, and outliers were identified using box-‐and-‐whisker plots and removed (data with values >( 3 rd quartile + 1.5*(interquartile range)) and <(1 st quartile – 1.5*(interquartile range)) were removed). Given that stress was a discrete variable, there were not any outliers removed. The longitudinal model has two levels: Level 1 describes within-‐person differences and Level 2 describes between-‐person differences. (J. D. Singer & Willett, 2003) The Level 1 model is specified as follows: Yij = π0i + π1iZij + εij 72 where Yij is reported stress for individual i at time j, π0i is the true initial status for individual i, π1i is the slope representing the effect of time-‐varying predictor terms Zij, and εij is random measurement error. The Level 2 model is then specified as follows: π0i = γ00 + γ0kZi + ζ0i π1i = γ10 + γ1kZi + ζ1i where γ00 and γ10 are the population average initial status and rate of change, respectively; γ0k and γ0k are slopes representing the effect of predictor terms Zi; and ζ0i and ζ1i are the residuals (individual deviations). Goodness of fit was established by checking normality and homoscedasticity of error terms. (J. D. Singer & Willett, 2003) A priori individual covariates included age, sex, ethnicity (dummy coded with multiple levels), annual household income (a continuous variable), neighborhood group (Smart Growth Community versus control), BMI percentile, and PA. Between-‐person covariates were entered in the Level 2 model (within-‐person PA was included in the Level 1 model). Temporal covariates included time of day, day of week (categorical, dummy coded with reference=Friday), day of study (ordinal), and prompt number (entered in Level 1 of the model). For model parsimony, backward selection was employed to remove non-‐contributing covariates. Covariates were left in the model when removal of the covariate of interest resulted in a >20% change in a coefficient of interest, or when statistically significant. Follow-‐up models examined stress as a binary outcome (no stress versus at least some level of stress) and green space predictors as categorical variables. Categories were determined by examining univariate distributions and observations were assigned to one of three levels. For momentary and 30-‐minute NVDI variables, the following categories were used: NDVI=0, 0<NVDI≤0.25, NDVI<0.25 (maximum NDVI value for both 30 minute and momentary NDVI was 0.52). For 30-‐minute ORS use, the following categories were used: %ORS=0, 0<%ORS<1, %ORS=1. (Momentary ORS use was not categorized given it is a binary variable.) Given the longitudinal nature of this study, missingness is a concern in these analyses, as this can impact parameter estimates and interpretations. (J. D. Singer & Willett, 2003) In this study, data were presumed to be missing at random (MAR), such that missingness may depend on any of observed data, but does not depend on any unobserved variables. (J. D. 73 Singer & Willett, 2003) According to Graham, once variables known to predict missingness are included in the model, then the missing completely at random (MCAR) criteria is essentially met. (Graham, 2012) To test this, a binary variable for missingness of an EMA stress measure was created, and all of the above aforementioned covariates were examined as possible predictors of missingness in separate multilevel logistic regression models. Prior missingness and measures of positive and negative affect (measured via EMA) were also considered as possible predictors of missing an EMA prompt, based on theoretical considerations. Those that were significant predictors of missingness were included as covariates in main outcomes models, and were not removed during backward selection. Analyses were conducted using SAS PROC MIXED version 9.3 (SAS Institute, Inc., Cary, NC), using restricted maximum likelihood as the method of estimation, and an unstructured covariance matrix. In tests for predictors of missingness, SAS PROC GLIMMIX with a logit link function and binomial distribution was used. Results One hundred and nine children (51% male, age 10.9±1.2 years, Table 4-‐1) had at least partial data, and there were 979 valid observations with a stress measure (range 2-‐12 observations with corresponding GPS data within participant; mean number of observations per participant 8.9±3.0). Latinos comprised the largest ethnic group at 31.4% of participants. Among participating children, 39.6% were overweight (BMI ≥ 85 th percentile for age and sex), and 20.8% were obese (BMI ≥ 95 th percentile). Reported stress was generally low (mean 0.4±0.6), and participants spent 10.8% of their time during the observation period in ORS (9.7% of time in 30 minutes prior to prompts was in ORS). Mean NDVI over the entire observation period was 0.18±0.12, and mean NDVI in 30 minutes prior to prompts was 0.12±0.19. Tests for predictors of missingness revealed that two variables were significant predictors of the likelihood of failing to respond to any given EMA prompt: the total number of prior EMA prompts missed was positively related to missingness, (between-‐subject prior missingness; p<0.0001), such that those who missed more prior prompts on average had a greater likelihood of failing to respond to a prompt; and the mean score for feeling energetic was positively related to missingness (an affect measure; between-‐subject energetic; p=0.01), 74 such that those who reported feeling more energetic overall had a greater likelihood of failing to respond to a prompt. Both of these covariates were included in all final models. From partial correlations, there were no significant associations between mean reported stress and either overall ORS use or NDVI (r = 0.21, p=0.09; r = 0.03, p=0.81; respectively). There were no associations between EMA-‐reported stress and either momentary NDVI or ORS use (for momentary NDVI, between-‐person: -‐0.05±0.40, p=0.90, within-‐person: -‐0.19±0.17, p=0.26; for momentary ORS, between person: 0.35±0.28, p=0.21, within-‐person: -‐0.09±0.06, p=0.26; Tables 4-‐2 and 4-‐3, respectively). Covariates for the test of momentary NDVI included the aforementioned from the test for predictors of missingness, ethnicity, physical activity, and day. Covariates for the test of momentary ORS included those from the test for predictors of missingness, physical activity, day, and prompt number. There were also no associations between EMA-‐reported stress and either 30-‐minute NDVI or ORS use (for 30-‐minute NDVI, between-‐person: -‐0.22±0.33, p=0.51, within-‐person: 0.07±0.18, p=0.69; for 30-‐minute ORS, between person: 0.18±0.18, p=0.32, within-‐person: -‐ 0.02±0.13, p=0.87; Tables 4-‐4 and 4-‐5, respectively). Covariates for the test of 30-‐minute NDVI included the aforementioned from the test for predictors of missingness and day. Covariates for the test of 30-‐minute ORS included those from the test for predictors of missingness, day, and prompt number. There were no significant effects when stress was examined as a binary variable, or when green space variables were categorized (data not shown). Fit was acceptable for all final models. Discussion This study of 109 youth in Southern California examined the associations between reported stress and green space use, measured via NDVI and ORS use. This study is one of the few studies to examine whether time in nature has a therapeutic effect on stress in children. The following relationships were examined: overall green space use and mean stress over the entire 4-‐day observation period, stress and green space use at the time of self-‐report, and stress and green space use in the 30 minutes prior to self-‐report. This study found that there was no association between green space use and stress in youth. As mentioned, there is little prior research examining the effect of green space on stress in youth. To our knowledge, a study by Feda et al. is the only other study to examine 75 neighborhood green space and stress in children. This study examined percent neighborhood park area and stress via the 14-‐item perceived stress scale in 68 adolescents, and found there to be an inverse association between neighborhood park area and stress (β=-‐62.6, p<0.03). (Feda et al., 2014) The present study differs from this study in that it measures utilization of green space by children, whereas the study by Feda et al. only measures the presence of neighborhood green space. The lack of significant effects in this study is surprising given the theoretical support for this relationship. (Bratman et al., 2012; R. Kaplan & Kaplan, 1989; Ulrich, 1986) Also, there is data from experimental studies demonstrating that nature can be restorative after performing a stressful task, yet these data were collected in adults. (Hartig et al., 2003) A limitation in this study is that stress was measured at random times over regular intervals, and not necessarily when individuals were experiencing high stress. Participants said they felt ‘extremely’ stressed in only 3.1% of observations, and they reported feeling ‘quite a bit’ stressed in only 5.6% of observations. If more data were available during times of high stress, it may be possible to further examine the environmental influences on stress and coping. Another limitation of this study was that participants spent minimal time in green areas. Participants were in areas with zero greenness at the time of self-‐report 69.4% of the time, and in areas with zero greenness in the 30 minutes prior in 68.9% of the time. Yet, in Southern California, where these data were collected, there are many popular natural areas with low vegetation density, such desert and coastal areas. Mountains and foothills in the area, where the primary vegetation is brush and chaparral due to the semi-‐arid climate, would also be expected to have a low amount of greenness. Furthermore, given chronic drought conditions in Southern California, many parks may have implemented a reduced-‐ watering schedule, even when these data were collected in 2010. However, participants were in open recreational spaces at the time of self-‐report only 7.0% of the time, and participants spent some time in an ORS in the 30 minutes prior to self-‐report only 24.3% of times. Also, it is not possible to tell from these data whether participants were indoors or outdoors, which would likely affect the relationship between area vegetation and stress. (However, it has been shown that even looking out a window onto a natural landscape can be beneficial for stress reduction). (Hartig et al., 2003) Participants were asked during select prompts whether or not they were outdoors, but there were only 62 instances when both this question and the question on stress were asked. Just as it would be helpful to have more data available during 76 times when participants experienced moderate-‐to-‐high stress, it may have been possible to learn more about the relationship between nature and stress had there been more data available during times when participants were in natural areas with higher vegetation density. Mean NDVI over the entire monitoring period was 0.18, whereas it was only 0.12 on average during the 30 minutes prior to self-‐report, indicating that not all experiences in (relatively) higher vegetation density were captured. This research question can be better-‐ addressed with a study design that collects environmental information from individuals during times of high stress, or one that collects stress data upon entering or leaving a natural area. Another consideration when interpreting these findings is the lack of information on the social environment. In this study, there was no significant difference in reported stress when children reported being with parents, siblings, other family, friends, classmates or strangers (data not shown). However, interpersonal dynamics are very complex, and it is possible for individuals to feel high stress or great happiness and serenity with the same individual at different times. Youth may be particularly susceptible to stress from their social environment, (Moksnes et al., 2010b) for example from challenging family dynamics, pressure from peers, or the presence of strangers. Also, some natural areas may not be as calming as others. For example, parks next to a noisy freeway, or with homelessness or gang activity, are not as likely to be calming or restorative. There are some other limitations in this study. One limitation was the measure of reported stress, which was limited to a single item with four response options. Also, the bidirectional relationship between stress and green space use was not examined in this study, and given the measurement schedule of the stress variable, it is not known whether or not participants ever sought out green space to cope with stress. Furthermore, data were collected over a 4-‐day period, which may not be fully representative of children’s lives. However, this study was the first, to our knowledge, to measure stress and geographic environment in real time in children. Also, physical activity is well-‐controlled through the use of accelerometers. Although there was no significant association found in this study between time in green space and self-‐reported stress in children, this study design was not highly appropriate to use to measure this relationship. The use of real-‐time data was a major strength, but the small amount of data captured during pivotal instances make it difficult to draw conclusions 77 from these findings. More observational studies are needed to examine the effect of green space on stress, especially in children, but more deliberate action must be taken to ensure that data is collected during critical times. One possible approach could be to ask participants to report on their environment when they feel stressed, but they are likely to be distracted during these times and may not be reliable in performing self-‐report. It is also possible for individuals to be distracted in a different manner when they visit natural areas, such as when they are fully present in taking in a beautiful setting. However, as data collection tools improve through the use of mobile-‐monitoring prompt triggers (such as GPS alerts when entering or leaving a park, or an increase in heart rate as proxy for higher stress), bias from the need to initiate self-‐report can be reduced. Studies that employ these new technologies to examine the effect of green space on stress are needed in order to inform recommendations for child stress and obesity prevention. Table 4-‐1: Baseline characteristics of 109 3 rd -‐8 th grade children in Chino, CA Demographic characteristics n (%) Sex (male) 54 (50.9%) Ethnicity: Black 10 (9.5%) Asian 15 (14.3%) Latino 33 (31.4%) White 23 (21.9%) Mixed 17 (16.2%) Other 7 (6.7%) mean (SD) Age (years) 10.9 (1.2) BMI percentile 64.4 (30.8) Annual household income (dollars) 78,507 (39,829) Individual modifiable characteristics Perceived stress (range 0-‐3) 0.4 (0.6) Physical activity (steps/30 min prior to prompt) 335.7 (510.4) Environmental characteristics (over entire observation period) Normalized Difference Vegetation Index (range 0-‐1) 0.18 (0.12) Percent of time over entire observation period Open Recreational Space use 10.8 (20.2) 78 Table 4-‐2: Momentary normalized difference vegetation index (NDVI) as a predictor of stress in 3 rd -‐8 th grade children in Chino, CA (n=964 observations) Parameter Coefficient (SE) p-‐value Intercept 0.72 (0.16) <0.0001 Ethnicity (Black) -‐0.53 (0.21) 0.01 Ethnicity (Asian) -‐0.11 (0.19) 0.56 a Ethnicity (Latino) -‐0.36 (0.18) 0.04 Ethnicity (White) -‐0.37 (0.18) 0.04 Ethnicity (Mixed) -‐0.34 (0.18) 0.06 Between-‐subject energetic -‐0.02 (0.07) 0.72 Between-‐subject prior missingness -‐0.47 (0.29) 0.11 Between-‐subject physical activity -‐0.08 (0.04) 0.03 Day (Friday) -‐0.04 (0.06) 0.52 a Day (Monday) 0.16 (0.07) 0.02 Day (Saturday) -‐0.06 (0.05 0.25 a Between-‐subject momentary NDVI -‐0.05 (0.40) 0.90 Within-‐subject momentary NDVI 0.19 (0.17) 0.26 a F-‐test p-‐value < 0.05 Table 4-‐3: Momentary open recreational space (ORS) use as a predictor of stress in 3 rd -‐8 th grade children in Chino, CA (n=573 observations) Parameter Coefficient (SE) p-‐value Intercept 0.79 (0.21) <0.001 Between-‐subject energetic 0.02 (0.08) 0.77 Between-‐subject prior missingness -‐0.51 (0.35) 0.15 Between-‐subject physical activity -‐0.08 (0.04) 0.05 Day (Friday) -‐0.42 (0.19) 0.02 Day (Monday) 0.39 (0.11) <0.001 Day (Saturday) -‐0.28 (0.12) 0.02 Prompt number -‐0.03 (0.01) 0.04 Between-‐subject momentary ORS use 0.35 (0.28) 0.21 Within-‐subject momentary ORS use -‐0.09 (0.06) 0.26 79 Table 4-‐4: Thirty-‐minute normalized difference vegetation index (NDVI) as a predictor of stress in 3 rd -‐8 th grade children in Chino, CA (n=969 observations) Parameter Coefficient (SE) p-‐value Intercept 0.66 (0.16) <0.0001 Between-‐subject energetic -‐0.001 (0.07) 0.98 Between-‐subject prior missingness -‐0.36 (0.30) 0.22 Day (Friday) -‐0.25 (0.14) 0.08 a Day (Monday) 0.25 (0.09) <0.01 Day (Saturday) -‐0.19 (0.09) 0.04 Between-‐subject 30-‐minute NDVI -‐0.22 (0.33) 0.51 Within-‐subject 30-‐minute NDVI 0.07 (0.18) 0.69 a F-‐test p-‐value < 0.05 Table 4-‐5: Thirty-‐minute open recreational space (ORS) use as a predictor of stress in 3 rd -‐8 th grade children in Chino, CA (n=706 observations) Parameter Coefficient (SE) p-‐value Intercept 0.80 (0.19) <0.0001 Between-‐subject energetic -‐0.02 (0.07) 0.75 Between-‐subject prior missingness -‐0.41 (0.32) 0.21 Day (Friday) -‐0.39 (0.17) 0.02 Day (Monday) 0.35 (0.10) <0.001 Day (Saturday) -‐0.27 (0.11) 0.01 Prompt number -‐0.03 (0.01) 0.03 Between-‐subject 30-‐minute ORS use 0.18 (0.18) 0.32 Within-‐subject 30-‐minute ORS use -‐0.02 (0.13) 0.87 80 CHAPTER 5: Discussion Summary of Findings Childhood obesity is a major public health issue in the United States (US), and of the estimated 74.5 million children living in the US, approximately 24 million are estimated to be overweight or obese. (childstats.gov, 2013; Ogden, 2012) Understanding opportunities for prevention and treatment is imperative in order to reduce obesity and associated comorbidities in children. Although there are many non-‐modifiable characteristics contributing to childhood obesity, such as genetics, individual behaviors (including diet and physical activity) and other modifiable individual-‐level characteristics (such as stress) play a major role in obesity development. (Centers for Disease Control and Prevention (CDC), 2013a) The goal of this body of work was to understand how differing environmental levels, the social, structured learning, and physical environments, affect obesity-‐related behaviors in children. The first objective was to examine patterns of similarity among adolescent friends in their dietary intake, specifically fruit, vegetables (FV) and soda. These three food types are important as they are associated with risk for obesity, and are consumed in amounts highly disproportional to their nutritional value. (He et al., 2004; Kimmons et al., 2009; Malik et al., 2006; Nielsen & Popkin, 2004) The second objective was to examine the impact of a school garden-‐based nutritional education program on FV intake as a mediator between FV determinants and BMI. The third objective was to examine the impact of green space use on child perceived stress in their everyday living environments. Results from Study 1 indicate that there were not any overall similarities among middle school friends in their intake of fruit, vegetables or soda. It was also found that there were not any significant sender or receiver effects, such that the dietary behaviors, as measured in this study, did not affect the number of friends nominated or the number of times a person was nominated. Since it may not be plausible to expect that all friendships share similar dietary patterns, smaller communities within each classroom were identified, and similarities among friends in dietary behaviors were examined in these groups. Surprisingly, there were very few communities with significant homophily on dietary intake among members. Only two communities had significant homphily on vegetable intake, and one community each had significant homophily on fruit and soda intake. However, homophily as measured by the absolute difference between those who share direct connections, as was 81 done in this study, is likely not an appropriate way to examine relationships within communities. Presumably, all community members interact in the small social group, and second-‐ and third-‐degree connections may be highly relevant, especially in a classroom setting where community members share common spaces and weekday activities (unlike many communities in adult networks). Study 2 found that the structured learning environment of a garden-‐based intervention for obesity prevention did not significantly effect instances of co-‐occurring change in determinants of FV intake and change in FV intake or BMI in elementary school children. The addition of gardening, cooking and nutrition information to an education program has previously been shown to have a positive effect on dietary determinants, intake and measures of obesity in this group, (J. N. Davis et al., in press; Gatto et al., under review) yet this study examined which changes between pre-‐ and post-‐test occurred simultaneously. This study found that students who changed their controlled motivation to eat FV, willingness to try FV, and vegetable preferences had a positively associated change in their vegetable intake. Also, change in vegetables preferences was positively associated with change in fruit intake. This suggests that a structured learning environment does not alter the strength of association between simultaneous change in dietary determinants and intake, but can be one mechanism through which behaviors and determinants are altered. Data from Study 3 did not show an association between stress and green space as measured by Normalized Difference Vegetation Index (NDVI) or Open Recreational Space (ORS) use among children. The use of a semi-‐random prompt schedule was not most appropriate to address this research question, as children often reported little or not stress, and were infrequently in open spaces or areas with high vegetation density. A study design that allows for the collection of data during these critical times is needed in order to have a more robust understanding of this relationship. New methodologies that include on-‐body stress measuring devices and GPS tracking to trigger Ecological Momentary Assessment (EMA) prompts should be used to fully capture these instances in a real-‐world environment. These studies contribute to the literature in various ways. Study 1 is the first, to our knowledge, to measure dietary intake of FV and soda with complete social networks. Although these findings we null, it is also the first to consider whether or not communities within networks may have varying levels of homophily on dietary intake. Study 2 aids in identifying determinants of intake that have co-‐occurring change with vegetable intake. Given 82 limited longitudinal research on the impact of change in determinants on dietary behavior, (Guillaumie et al., 2010) these findings will help identify areas of focus for future interventions. Study 3 is the first to our knowledge to examine both ORS use and NDVI in relation to stress over time in children. The lack of significant findings provides a rationale for the use of more advanced data collection strategies to measure this association. Interaction of Environmental Levels Social Environment The social environment played an important role in all three studies, even though it was only examined in Study 1. For Study 2, the intervention was delivered in a classroom setting, and children were encouraged to eat snacks together and discuss their thoughts about the food. (Topics of conversation that were encouraged included ‘What did you like or not like about this snack?’, and ‘What other fruits or vegetables could you include in this snack?’.) Results from this study indicate that change in controlled motivation to eat FV had the largest strength of association in predicting change in vegetable intake. With controlled motivation, thoughts and actions are driven by factors outside of the individual, such as a desire to please others or to adhere to social norms. (Deci & Ryan, 2008) Since the LA Sprouts intervention was delivered in a group setting, this could reflect a desire for individuals to have similar dietary intake as their classmates, or to adhere to a specific dietary pattern because it was encouraged by the instructor. This significant association was also found in control participants, which suggest that desire to adhere to an instructor’s recommendations is not the sole driver of this relationship, but peers, such as those one eats lunch with, may also be important in influencing intake through mechanisms described in chapter 1 of this dissertation. In Study 3, it is likely that specific social context can influence the relationship between green space and stress in youth. For example, presence of strangers in a park, especially if there is homelessness or gang activity, could moderate the beneficial effects of open space. On the other hand, being with friends or loved ones in a natural setting could be especially relaxing for some people. Or, being alone in a natural setting could provide an even better opportunity to recover from stress and restore attentional reserves for some people, although this may be more relevant for adults, compared to children. 83 Structured Learning Environment The structured learning environment plays an especially important role in the lives of children, as it provides opportunities to not only teach them information and skills needed for a successful and healthy life, but is also the setting for the majority of opportunities for socializing with peers. In Study 1, social networks were based on friendships within classrooms, and generally students had many friendships with other students in their classroom (see Figure 2-‐2). Only 8 students out of the 617 middle school students in this sample did not report having any friends in their classroom, nor were they reported as being a friend of someone in their classroom. It is not clear to what extent obesity-‐related behaviors are influenced by classroom friends, relative to other friends outside one’s class, or family members. Parental modeling is a well-‐established determinant of dietary intake, (McClain et al., 2009) but may also be an easier relationship to measure, relative to complex social dynamics among youth friends. In addition to providing opportunities for friendship, schools constitute an important part of the physical environments in which children spend their time. Children eat 1/3 of all weekday meals at schools, and many children eat both breakfast and lunch at school. In the LA Sprouts study, 87% of all participants were eligible for free lunch at their school, (Gatto et al., under review) and school lunch can has substantial impact on dietary intake. (Gordon et al., 1995) With respect to Study 3, schools can provide opportunities for children to use open green space if this is available on campus. However, many urban schools in cities like Los Angeles do not have substantial playground and recreational areas, and also lack the budget and manpower to maintain safe, natural spaces. Installation of teaching gardens on blacktops, as was done in the LA Sprouts study, could provide a restorative and calming environment for schoolchildren, but findings from Study 3 suggest that greater density of vegetation does not reduce stress in children. It is therefore unclear what effects school gardens could have on stress in youth. Physical Environment It is not expected that the physical environment play a large role in the relationship between adolescent friendships and dietary behaviors, except in instances where friends eat together in specific locations. These may include settings where menu options are limited, such as during school lunches, as previously mentioned, or in fast food restaurants. Research 84 has shown that youth consume a similar number of meals at fast food restaurants as their friends. (Ali et al., 2011) Further research on the interaction between food sources and friendship groups on dietary intake in youth would make a valuable contribution towards our understanding of environmental influences on dietary behavior. This research topic is especially relevant to teenagers, as they are more autonomous in their eating behaviors than younger children. It is also unclear to what extent the physical environment affects structured learning as it relates to health behaviors. There has been research conducted on ways to optimize classroom organization, (Slavin, 1989) which could help facilitate the delivery of obesity-‐ related information. In the LA Sprouts study, it is unlikely that substantial physical changes were made to neighborhood, school or home environments over the 12-‐week study period (with the exception of the school garden), so it is unlikely that the physical environment could be a primary driver of changes seen in dietary determinants or intake in control students. Future Research For all three of these studies, more questions were raised than were answered. From the follow-‐up analyses in Study 1, there is some evidence of similar dietary behaviors within small communities of friends within a classroom network, but it is not clear how frequently this occurs. As mentioned, the technique utilized (absolute difference over direct ties) is limited in that it does not consider 2 nd and 3 rd degree connections, or the small community as a whole, where it can be assumed that most members are friends with varying degrees of closeness. One possible way to approach this research question would be to artificially inflate the density of each community, and to allow for individuals to have more than 5 outgoing ties (the limit imposed by the study design). Absolute difference could then be used, as it was in this study. Or, some measure of standard deviation from a community mean could possibly be informative. Once small communities with homophily on dietary behaviors are better identified, the prevalence can be estimated, and communities can be further examined in order to determine under which conditions homophily is more likely to occur. In the LA Sprouts study, it was found that the intervention resulted in improvements in child BMI compared to controls, but it is not clear what contributed to these findings, as there was no change in both BMI and FV intake or determinants of intake. It is possible that small changes were made in the consumption of many dietary items (for example, FV consumption 85 could increase by a very small amount, whole grains consumption could increase by a small amount, added sugar could decrease by a small amount, etc). Cooking and gardening behaviors were not explored in these analyses, and these could be responsible for the change observed in BMI, but it would not be through change in FV intake alone. Or, perhaps sedentary behavior decreased in the LA Sprouts participants through the gardening activities. A data mining approach could be used to determine how changes in individual dietary factors related to change in BMI. It could also be examined how and if this differed among children (for example, one child may have increased FV consumption, whereas another may have stopped consuming sodas, etc). If no relationship can be found with change in diet and change in BMI, it can be presumed that physical or sedentary activity played a major role in changes observed in BMI (unfortunately, activity was not measured in this study). In Study 3, a limitation is that there were few instances when participants experienced high stress, and even at these points, a single question with four possible response options was used to measure stress. A measure such as this limits ability to measure a stress response over time, let alone the dynamic relationship between environmental factors and recovery from stress. As mentioned previously, a more robust stress-‐monitoring tool is needed to identify when data on stress ought to be collected, and more detailed environmental data would also help elucidate the association between nature and stress. A theme for future research in all of these studies is heterogeneity. Individuals cannot be expected to have the same behavioral response to all environmental stimuli, and better understanding of in whom and under which conditions certain behaviors occur will help us to better understand health practices and to make more informed recommendations. In addition, the interaction between different environmental levels should be further explored, as mentioned in the previous section. This is especially valuable with interaction of the social and physical environments, as different environmental contexts may inspire various behavioral reactions in children. Limitations In addition to limitations specific to each study, as mentioned in individual study chapters, there are some overall limitations of this body of work. The first is that this dissertation offers a broad overview of environmental influences on obesity-‐related behaviors, but does not dive deep into the exploration of specific relationships. Three 86 environmental levels and three obesity-‐related behaviors were identified as being relevant to childhood obesity (See Figure 1-‐1), yet each environmental level is only examined once, and each with only one specific behavior (also, interactions of environmental levels are not measured, as discussed above). Physical activity is not examined in these studies, yet plays an important role in energy balance. (Centers for Disease Control and Prevention (CDC), 2013a) Furthermore, BMI is only examined in one study, although it is well-‐established that diet, stress, and physical activity all contribute to this overweight and obesity. (Centers for Disease Control and Prevention, 2013a; Spruijt-‐Metz et al., 2014) Although environmental measures in each study were generally robust and consistent with best practices, behavioral measures were not as strong. In Study 1, vegetable intake was measured with three items, yet fruit and soda intake were measured with one frequency item each. In Study 2, the Block Kids Food Screener was used because this provides an estimate of total calories consumed (although the kcal variable is not validated). (Garcia-‐Dominic et al., 2012) This screener is challenging for elementary school children and may be one reason why there was no significant association found between change in FV intake and change in BMI. In Study 3, stress was measured with a single item, although it was a momentary measure, which is a major strength. However, it was measured at quasi-‐random times and likely did not capture the environmental contexts of all experiences of stress in participants. Having strong behavioral measures is important because measures lacking robustness could bias findings towards null and limit exploration and understanding of complex relationships. Recommendations Based on findings from this body of work, the following recommendations can be made: 1) More research needs to be done on the relationship between adolescent friendships and dietary intake, and new techniques must be developed in order to measure the heterogeneity of this relationship across small communities of friends. 2) The following dietary determinants are valuable to address in interventions targeting vegetable intake: controlled motivation to eat FV, willingness to try FV, and vegetable preferences. Vegetable preferences may also be important to address in interventions targeting fruit intake, although additional research should confirm these findings. 87 3) Fruit and vegetable intake should not be considered the same dietary construct, and when examining determinants of intake, fruit and vegetables ought to be examined as two separate food types. 4) School-‐based obesity interventions should consider the role of peer groups when implementing curricula. 5) More robust measurement tools should be used to fully capture the dynamic association between green space and stress in children. 6) Research is needed on the interaction between the physical and social environment, and how this affects dietary intake and stress in youth. Conclusion This body of work provides additional insight into ways in which environmental factors affect obesity and related behaviors. Improving obesity-‐related behaviors is especially important in children, given the staggering amount of children who are overweight and obese, and the greater disease risks these children are susceptible to. (Bray & Bouchard, 2003; Ogden, 2012) Given the evidence from the studies conducted and the literature cited, it is clear that the environment, which includes social, structured learning, and physical levels, impacts diet, physical activity and stress. The general public has strongly debated the importance of ‘personal choice’ as the primary driver of obesity, (Brownell et al., 2010) yet the evidence here shows that individuals make choices based on circumstances and situations around them. More needs to be done to fully understand how the environment shapes our health-‐related activities, and actions must be taken to empower individuals in making better choices and to facilitate long-‐term health in children. 88 REFERENCES Ali, M. M., Amialchuk, A., & Heiland, F. W. (2011). Weight-‐related behavior among adolescents: the role of peer effects. PloS One, 6(6), e21179. http://doi.org/10.1371/journal.pone.0021179 Almanza, E., Jerrett, M., Dunton, G., Seto, E., & Pentz, M. A. (2012). A study of community design, greenness, and physical activity in children using satellite, GPS and accelerometer data. Health & Place, 18(1), 46–54. http://doi.org/10.1016/j.healthplace.2011.09.003 Anda, D. de, Baroni, S., Boskin, L., Buchwald, L., Morgan, J., Ow, J., et al. (2000). Stress, stressors and coping among high school students. Children and Youth Services Review, 22(6), 441– 463. Anzman-‐Frasca, S., Savage, J. S., Marini, M. E., Fisher, J. O., & Birch, L. L. (2012). Repeated exposure and associative conditioning promote preschool children's liking of vegetables. Appetite, 58(2), 543–553. http://doi.org/10.1016/j.appet.2011.11.012 Armitage, C. J., & Conner, M. (2000). Social cognition models and health behaviour: A structured review. Psychology & Health, 15(2), 173–189. http://doi.org/10.1080/08870440008400299 Arnett, J. J. (2006). Suffering, Selfish, Slackers? Myths and Reality About Emerging Adults. Journal of Youth and Adolescence, 36(1), 23–29. http://doi.org/10.1007/s10964-‐006-‐ 9157-‐z Badaly, D. (2013). Clinical Psychology Review. Clinical Psychology Review, 33(8), 1218–1236. http://doi.org/10.1016/j.cpr.2013.09.003 Bamia, C., Trichopoulou, A., Lenas, D., & Trichopoulos, D. (2004). Tobacco smoking in relation to body fat mass and distribution in a general population sample. International Journal of Obesity, 28(8), 1091–1096. http://doi.org/10.1038/sj.ijo.0802697 Bandura, A. (1977). Self-‐efficacy: toward a unifying theory of behavioral change. Psychological Review, 84(2), 191–215. Baranowski, T., Davis, M., Resnicow, K., Baranowski, J., Doyle, C., Lin, L. S., et al. (2000). Gimme 5 fruit, juice, and vegetables for fun and health: outcome evaluation. Health Education & Behavior, 27(1), 96–111. Barton, J., & Pretty, J. (2010). What is the Best Dose of Nature and Green Exercise for Improving Mental Health? A Multi-‐Study Analysis. Environmental Science & Technology, 44(10), 3947–3955. http://doi.org/10.1021/es903183r Basciano, H., Federico, L., & Adeli, K. (2005). Fructose, insulin resistance, and metabolic dyslipidemia. Nutrition & Metabolism, 2(1), 5. http://doi.org/10.1186/1743-‐7075-‐2-‐5 Bell, C. G., Walley, A. J., & Froguel, P. (2005). The genetics of human obesity. Nature Reviews Genetics, 6(3), 221–234. http://doi.org/10.1038/nrg1556 Bell, J. F., Wilson, J. S., & Liu, G. C. (2008). Neighborhood Greenness and 2-‐Year Changes in Body Mass Index of Children and Youth. Am J Prev Med, 35(6), 547–553. http://doi.org/10.1016/j.amepre.2008.07.006 Berkey, C. S., Rockett, H. R. H., Field, A. E., Gillman, M. W., & Colditz, G. A. (2004). Sugar-‐added beverages and adolescent weight change. Obesity Research, 12(5), 778–788. http://doi.org/10.1038/oby.2004.94 Berto, R. (2005). Exposure to restorative environments helps restore attentional capacity. Journal of Environmental Psychology, 25(3), 249–259. http://doi.org/10.1016/j.jenvp.2005.07.001 Bevelander, K. E., Anschütz, D. J., & Engels, R. C. M. E. (2012). Social norms in food intake 89 among normal weight and overweight x. Appetite, 58(3), 864–872. http://doi.org/10.1016/j.appet.2012.02.003 Birch, L. L., & Marlin, D. W. (1982). I don“t like it; I never tried it: Effects of exposure on two-‐ year-‐old children”s food preferences. Appetite, 3(4), 353–360. http://doi.org/10.1016/S0195-‐6663(82)80053-‐6 Birch, L. L., McPhee, L., Shoba, B. C., Pirok, E., & Steinberg, L. (1987). What kind of exposure reduces children's food neophobia? Looking vs. tasting. Appetite, 9(3), 171–178. Björntorp, P., Holm, G., & Rosmond, R. (1999). Hypothalamic arousal, insulin resistance and Type 2 diabetes mellitus. Diabetic Medicine : a Journal of the British Diabetic Association, 16(5), 373–383. http://doi.org/10.1046/j.14 Blair, D. (2009). The child in the garden: An evaluative review of the benefits of school gardening. The Journal of Environmental Education, 40(2), 15–38. Bolton, R. P., Heaton, K. W., & Burroughs, L. F. (1981). The role of dietary fiber in satiety, glucose, and insulin: studies with fruit and fruit juice. The American Journal of Clinical Nutrition, 34(2), 211–217. Booth, K. M., Pinkston, M. M., & Poston, W. S. C. (2005). Obesity and the Built Environment. Journal of the American Dietetic Association, 105(5), 110–117. http://doi.org/10.1016/j.jada.2005.02.045 Booth, M. L., Owen, N., Baum, A., Clavisi, O., & Leslie, E. (2000). Social–Cognitive and Perceived Environment Influences Associated with Physical Activity in Older Australians. Preventive Medicine, 31(1), 15–22. http://doi.org/10.1006/pmed.2000.0661 Bowler, D. E., Buyung-‐Ali, L. M., Knight, T. M., & Pullin, A. S. (2010). A systematic review of evidence for the added benefits to health of exposure to natural environments. BMC Public Health, 10(1), 456. http://doi.org/10.1186/1471-‐2458-‐10-‐456 Bratman, G. N., Hamilton, J. P., & Daily, G. C. (2012). The impacts of nature experience on human cognitive function and mental health. Annals of the New York Academy of Sciences, 1249(1), 118–136. http://doi.org/10.1111/j.1749-‐6632.2011.06400.x Bray, G. A., & Bouchard, C. (2003). Handbook of obesity: clinical applications. Bray, G. A., Nielsen, S. J., & Popkin, B. M. (2004). Consumption of high-‐fructose corn syrup in beverages may play a role in the epidemic of obesity. The American Journal of Clinical Nutrition, 79(4), 537–543. Bronfenbrenner, U. (1997). Ecological models of human development. Brown, B. B. (2004). Adolescents’ relationships with peers. In Handbook of adolescent psychology 2 (pp. 363–394). Brown, T., & Summerbell, C. (2009). Systematic review of school-‐based interventions that focus on changing dietary intake and physical activity levels to prevent childhood obesity: an update to the obesity guidance produced by the National Institute for Health and Clinical Excellence. Obesity Reviews, 10(1), 110–141. http://doi.org/10.1111/j.1467-‐ 789X.2008.00515.x Brownell, K. D., Kersh, R., Ludwig, D. S., Post, R. C., Puhl, R. M., Schwartz, M. B., & Willett, W. C. (2010). Personal Responsibility And Obesity: A Constructive Approach To A Controversial Issue. Health Affairs, 29(3), 379–387. http://doi.org/10.1377/hlthaff.2009.0739 Bruening, M., Eisenberg, M., MacLehose, R., Nanney, M. S., Story, M., & Neumark-‐Sztainer, D. (2012). Relationship between Adolescents’ and TheirFriends’ Eating Behaviors: Breakfast, Fruit,Vegetable, Whole-‐Grain, and Dairy Intake. Journal of the Academy of Nutrition and Dietetics, 112(10), 1608–1613. http://doi.org/10.1016/j.jand.2012.07.008 Bruening, M., MacLehose, R., Eisenberg, M. E., Nanney, M. S., Story, M., & Neumark-‐Sztainer, D. 90 (2014). Associations Between Sugar-‐Sweetened Beverage Consumption and Fast-‐Food Restaurant Frequency Among Adolescents and Their Friends. Journal of Nutrition Education and Behavior, 46(4), 277–285. http://doi.org/10.1016/j.jneb.2014.02.009 Butts, C. T. (2008). Social network analysis with sna. J. Stat. Softw, 24(6), 1–51. Butts, C. T. (2010). sna: Tools for social network analysis. R Package Version, 2. Butts, C. T. (2014). sna: Tools for Social Network Analysis. R package version 2.3-‐2. http://CRAN.R-‐project.org/package=sna. Caballero, B., Clay, T., Davis, S. M., Ethelbah, B., Rock, B. H., Lohman, T., et al. (2003). Pathways: a school-‐based, randomized controlled trial for the prevention of obesity in American Indian schoolchildren. The American Journal of Clinical Nutrition, 78(5), 1030–1038. Carney, P. A., Hamada, J. L., Rdesinski, R., Sprager, L., Nichols, K. R., Liu, B. Y., et al. (2011). Impact of a Community Gardening Project on Vegetable Intake, Food Security and Family Relationships: A Community-‐based Participatory Research Study. Journal of Community Health, 37(4), 874–881. http://doi.org/10.1007/s10900-‐011-‐9522-‐z Castro, D. C., Samuels, M., & Harman, A. E. (2013). Growing Healthy Kids: A community Garden-‐Based Obesity Prevention Program. American Journal of Preventive Medicine, 44(S3), S193–S199. http://doi.org/10.1016/j.amepre.2012.11.024 Cawley, J. (2010). The Economics Of Childhood Obesity. Health Affairs, 29(3), 364–371. http://doi.org/10.1377/hlthaff.2009.0721 Centers for Disease Control and Prevention (CDC). (2013a). Overweight and Obesity. Retrieved January 30, 2015, from http://www.cdc.gov/obesity/childhood/problem.html Centers for Disease Control and Prevention (CDC). (2013b). Tobacco product use among middle and high school students-‐-‐United States, 2011 and 2012. MMWR. Morbidity and Mortality Weekly Report, 62(45), 893–897. Cervinka, R., Roderer, K., & Hefler, E. (2012). Are nature lovers happy? On various indicators of well-‐being and connectedness with nature. Journal of Health Psychology, 17(3), 379– 388. http://doi.org/10.1177/1359105311416873 Chen, X., Beydoun, M. A., & Wang, Y. (2008). Is Sleep Duration Associated With Childhood Obesity? A Systematic Review and Meta-‐analysis. Obesity, 16(2), 265–274. http://doi.org/10.1038/oby.2007.63 Child population. (2013). Child population. Retrieved February 10, 2015, from http://www.childstats.gov/americaschildren/tables/pop1.asp Chiolero, A., Faeh, D., Paccaud, F., & Cornuz, J. (2008). Consequences of smoking for body weight, body fat distribution, and insulin resistance. American Journal of Clinical Nutrition, 87(4), 801–809. Christakis, N. A., & Fowler, J. H. (2007). The spread of obesity in a large social network over 32 years. The New England Journal of Medicine, 357(4), 370–379. http://doi.org/10.1056/NEJMsa066082 Clair, C., Chiolero, A., Faeh, D., Cornuz, J., Marques-‐Vidal, P., Paccaud, F., et al. (2011). Dose-‐ dependent positive association between cigarette smoking, abdominal obesity and body fat: cross-‐sectional data from a population-‐based survey. BMC Public Health, 11(1), 23. http://doi.org/10.1186/1471-‐2458-‐11-‐23 Contento, I. R., Williams, S. S., Michela, J. L., & Franklin, A. B. (2006). Understanding the food choice process of adolescents in the context of family and friends. Journal of Adolescent Health, 38(5), 575–582. http://doi.org/10.1016/j.jadohealth.2005.05.025 Cook, L. T., O'Reilly, G. A., DeRosa, C. J., Rohrbach, L. A., & Spruijt-‐Metz, D. (2014a). Association between home availability and vegetable consumption in youth: a review. Public Health 91 Nutrition, 1–9. http://doi.org/10.1017/S1368980014000664 Cook, L. T., O'Reilly, G. A., Goran, M. I., Weigensberg, M. J., Spruijt-‐Metz, D., & Davis, J. N. (2014b). Vegetable Consumption Is Linked to DecreasedVisceral and Liver Fat and Improved InsulinResistance in Overweight Latino Youth. Journal of the Academy of Nutrition and Dietetics, 1–8. http://doi.org/10.1016/j.jand.2014.01.017 Cooke, L. (2007). The importance of exposure for healthy eating in childhood: a review. J Hum Nutr Diet, 20, Pp. 294–301, 1–8. Cooper, A. J., Sharp, S. J., Lentjes, M. A. H., Luben, R. N., Khaw, K. T., Wareham, N. J., & Forouhi, N. G. (2012). A Prospective Study of the Association Between Quantity and Variety of Fruit and Vegetable Intake and Incident Type 2 Diabetes. Diabetes Care, 35(6), 1293–1300. http://doi.org/10.2337/dc11-‐2388 Cowie, C. C., Rust, K. F., Ford, E. S., Eberhardt, M. S., Byrd-‐Holt, D. D., Li, C., et al. (2009). Full Accounting of Diabetes and Pre-‐Diabetes in the US Population in 1988-‐1994 and 2005-‐ 2006. Diabetes Care, 32(2), 287–294. http://doi.org/10.2337/dc08-‐1296 Cradock, A. L., Melly, S. J., Allen, J. G., Morris, J. S., & Gortmaker, S. L. (2007). Characteristics of School Campuses and Physical Activity Among Youth. American Journal of Preventive Medicine, 33(2), 106–113.e1. http://doi.org/10.1016/j.amepre.2007.04.009 Cryer, P. E., Haymond, M. W., Santiago, J. V., & Shah, S. D. (1976). Norepinephrine and epinephrine release and adrenergic mediation of smoking-‐associated hemodynamic and metabolic events. The New England Journal of Medicine, 295(11), 573–577. http://doi.org/10.1056/NEJM197609092951101 Csardi, G., & Nepusz, T. (2006). The igraph software package for complex network research. InterJournal, Complex Systems, 1695(5). Retrieved from http://igraph.org Cummins, S. (2005). Food environments and obesity-‐-‐neighbourhood or nation? International Journal of Epidemiology, 35(1), 100–104. http://doi.org/10.1093/ije/dyi276 Dabbaghian, V., Mago, V. K., Wu, T., Fritz, C., & Alimadad, A. (2012). Social interactions of eating behaviour among high school students: a cellular automata approach. BMC Medical Research Methodology, 12, 155. http://doi.org/10.1186/1471-‐2288-‐12-‐155 Dallman, M. F., Pecoraro, N., Akana, S. F., La Fleur, S. E., Gomez, F., Houshyar, H., et al. (2003). Chronic stress and obesity: a new view of "comfort food". Proceedings of the National Academy of Sciences of the United States of America, 100(20), 11696–11701. http://doi.org/10.1073/pnas.1934666100 Daniels, S. R. (2009). REVIEWComplications of obesity in children and adolescents, 33(S1), S60–S65. http://doi.org/10.1038/ijo.2009.20 Davis, J. N., Martinez, L. C., Spruijt-‐Metz, D., & Gatto, N. M. (in press). LA Sprouts: A nutrition and gardening program improves determinants of dietary intake. Journal of Nutrition Education and Behavior. Davis, J. N., Ventura, E. E., Alexander, K. E., Salguero, L. E., Weigensberg, M. J., Crespo, N. C., et al. (2007). Feasibility of a home-‐based versus classroom-‐based nutrition intervention to reduce obesity and type 2 diabetes in Latino youth. International Journal of Pediatric Obesity, 2(1), 22–30. http://doi.org/10.1080/17477160601133077 Davis, J. N., Ventura, E. E., Cook, L. T., Gyllenhammer, L. E., & Gatto, N. M. (2011). LA Sprouts: a gardening, nutrition, and cooking intervention for Latino youth improves diet and reduces obesity. Journal of the American Dietetic Association, 111(8), 1224–1230. http://doi.org/10.1016/j.jada.2011.05.009 Davison, K. K., & Lawson, C. T. (2006). Do attributes in the physical environment influence children's physical activity? A review of the literature. International Journal of Behavioral 92 Nutrition and Physical Activity, 3(1), 19. http://doi.org/10.1186/1479 de Castro, J. M. (1994). Family and friends produce greater social facilitation of food intake than other companions. Physiology & Behavior, 56(3), 445–445. de la Haye, de, K., Robins, G., Mohr, P., & Wilson, C. (2010). Social Networks. Social Networks, 32(3), 161–167. http://doi.org/10.1016/j.socnet.2009.09.001 de la Haye, de, K., Robins, G., Mohr, P., & Wilson, C. (2013). Adolescents’ Intake of Junk Food: Processes and Mechanisms Driving Consumption Similarities Among Friends. Journal of Research on Adolescence, 23(3), 524–536. http://doi.org/10.1111/jora.12045 Deci, E. L., & Ryan, R. M. (2008). Self-‐determination theory: A macrotheory of human motivation, development, and health. Canadian Psychology/Psychologie …. Dennison, B. A., Erb, T. A., & Jenkins, P. L. (2002). Television viewing and television in bedroom associated with overweight risk among low-‐income preschool children. Pediatrics, 109(6), 1028–1035. Dietz, W. H. (1998). Health consequences of obesity in youth: childhood predictors of adult disease. Pediatrics, 101(3 Pt 2), 518–525. Ding, D., & Gebel, K. (2012). Health & Place. Health & Place, 18(1), 100–105. http://doi.org/10.1016/j.healthplace.2011.08.021 Ding, D., Sallis, J. F., Kerr, J., Lee, S., & Rosenberg, D. E. (2011). Neighborhood Environment and PhysicalActivity Among Youth. Am J Prev Med, 41(4), 442–455. http://doi.org/10.1016/j.amepre.2011.06.036 Domel, S. B., Baranowski, T., Davis, H., Leonard, S. B., Riley, P., & Baranowski, J. (1993). Measuring fruit and vegetable preferences among 4th-‐ and 5th-‐grade students. Preventive Medicine, 22(6), 866–879. http://doi.org/10.1006/pmed.1993.1078 Doyle, W., & Rutherford, B. (1984). Classroom research on matching learning and teaching styles. Theory Into Practice, 23(1), 20–25. Dunton, G. F., Huh, J., Leventhal, A. M., Riggs, N., Hedeker, D., Spruijt-‐Metz, D., & Pentz, M. A. (2014). Momentary assessment of affect, physical feeling states, and physical activity in children. Health Psychology, 33(3), 255–263. http://doi.org/10.1037/a0032640 Dunton, G. F., Kaplan, J., Wolch, J., Jerrett, M., & Reynolds, K. D. (2009a). Physical environmental correlates of childhood obesity: a systematic review. Obesity Reviews, 10(4), 393–402. http://doi.org/10.1111/j.1467-‐789X.2009.00572.x Dunton, G. F., Kawabata, K., Intille, S., Wolch, J., & Pentz, M. A. (2012). Assessing the social and physical contexts of children's leisure-‐time physical activity: an ecological momentary assessment study. American Journal of Health Promotion : AJHP, 26(3), 135–142. http://doi.org/10.4278/ajhp.100211-‐QUAN-‐43 Dunton, G. F., Liao, Y., Intille, S. S., Spruijt-‐Metz, D., & Pentz, M. (2009b). Investigating Children's Physical Activity and Sedentary Behavior Using Ecological Momentary Assessment With Mobile Phones. Obesity, 19(6), 1205–1212. http://doi.org/10.1038/oby.2010.302 Dunton, G. F., Liao, Y., Intille, S., Wolch, J., & Pentz, M. A. (2011). Physical and social contextual influences on children's leisure-‐time physical activity: an ecological momentary assessment study. Journal of Physical Activity & Health, 8 Suppl 1, S103–8. Dyment, J. E., & Bell, A. C. (2007). Grounds for movement: green school grounds as sites for promoting physical activity. Health Education Research, 23(6), 952–962. http://doi.org/10.1093/her/cym059 Ebbeling, C. B., Pawlak, D. B., & Ludwig, D. S. (2002). Childhood obesity: public-‐health crisis, common sense cure. Lancet, 360(9331), 473–482. http://doi.org/10.1016/S0140-‐ 93 6736(02)09678-‐2 Eisenberg, M. E., & Neumark-‐Sztainer, D. (2010). Friends’ Dieting and Disordered Eating Behaviors Among Adolescents Five Years Later: Findings From Project EAT. Journal of Adolescent Health, 47(1), 67–73. http://doi.org/10.1016/j.jadohealth.2009.12.030 Eriksson, T., Westerberg, Y., & Jonsson, H. (2011). Experiences of women with stress-‐related ill health in a therapeutic gardening program. Canadian Journal of Occupational Therapy, 78(5). http://doi.org/10.2182/cjot.2011.78.5.2 Esmaillzadeh, A., Kimiagar, M., Mehrabi, Y., Azadbakht, L., Hu, F. B., & Willett, W. C. (2006). Fruit and vegetable intakes, C-‐reactive protein, and the metabolic syndrome. The American Journal of Clinical Nutrition, 84(6), 1489–1497. Evans, A., Ranjit, N., Rutledge, R., Medina, J., Jennings, R., Smiley, A., et al. (2012). Exposure to Multiple Components of a Garden-‐Based Intervention for Middle School Students Increases Fruit and Vegetable Consumption. Health Promotion Practice, 13(5), 608–616. http://doi.org/10.1177/1524839910390357 Evans, G. W. (2003). The built environment and mental health. Journal of Urban Health, 80(4), 536–555. Evans, G. W., & English, K. (2002). The environment of poverty: multiple stressor exposure, psychophysiological stress, and socioemotional adjustment. Child Development, 73(4), 1238–1248. Evans, G. W., Wells, N. M., & Moch, A. (2003). Housing and mental health: A review of the evidence and a methodological and conceptual critique. Journal of Social Issues, 59(3), 475–500. Feda, D. M., Seelbinder, A., Baek, S., Raja, S., Yin, L., & Roemmich, J. N. (2014). Neighbourhood parks and reduction in stress among adolescents: results from Buffalo, New York. Indoor and Built Environment. http://doi.org/10.1177/1420326X14535791 Feunekes, G. I., de Graaf, C., Meyboom, S., & van Staveren, W. A. (1998). Food choice and fat intake of adolescents and adults: associations of intakes within social networks. Preventive Medicine, 27(5 Pt 1), 645–656. http://doi.org/10.1006/pmed.1998.0341 Fitzgerald, A., Heary, C., Kelly, C., Nixon, E., & Shevlin, M. (2013). Self-‐efficacy for healthy eating and peer support for unhealthy eating are associated with adolescents food intake patterns. Appetite, 63(C), 48–58. http://doi.org/10.1016/j.appet.2012.12.011 Fletcher, A., Bonell, C., & Sorhaindo, A. (2011). You are what your friends eat: systematic review of social network analyses of young people's eating behaviours and bodyweight. Journal of Epidemiology & Community Health, 65(6), 548–555. http://doi.org/10.1136/jech.2010.113936 Ford, E. S., & Mokdad, A. H. (2001). Fruit and Vegetable Consumption and Diabetes Mellitus Incidence among U.S. Adults. Preventive Medicine, 32(1), 33–39. http://doi.org/10.1006/pmed.2000.0772 Ford, E. S., Ajani, U. A., McGuire, L. C., & Liu, S. (2005). Concentrations of serum vitamin D and the metabolic syndrome among U.S. adults. Diabetes Care, 28(5), 1228–1230. Fox, K. R. (1999). The influence of physical activity on mental well-‐being. Public Health Nutrition, 2(3A), 411–418. Freedman, D. S., Khan, L. K., Serdula, M. K., Dietz, W. H., Srinivasan, S. R., & Berenson, G. S. (2003). Inter-‐relationships among childhood BMI, childhood height, and adult obesity: the Bogalusa Heart Study. International Journal of Obesity, 28(1), 10–16. http://doi.org/10.1038/sj.ijo.0802544 Freedman, D. S., Srinivasan, S. R., Burke, G. L., Shear, C. L., Smoak, C. G., Harsha, D. W., et al. 94 (1987). Relation of Body-‐Fat Distribution to Hyperinsulinemia in Children and Adolescents -‐ the Bogalusa-‐Heart-‐Study. The American Journal of Clinical Nutrition, 46(3), 403–410. French, S. A., Story, M., & Jeffery, R. W. (2001). Environmental influences on eating and physical activity. Annual Review of Public Health, 22(1), 309–335. Gangwisch, J. E., Heymsfield, S. B., Boden-‐Albala, B., Buijs, R. M., Kreier, F., Pickering, T. G., et al. (2006). Short Sleep Duration as a Risk Factor for Hypertension: Analyses of the First National Health and Nutrition Examination Survey. Hypertension, 47(5), 833–839. http://doi.org/10.1161/01.HYP.0000217362.34748.e0 Garcia-‐Dominic, O., Trevino, R. P., Echon, R. M., Mobley, C., Block, T., Bizzari, A., & Michalek, J. (2012). Improving Quality of Food Frequency Questionnaire Response in Low-‐Income Mexican American Children. Health Promotion Practice, 13(6), 763–771. http://doi.org/10.1177/1524839911405847 Gatto, N. M., Martinez, L. C., Spruijt-‐Metz, D., & Davis, J. N. (under review). LA sprouts randomized controlled nutrition and gardening program reduces obesity and metabolic risk in latino youth. Gatto, N. M., Ventura, E. E., Cook, L. T., Gyllenhammer, L. E., & Davis, J. N. (2012). LA Sprouts: a garden-‐based nutrition intervention pilot program influences motivation and preferences for fruits and vegetables in Latino youth. Journal of the Academy of Nutrition and Dietetics, 112(6), 913–920. http://doi.org/10.1016/j.jand.2012.01.014 George, L. K. (2006). Perceived quality of life. Handbook of Aging and the Social Sciences, 6, 320–336. Gibbs, L., Staiger, P. K., Johnson, B., Block, K., Macfarlane, S., Gold, L., et al. (2013). Expanding Children’s Food Experiences: The Impact of a School-‐Based Kitchen Garden Program. Journal of Nutrition Education and Behavior, 45(2), 137–146. http://doi.org/10.1016/j.jneb.2012.09.004 Glasson, C., Chapman, K., & James, E. (2010). Fruit and vegetables should be targeted separately in health promotion programmes: differences in consumption levels, barriers, knowledge and stages of readiness for change. Public Health Nutrition, 14(04), 694–701. http://doi.org/10.1017/S1368980010001643 Gleason, P. M., & Suitor, C. W. (2003). Eating at School: How the National School Lunch Program affects children's diets. Am J Agr Econ, 85(4), 1047–1061. Gonzalez-‐Suarez, C., Worley, A., Grimmer-‐Somers, K., & Dones, V. (2009). School-‐Based Interventions on Childhood Obesity. Am J Prev Med, 37(5), 418–427. http://doi.org/10.1016/j.amepre.2009.07.012 Goran, M. I., Dumke, K., Bouret, S. G., Kayser, B., Walker, R. W., & Blumberg, B. (2013). The obesogenic effect of high fructose exposure during early development. Nature Publishing Group, 9(8), 494–500. http://doi.org/10.1038/nrendo.2013.108 Goran, M. I., Lane, C., Toledo-‐Corral, C., & Weigensberg, M. J. (2008). Persistence of Pre-‐ Diabetes in Overweight and Obese Hispanic Children: Association With Progressive Insulin Resistance, Poor -‐Cell Function, and Increasing Visceral Fat. Diabetes, 57(11), 3007–3012. http://doi.org/10.2337/db08-‐0445 Goran, M. I., Walker, R., Le, K. A., Mahurkar, S., Vikman, S., Davis, J. N., et al. (2010). Effects of PNPLA3 on Liver Fat and Metabolic Profile in Hispanic Children and Adolescents. Diabetes, 59(12), 3127–3130. http://doi.org/10.2337/db10-‐0554 Gordon, A. R., Devaney, B. L., & Burghardt, J. A. (1995). Dietary effects of the National School Lunch Program and the School Breakfast Program. The American Journal of Clinical 95 Nutrition, 61(1 Suppl), 221S–231S. Graham, J. W. (2012). Missing Data Theory (pp. 3–46). New York, NY: Springer New York. http://doi.org/10.1007/978-‐1-‐4614-‐4018-‐5_1 Grossman, P., Niemann, L., Schmidt, S., & Walach, H. (2004). Mindfulness-‐based stress reduction and health benefits. Journal of Psychosomatic Research, 57(1), 35–43. http://doi.org/10.1016/S0022-‐3999(03)00573-‐7 Guillaumie, L., Godin, G., & Vezina-‐Im, L.-‐A. (2010). Psychosocial determinants of fruit and vegetable intake in adult population: a systematic review. International Journal of Behavioral Nutrition and Physical Activity, 7(1), 12. http://doi.org/10.1186/1479-‐5868-‐7-‐ 12 Gutin, B., Islam, S., Manos, T., Cucuzzo, N., Smith, C., & Stachura, M. E. (1994). Relation of Percentage of Body-‐Fat and Maximal Aerobic Capacity to Risk-‐Factors for Atherosclerosis and Diabetes in Black-‐and-‐White 7-‐Year-‐Old to 11-‐Year-‐Old Children. Journal of Pediatrics, 125(6), 847–852. Handcock, M. S., Hunter, D. R., Butts, C. T., Goodreau, S. M., & Morris, M. (2003). statnet: Software tools for the Statistical Modeling of Network Data. Retrieved January 15, 2015, from http://statnetproject.org Harnack, L., Stang, J., & Story, M. (1999). Soft Drink Consumption Among US Children and Adolescents. Journal of the American Dietetic Association, 99(4), 436–441. http://doi.org/10.1016/S0002-‐8223(99)00106-‐6 Hartig, T., Evans, G. W., Jamner, L. D., Davis, D. S., & Gärling, T. (2003). Tracking restoration in natural and urban field settings. Journal of Environmental Psychology, 23(2), 109–123. http://doi.org/10.1016/S0272-‐4944(02)00109-‐3 Haug, E., Torsheim, T., & Samdal, O. (2008). Physical environmental characteristics and individual interests as correlates of physical activity in Norwegian secondary schools: The health behaviour in school-‐aged children study. International Journal of Behavioral Nutrition and Physical Activity, 5(1), 47. http://doi.org/10.1186/1479-‐5868-‐5-‐47 Haug, E., Torsheim, T., Sallis, J. F., & Samdal, O. (2010). The characteristics of the outdoor school environment associated with physical activity. Health Education Research, 25(2), 248–256. http://doi.org/10.1093/her/cyn050 He, K., Hu, F. B., Colditz, G. A., Manson, J. E., Willett, W. C., & Liu, S. (2004). Changes in intake of fruits and vegetables in relation to risk of obesity and weight gain among middle-‐aged women. International Journal of Obesity, 28(12), 1569–1574. http://doi.org/10.1038/sj.ijo.0802795 Hearn, M. D., Baranowski, T., Baranowski, J., Doyle, C., Smith, M., Lin, L. S., & Resnicow, K. (1998). Environmental Influences on Dietary Behavior among Children: Availability and Accessibility of Fruits and Vegetables Enable Consumption. Journal of Health Education, 29(1), 26–32. http://doi.org/10.1080/10556699.1998.10603294 Hedeker, D., Mermelstein, R. J., & Demirtas, H. (2012). Modeling between-‐subject and within-‐ subject variances in ecological momentary assessment data using mixed-‐effects location scale models. Statistics in Medicine, 31(27), 3328–3336. http://doi.org/10.1002/sim.5338 Hedeker, D., Mermelstein, R. J., Berbaum, M. L., & Campbell, R. T. (2009). Modeling mood variation associated with smoking: an application of a heterogeneous mixed-‐effects model for analysis of ecological momentary assessment (EMA) data. Addiction, 104(2), 297–307. http://doi.org/10.1111/j.1360-‐0443.2008.02435.x Heim, S., Stang, J., & Ireland, M. (2009). A Garden Pilot Project Enhances Fruit and Vegetable Consumption among Children. Journal of the American Dietetic Association, 109(7), 1220– 96 1226. http://doi.org/10.1016/j.jada.2009.04.009 Herman, C. P., Roth, D. A., & Polivy, J. (2003). Effects of the Presence of Others on Food Intake: A Normative Interpretation. Psychological Bulletin, 129(6), 873–886. http://doi.org/10.1037/0033-‐2909.129.6.873 Hermann, J. R., Parker, S. P., Brown, B. J., Siewe, Y. J., Denney, B. A., & Walker, S. J. (2006). After-‐ School Gardening Improves Children’s Reported Vegetable Intake and Physical Activity. Journal of Nutrition Education and Behavior, 38(3), 201–202. http://doi.org/10.1016/j.jneb.2006.02.002 Herzog, T. R., Black, A. M., Fountaine, K. A., & Knotts, D. J. (1997). Reflection and attentional recovery as distinctive benefits of restorative environments. Journal of Environmental Psychology, 17(2), 165–170. Howell, A. J., Dopko, R. L., Passmore, H.-‐A., & Buro, K. (2011). Nature connectedness: Associations with well-‐being and mindfulness. Personality and Individual Differences, 51(2), 166–171. http://doi.org/10.1016/j.paid.2011.03.037 Hruschka, D. J., Brewis, A. A., Wutich, A., & Morin, B. (2011). Shared Norms and Their Explanation for the Social Clustering of Obesity. American Journal of Public Health, 101(S1), S295–S300. http://doi.org/10.2105/AJPH.2010.300053 Hu, F. B., & Malik, V. S. (2010). Physiology & Behavior. Physiology & Behavior, 100(1), 47–54. http://doi.org/10.1016/j.physbeh.2010.01.036 Hung, H. C., Joshipura, K. J., Jiang, R., Hu, F. B., Hunter, D., Smith-‐Warner, S. A., et al. (2004). Fruit and Vegetable Intake and Risk of Major Chronic Disease. JNCI Journal of the National Cancer Institute, 96(21), 1577–1584. http://doi.org/10.1093/jnci/djh296 Hutchinson, D., & Rapee, R. (2007). Do friends share similar body image and eating problems? The role of social networks and peer influences in early adolescence. Behaviour Research and Therapy, 45(7), 1557–1577. http://doi.org/10.1016/j.brat.2006.11.007 Ihaka, R., & Gentleman, R. (1996). R: A Language for Data Analysis and Graphics. Journal of Computational and Graphical Statistics, 5(3), 299–314. http://doi.org/10.1080/10618600.1996.10474713 Jago, R., Baranowski, T., & Baranowski, J. C. (2007). Fruit and vegetable availability: a micro environmental mediating variable? Public Health Nutrition, 10(07). http://doi.org/10.1017/S1368980007441441 Kahn, S. E., Hull, R. L., & Utzschneider, K. M. (2006). Mechanisms linking obesity to insulin resistance and type 2 diabetes. Nature, 444(7121), 840–846. http://doi.org/10.1038/nature05482 Kaplan, R., & Kaplan, S. (1989). The experience of nature: A psychological perspective. CUP Archive. Kaplan, S. (1995). The restorative benefits of nature: Toward an integrative framework. Journal of Environmental Psychology, 15(3), 169–182. Kimmons, J., Gillespie, C., Seymour, J., Serdula, M., & Blanck, H. M. (2009). Fruit and vegetable intake among adolescents and adults in the United States: percentage meeting individualized recommendations. Medscape Journal of Medicine, 11(1), 26. Kuczmarski, R. J., Ogden, C. L., Guo, S. S., Grummer-‐Strawn, L. M., Flegal, K. M., Mei, Z., et al. (2002). 2000 CDC Growth Charts for the United States: methods and development. Vital and Health Statistics. Series 11, Data From the National Health Survey, (246), 1–190. Kwan, M.-‐P. (2012a). How GIS can help address the uncertain geographic context problem in social science research. Annals of GIS, 18(4), 245–255. http://doi.org/10.1080/19475683.2012.727867 97 Kwan, M.-‐P. (2012b). The Uncertain Geographic Context Problem. Annals of the Association of American Geographers, 102(5), 958–968. http://doi.org/10.1080/00045608.2012.687349 Lancaster, B. P. (1999). Defining and Interpreting Suppressor Effects: Advantages and Limitations. Larson, N. I., Story, M. T., & Nelson, M. C. (2009). Neighborhood Environments. Am J Prev Med, 36(1), 74–81.e10. http://doi.org/10.1016/j.amepre.2008.09.025 Lattimer, J. M., & Haub, M. D. (2010). Effects of Dietary Fiber and Its Components on Metabolic Health. Nutrients, 2(12), 1266–1289. http://doi.org/10.3390/nu2121266 Lautenschlager, L., & Smith, C. (2007). Beliefs, knowledge, and values held by inner-‐city youth about gardening, nutrition, and cooking. Agriculture and Human Values, 24(2), 245–258. http://doi.org/10.1007/s10460-‐006-‐9051-‐z Leary, M. R., & Kowalski, R. M. (1990). Impression management: A literature review and two-‐ component model. Psychological Bulletin, 107(1), 34. Ledoux, T. A., Hingle, M. D., & Baranowski, T. (2010). Relationship of fruit and vegetable intake with adiposity: a systematic review. Obesity Reviews, 12(5), e143–e150. http://doi.org/10.1111/j.1467-‐789X.2010.00786.x Lee, A. C. K., & Maheswaran, R. (2011). The health benefits of urban green spaces: a review of the evidence. Journal of Public Health, 33(2), 212–222. http://doi.org/10.1093/pubmed/fdq068 Lee, J., Park, B. J., Tsunetsugu, Y., Ohira, T., Kagawa, T., & Miyazaki, Y. (2011). Effect of forest bathing on physiological and psychological responses in young Japanese male subjects. Public Health, 125(2), 93–100. http://doi.org/10.1016/j.puhe.2010.09.005 Let's Move! (2015). Let's Move! Retrieved January 30, 2015, from http://www.letsmove.gov Lineberger, S. E., & Zajicek, J. M. (2000). School gardens: Can a hands-‐on teaching tool affect students' attitudes and behaviors regarding fruit and vegetables? Horttechnology, 10(3), 593–597. Liu, S., Serdula, M., Janket, S.-‐J., Cook, N. R., Sesso, H. D., Willett, W. C., et al. (2004). A prospective study of fruit and vegetable intake and the risk of type 2 diabetes in women. Diabetes Care, 27(12), 2993–2996. Lobstein, T., Baur, L., Uauy, R., IASO International Obesity TaskForce. (2004). Obesity in children and young people: a crisis in public health. Obesity Reviews : an Official Journal of the International Association for the Study of Obesity, 5 Suppl 1, 4–104. http://doi.org/10.1111/j.1467-‐789X.2004.00133.x Lopes, V. P., Vasques, C. M. S., Maia, J. A. R., & Ferreira, J. C. V. (2007). Habitual physical activity levels in childhood and adolescence assessed with accelerometry. The Journal of Sports Medicine and Physical Fitness, 47(2), 217–222. Lubans, D. R., Foster, C., & Biddle, S. J. H. (2008). Preventive Medicine. Preventive Medicine, 47(5), 463–470. http://doi.org/10.1016/j.ypmed.2008.07.011 Ludwig, D. S. (2002). The glycemic index: physiological mechanisms relating to obesity, diabetes, and cardiovascular disease. JAMA: the Journal of the American Medical Association, 287(18), 2414–2423. Ludwig, D. S., Peterson, K. E., & Gortmaker, S. L. (2001). Relation between consumption of sugar-‐sweetened drinks and childhood obesity: a prospective, observational analysis. Lancet, 357(9255), 505–508. http://doi.org/10.1016/S0140-‐6736(00)04041-‐1 Lumeng, J. C., Somashekar, D., Appugliese, D., Kaciroti, N., Corwyn, R. F., & Bradley, R. H. (2007). Shorter Sleep Duration Is Associated With Increased Risk for Being Overweight at 98 Ages 9 to 12 Years. Pediatrics, 120(5), 1020–1029. http://doi.org/10.1542/peds.2006-‐ 3295 Lusher, D., Koskinen, J., & Robins, G. (Eds.). (2013). Exponential random graph models for social networks: Theory, Methods, and Applications. Cambridge University Press. Lustig, R. H., Schmidt, L. A., & Brindis, C. D. (2012). Public health: The toxic truth about sugar. Nature, 482(7383), 27–29. http://doi.org/10.1038/482027a Lytle, L. A., Stone, E. J., Nichaman, M. Z., Perry, C. L., Montgomery, D. H., Nicklas, T. A., et al. (1996). Changes in nutrient intakes of elementary school children following a school-‐ based intervention: results from the CATCH Study. Preventive Medicine, 25(4), 465–477. http://doi.org/10.1006/pmed.1996.0078 Maas, J., Verheij, R. A., Spreeuwenberg, P., & Groenewegen, P. P. (2008). Physical activity as a possible mechanism behind the relationship between green space and health: A multilevel analysis. BMC Public Health, 8(1), 206. http://doi.org/10.1186/1471-‐2458-‐8-‐206 Malik, V. S., Popkin, B. M., Bray, G. A., Despres, J. P., Willett, W. C., & Hu, F. B. (2010). Sugar-‐ Sweetened Beverages and Risk of Metabolic Syndrome and Type 2 Diabetes: A meta-‐ analysis. Diabetes Care, 33(11), 2477–2483. http://doi.org/10.2337/dc10-‐1079 Malik, V. S., Schulze, M. B., & Hu, F. B. (2006). Intake of sugar-‐sweetened beverages and weight gain: a systematic review. The American Journal of Clinical Nutrition, 84(2), 274–288. Martinez, L. C., Gatto, N. M., Chou, C.-‐P., O'Reilly, G. A., Spruijt-‐Metz, D., & Davis, J. N. (under review). Determinants of fruit and vegetable preferences and intake in Hispanic/Latino children. Appetite. Martinez, L. C., Gatto, N. M., Spruijt-‐Metz, D., & Davis, J. N. (2015). Design and methodology of the LA Sprouts nutrition, cooking and gardening program for Latino youth: a randomized controlled intervention. Contemporary Clinical Trials, 219–227. Matthews, C. E., Chen, K. Y., Freedson, P. S., Buchowski, M. S., Beech, B. M., Pate, R. R., & Troiano, R. P. (2008). Amount of Time Spent in Sedentary Behaviors in the United States, 2003-‐2004. American Journal of Epidemiology, 167(7), 875–881. http://doi.org/10.1093/aje/kwm390 Matthews, S. A., & Yang, T.-‐C. (2010). Exploring the Role of the Built and Social Neighborhood Environment in Moderating Stress and Health. Annals of Behavioral Medicine : a Publication of the Society of Behavioral Medicine, 39(2), 170–183. http://doi.org/10.1007/s12160-‐010-‐9175-‐7 McAleese, J. D., & Rankin, L. L. (2007). Garden-‐Based Nutrition Education Affects Fruit and Vegetable Consumption in Sixth-‐Grade Adolescents. Journal of the American Dietetic Association, 107(4), 662–665. http://doi.org/10.1016/j.jada.2007.01.015 McClain, A. D., Chappuis, C., Nguyen-‐Rodriguez, S. T., Yaroch, A. L., & Spruijt-‐Metz, D. (2009). Psychosocial correlates of eating behavior in children and adolescents: a review. International Journal of Behavioral Nutrition and Physical Activity, 6(1), 54. http://doi.org/10.1186/1479-‐5868-‐6-‐54 McEwen, B. S. (2006). Protective and damaging effects of stress mediators: central role of the brain. Dialogues in Clinical Neuroscience, 8(4), 367–381. McEwen, B. S. (2008). Central effects of stress hormones in health and disease: Understanding the protective and damaging effects of stress and stress mediators. European Journal of Pharmacology, 583(2-‐3), 174–185. http://doi.org/10.1016/j.ejphar.2007.11.071 McLaren, L. (2007). Socioeconomic Status and Obesity. Epidemiologic Reviews, 29(1), 29–48. http://doi.org/10.1093/epirev/mxm001 Mendelson, T., Greenberg, M. T., Dariotis, J. K., Gould, L. F., Rhoades, B. L., & Leaf, P. J. (2010). 99 Feasibility and Preliminary Outcomes of a School-‐Based Mindfulness Intervention for Urban Youth. Journal of Abnormal Child Psychology, 38(7), 985–994. http://doi.org/10.1007/s10802-‐010-‐9418-‐x Millimet, D. L., Tchernis, R., & husain, M. (2010). School Nutrition Programs and the Incidence of Childhood Obesity. J Human Resources, 45(3), 640–654. Mitchell, R. (2013). Is physical activity in natural environments better for mental health than physical activity in other environments? Social Science & Medicine, 91(C), 130–134. http://doi.org/10.1016/j.socscimed.2012.04.012 Moksnes, U. K., Byrne, D. G., Mazanov, J., & Espnes, G. A. (2010a). Adolescent stress: Evaluation of the factor structure of the Adolescent Stress Questionnaire (ASQ-‐N). Scandinavian Journal of Psychology, 51(3), 203–209. http://doi.org/10.1111/j.1467-‐9450.2009.00803.x Moksnes, U. K., Moljord, I. E. O., Espnes, G. A., & Byrne, D. G. (2010b). Personality and Individual Differences. Personality and Individual Differences, 49(5), 430–435. http://doi.org/10.1016/j.paid.2010.04.012 Montonen, J., Järvinen, R., Heliövaara, M., Reunanen, A., Aromaa, A., & Knekt, P. (2005). Food consumption and the incidence of type II diabetes mellitus. European Journal of Clinical Nutrition, 59(3), 441–448. http://doi.org/10.1038/sj.ejcn.1602094 Morgan, P. J., Warren, J. M., Lubans, D. R., Saunders, K. L., Quick, G. I., & Collins, C. E. (2010). The impact of nutrition education with and without a school garden on knowledge, vegetable intake and preferences and quality of school life among primary-‐school students. Public Health Nutrition, 13(11), 1931–1940. http://doi.org/10.1017/S1368980010000959 Morris, J. L., & Zidenberg-‐Cherr, S. (2002). Garden-‐enhanced nutrition curriculum improves fourth-‐grade school children's knowledge of nutrition and preferences for some vegetables. Journal of the American Dietetic Association, 102(1), 91–93. http://doi.org/10.1016/S0002-‐8223(02)90027-‐1 Nguyen-‐Michel, S. T., Unger, J. B., & Spruijt-‐Metz, D. (2007). Dietary correlates of emotional eating in adolescence. Appetite, 49(2), 494–499. http://doi.org/10.1016/j.appet.2007.03.005 Nichol, M. E., Pickett, W., & Janssen, I. (2009). Associations between school recreational environments and physical activity. The Journal of School Health, 79(6), 247–254. http://doi.org/10.1111/j.1746-‐1561.2009.00406.x Nielsen, S. J., & Popkin, B. M. (2004). Changes in beverage intake between 1977 and 2001. American Journal of Preventive Medicine, 27(3), 205–210. http://doi.org/10.1016/j.amepre.2004.05.005 O'Reilly, G. A., Cook, L., Spruijt-‐Metz, D., & Black, D. S. (2014). Mindfulness-‐based interventions for obesity-‐related eating behaviours: a literature review. Obesity Reviews, 15(6), 453– 461. http://doi.org/10.1111/obr.12156 Ogden, C. L. (2012). Prevalence of Obesity and Trends in Body Mass Index Among US Children and Adolescents, 1999-‐2010. JAMA: the Journal of the American Medical Association, 307(5), 483. http://doi.org/10.1001/jama.2012.40 Owen, N., Healy, G. N., Matthews, C. E., & Dunstan, D. W. (2010). Too Much Sitting. Exercise and Sport Sciences Reviews, 38(3), 105–113. http://doi.org/10.1097/JES.0b013e3181e373a2 Pan, Y., & Pratt, C. A. (2008). Metabolic Syndrome and Its Association with Diet and Physical Activity in US Adolescents. Journal of the American Dietetic Association, 108(2), 276–286. http://doi.org/10.1016/j.jada.2007.10.049 Papas, M. A., Alberg, A. J., Ewing, R., Helzlsouer, K. J., Gary, T. L., & Klassen, A. C. (2007). The 100 Built Environment and Obesity. Epidemiologic Reviews, 29(1), 129–143. http://doi.org/10.1093/epirev/mxm009 Park, B.-‐J., Furuya, K., Kasetani, T., Takayama, N., Kagawa, T., & Miyazaki, Y. (2011). Relationship between sychological responses and physical environments in forest settings. Landscape and Urban Planning, 102(1), 24–32. http://doi.org/10.1016/j.landurbplan.2011.03.005 Parmer, S. M., Salisbury-‐Glennon, J., Shannon, D., & Struempler, B. (2009). School Gardens: An Experiential Learning Approach for a Nutrition Education Program to Increase Fruit and Vegetable Knowledge, Preference, and Consumption among Second-‐grade Students. Journal of Nutrition Education and Behavior, 41(3), 212–217. http://doi.org/10.1016/j.jneb.2008.06.002 Pearson, N., Biddle, S. J., & Gorely, T. (2008). Family correlates of fruit and vegetable consumption in children and adolescents: a systematic review. Public Health Nutrition, 12(02), 267. http://doi.org/10.1017/S1368980008002589 Pliner, P., & Hobden, K. (1992). Development of a scale to measure the trait of food neophobia in humans. Appetite, 19(2), 105–120. Pliner, P., Bell, R., Hirsch, E. S., & Kinchla, M. (2006). Meal duration mediates the effect of “social facilitation” on eating in humans. Appetite, 46(2), 189–198. http://doi.org/10.1016/j.appet.2005.12.003 Pons, P., & Latapy, M. (2006). Computing communities in large networks using random walks. J Graph Algorithms Appl. Power, C., Lake, J. K., & Cole, T. J. (1997). Measurement and long-‐term health risks of child and adolescent fatness. International Journal of Obesity, 21(7), 507–526. Qureshi, M. M., Singer, M. R., & Moore, L. L. (2009). A cross-‐sectional study of food group intake and C-‐reactive protein among children. Nutrition & Metabolism, 6(1), 40. http://doi.org/10.1186/1743-‐7075-‐6-‐40 Ramirez, A. J., Graham, J., Richards, M. A., Cull, A., & Gregory, W. M. (1996). Mental health of hospital consultants: the effects of stress and satisfaction at work. Lancet, 347(9003), 724–728. Rasmussen, M., Krølner, R., Klepp, K.-‐I., Lytle, L., Brug, J., Bere, E., & Due, P. (2006). Determinants of fruit and vegetable consumption among children and adolescents: a review of the literature. Part I: quantitative studies. International Journal of Behavioral Nutrition and Physical Activity, 3(1), 22. Ratcliffe, M. M., Merrigan, K. A., Rogers, B. L., & Goldberg, J. P. (2010). The Effects of School Garden Experiences on Middle School-‐Aged Students' Knowledge, Attitudes, and Behaviors Associated With Vegetable Consumption. Health Promotion Practice, 12(1), 36– 43. http://doi.org/10.1177/1524839909349182 Ray, T. K., Mansell, K. M., Knight, L. C., Malmud, L. S., Owen, O. E., & Boden, G. (1983). Long-‐ term effects of dietary fiber on glucose tolerance and gastric emptying in noninsulin-‐ dependent diabetic patients. The American Journal of Clinical Nutrition, 37(3), 376–381. Reibel, D. K., Greeson, J. M., Brainard, G. C., & Rosenzweig, S. (2001). Mindfulness-‐based stress reduction and health-‐related quality of life in a heterogeneous patient population. General Hospital Psychiatry, 23(4), 183–192. Rennie, K. L., Johnson, L., & Jebb, S. A. (2005). Behavioural determinants of obesity. Best Practice & Research Clinical Endocrinology & Metabolism, 19(3), 343–358. http://doi.org/10.1016/j.beem.2005.04.003 Robinson-‐O'Brien, R., Story, M., & Heim, S. (2009). Impact of Garden-‐Based Youth Nutrition 101 Intervention Programs: A Review. Journal of the American Dietetic Association, 109(2), 273–280. http://doi.org/10.1016/j.jada.2008.10.051 Roemmich, J. N., Epstein, L. H., Raja, S., & Yin, L. (2007). The neighborhood and home environments: disparate relationships with physical activity and sedentary behaviors in youth. Annals of Behavioral Medicine : a Publication of the Society of Behavioral Medicine, 33(1), 29–38. http://doi.org/10.1207/s15324796abm3301_4 Rosmond, R. (2003). Stress induced disturbances of the HPA axis: a pathway to Type 2 diabetes? Medical Science Monitor : International Medical Journal of Experimental and Clinical Research, 9(2), RA35–9. Ryan, R. M., & Connell, J. P. (1989). Perceived locus of causality and internalization: examining reasons for acting in two domains. Journal of Personality and Social Psychology, 57(5), 749–761. Saelens, B. E., Sallis, J. F., Nader, P. R., Broyles, S. L., Berry, C. C., & Taras, H. L. (2002). Home environmental influences on children’s television watching from early to middle childhood. Journal of Developmental & Behavioral Pediatrics, 23(3), 127–132. Sallis, J. F., Prochaska, J. J., & Taylor, W. C. (2000). A review of correlates of physical activity of children and adolescents. Medicine and Science in Sports and Exercise, 32(5), 963–975. Salvy, S.-‐J., de la Haye, K., Bowker, J. C., & Hermans, R. C. J. (2012). Influence of peers and friends on children“s and adolescents” eating and activity behaviors. Physiology & Behavior, 106(3), 369–378. http://doi.org/10.1016/j.physbeh.2012.03.022 Samaha, F. F., Iqbal, N., Seshadri, P., Chicano, K. L., Daily, D. A., McGrory, J., et al. (2003). A low-‐ carbohydrate as compared with a low-‐fat diet in severe obesity. The New England Journal of Medicine, 348(21), 2074–2081. http://doi.org/10.1056/NEJMoa022637 Sandel, M., & Wright, R. J. (2006). When home is where the stress is: expanding the dimensions of housing that influence asthma morbidity. Archives of Disease in Childhood, 91(11), 942–948. http://doi.org/10.1136/adc.2006.098376 Schanzenbach, D. W. (2009). Do School Lunches Contribute to Childhood Obesity? J Human Resources, 44(3), 684–709. Schulze, M. B., Manson, J. E., Ludwig, D. S., Colditz, G. A., Stampfer, M. J., Willett, W. C., & Hu, F. B. (2004). Sugar-‐sweetened beverages, weight gain, and incidence of type 2 diabetes in young and middle-‐aged women. JAMA: the Journal of the American Medical Association, 292(8), 927–934. http://doi.org/10.1001/jama.292.8.927 Schwimmer, J. B., Deutsch, R., Kahen, T., Lavine, J. E., Stanley, C., & Behling, C. (2006). Prevalence of Fatty Liver in Children and Adolescents. Pediatrics, 118(4), 1388–1393. http://doi.org/10.1542/peds.2006-‐1212 Schwimmer, J. B., Deutsch, R., Rauch, J. B., Behling, C., Newbury, R., & Lavine, J. E. (2003). Obesity, insulin resistance, and other clinicopathological correlates of pediatric nonalcoholic fatty liver disease. The Journal of Pediatrics, 143(4), 500–505. http://doi.org/10.1067/S0022-‐3476(03)00325-‐1 Scott, K. A., Melhorn, S. J., & Sakai, R. R. (2012). Effects of Chronic Social Stress on Obesity. Current Obesity Reports, 1(1), 16–25. http://doi.org/10.1007/s13679-‐011-‐0006-‐3 SEARCH for Diabetes in Youth Study Group. (2006). The Burden of Diabetes Mellitus Among US Youth: Prevalence Estimates From the SEARCH for Diabetes in Youth Study. Pediatrics, 118(4), 1510–1518. http://doi.org/10.1542/peds.2006-‐0690 Shih, M., Dumke, K. A., Goran, M. I., & Simon, P. A. (2013). The association between community-‐level economic hardship and childhood obesity prevalence in Los Angeles. Pediatric Obesity, 8(6), 411–417. http://doi.org/10.1111/j.2047-‐6310.2012.00123.x 102 Shrum, W., Cheek, N. H., Jr, & Hunter, S. M. (1988). Friendship in school: Gender and racial homophily. Sociology of Education, 227–239. Singer, J. D., & Willett, J. B. (2003). Applied Longitudinal Data Analysis. Oxford Univ Press. Sisson, S. B., Church, T. S., Martin, C. K., Tudor-‐Locke, C., Smith, S. R., Bouchard, C., et al. (2009). Profiles of sedentary behavior in children and adolescents: The US National Health and Nutrition Examination Survey, 2001–2006. International Journal of Pediatric Obesity, 4(4), 353–359. http://doi.org/10.3109/17477160902934777 Slavin, R. E. (1989). School and classroom organization. Smyth, J. M., & Stone, A. A. (2003). Ecological momentary assessment research in behavioral medicine. Journal of Happiness Studies, 4(1), 35–52. Snijders, Tom AB, & Baerveldt, C. (2003). A multilevel network study of the effects of delinquent behavior on friendship evolution. Journal of Mathematical Sociology, 27(2-‐3), 123–151. http://doi.org/10.1080/00222500390213119 Snijders, Tom AB, Pattison, P. E., Robins, G. L., & Handcock, M. S. (2006). New specifications for exponential random graph models. Sociological Methodology, 36(1), 99–153. Somerset, S., & Markwell, K. (2008). Impact of a school-‐based food garden on attitudes and identification skills regarding vegetables and fruit: a 12-‐month intervention trial. Public Health Nutrition, 12(02), 214. http://doi.org/10.1017/S1368980008003327 Southern California Association of GovernmentsSCAG Aerial Land Use data description. SCAG Aerial Land Use data description. Southern California Association of Governments. Speiser, P. W., Rudolf, M. C. J., Anhalt, H., Camacho-‐Hubner, C., Chiarelli, F., Eliakim, A., et al. (2005). Childhood Obesity. The Journal of Clinical Endocrinology & Metabolism, 90(3), 1871–1887. http://doi.org/10.1210/jc.2004-‐1389 Spruijt-‐Metz, D. (2011). Etiology, Treatment, and Prevention of Obesity in Childhood and Adolescence: A Decade in Review. Journal of Research on Adolescence, 21(1), 129–152. http://doi.org/10.1111/j.1532-‐7795.2010.00719.x Spruijt-‐Metz, D., Belcher, B. R., Hsu, Y. W., McClain, A. D., Chou, C. P., Nguyen-‐Rodriguez, S., et al. (2013). Temporal Relationship Between Insulin Sensitivity and the Pubertal Decline in Physical Activity in Peripubertal Hispanic and African American Females. Diabetes Care, 36(11), 3739–3745. http://doi.org/10.2337/dc13-‐0083 Spruijt-‐Metz, D., Nguyen-‐Michel, S. T., Goran, M. I., Chou, C.-‐P., & Huang, T. T.-‐K. (2008). Reducing sedentary behavior in minority girls via a theory-‐based, tailored classroom media intervention. International Journal of Pediatric Obesity, 3(4), 240–248. http://doi.org/10.1080/17477160802113415 Spruijt-‐Metz, D., O'Reilly, G. A., Cook, L., Page, K. A., & Quinn, C. (2014). Behavioral Contributions to the Pathogenesis of Type 2 Diabetes. Current Diabetes Reports, 14(4), 475. http://doi.org/10.1007/s11892-‐014-‐0475-‐3 Stang, J., & Story, M. (2005). Adolescent growth and development. Guidelines for Adolescent Nutrition Services. Minneapolis, MN: Center for Leadership, Education and Training in Maternal and Child Nutrition, Division of Epidemiology and Community Health, School of Public Health, University of Minnesota, 1–8. Stigsdotter, U. K., Ekholm, O., Schipperijn, J., Toftager, M., Kamper-‐Jorgensen, F., & Randrup, T. B. (2010). Health promoting outdoor environments -‐ Associations between green space, and health, health-‐related quality of life and stress based on a Danish national representative survey. Scandinavian Journal of Public Health, 38(4), 411–417. http://doi.org/10.1177/1403494810367468 Stone, A. A., & Shiffman, S. (1994). Ecological momentary assessment (EMA) in behavorial 103 medicine. Annals of Behavioral Medicine : a Publication of the Society of Behavioral Medicine, 16(3), 119–202. Storch, E. A., Milsom, V. A., DeBraganza, N., Lewin, A. B., Geffken, G. R., & Silverstein, J. H. (2006). Peer Victimization, Psychosocial Adjustment, and Physical Activity in Overweight and At-‐Risk-‐For-‐Overweight Youth. Journal of Pediatric Psychology, 32(1), 80–89. http://doi.org/10.1093/jpepsy/jsj113 Story, M. (1999). School-‐based approaches for preventing and treating obesity. International Journal of Obesity and Related Metabolic Disorders : Journal of the International Association for the Study of Obesity, 23 Suppl 2, S43–51. Stults-‐Kolehmainen, M. A., & Sinha, R. (2013). The Effects of Stress on Physical Activity and Exercise. Sports Medicine, 44(1), 81–121. http://doi.org/10.1007/s40279-‐013-‐0090-‐5 Sugiyama, T., Leslie, E., Giles-‐Corti, B., & Owen, N. (2008). Associations of neighbourhood greenness with physical and mental health: do walking, social coherence and local social interaction explain the relationships? Journal of Epidemiology & Community Health, 62(5), e9–e9. http://doi.org/10.1136/jech.2007.064287 Tate, E., Wood, W., Liao, Y., & Dunton, G. (2014). Do stressed mothers have heavier children?: A meta-‐analysis on the relationship between maternal stress and child body mass index. Obesity Reviews, 1–42. Thompson, C. W., Aspinall, P., & Montarzino, A. (2007). The Childhood Factor: Adult Visits to Green Places and the Significance of Childhood Experience. Environment and Behavior, 40(1), 111–143. http://doi.org/10.1177/0013916507300119 Thompson, C. W., Roe, J., Aspinall, P., Mitchell, R., Clow, A., & Miller, D. (2012). More green space is linked to less stress in deprived communities: Evidence from salivary cortisol patterns. Landscape and Urban Planning, 105(3), 221–229. http://doi.org/10.1016/j.landurbplan.2011.12.015 Tohill, B. C., Seymour, J., Serdula, M., Kettel-‐Khan, L., & Rolls, B. J. (2004). What Epidemiologic Studies Tell Us about the Relationship between Fruit and Vegetable Consumption and Body Weight. Nutrition Reviews, 62(10), 365–374. http://doi.org/10.1301/nr.2004.oct.365-‐374 Torsheim, T., Aaroe, L. E., & Wold, B. (2003). School-‐related stress, social support, and distress: prospective analysis of reciprocal and multilevel relationships. Scandinavian Journal of Psychology, 44(2), 153–159. Trogdon, J. G., Nonnemaker, J., & Pais, J. (2008). Peer effects in adolescent overweight. Journal of Health Economics, 27(5), 1388–1399. http://doi.org/10.1016/j.jhealeco.2008.05.003 Troiano, R. P., Berrigan, D., Dodd, K. W., Mâsse, L. C., Tilert, T., & McDowell, M. (2008). Physical activity in the United States measured by accelerometer. Medicine and Science in Sports and Exercise, 40(1), 181–188. http://doi.org/10.1249/mss.0b013e31815a51b3 Troiano, R. P., Briefel, R. R., Carroll, M. D., & Bialostosky, K. (2000). Energy and fat intakes of children and adolescents in the united states: data from the national health and nutrition examination surveys. The American Journal of Clinical Nutrition, 72(5 Suppl), 1343S– 1353S. Ulrich, R. S. (1986). Human responses to vegetation and landscapes. Landscape and Urban Planning, 13, 29–44. Ulrich, R. S., Simons, R. F., Losito, B. D., Fiorito, E., Miles, M. A., & Zelson, M. (2014). Stress recovery during exposure to natural and urban environments. Journal of Environmental Psychology, 11(3), 201–230. Unger, J. B., Reynolds, K., Shakib, S., Spruijt-‐Metz, D., Sun, P., & Johnson, C. A. (2004). 104 Acculturation, physical activity, and fast-‐food consumption among Asian-‐American and Hispanic adolescents. Journal of Community Health, 29(6), 467–481. United States Department of Agriculture. My Plate, choosemyplate.gov 1–5 (pp. 1–5). Retrieved from http://choosemyplate.gov/food-‐groups/ Valente, T. W., Hoffman, B. R., Ritt-‐Olson, A., Lichtman, K., & Johnson, C. A. (2003). Effects of a social-‐network method for group assignment strategies on peer-‐led tobacco prevention programs in schools. American Journal of Public Health, 93(11), 1837–1843. Valente, T. W. (2010). Social networks and health: Models, methods, and applications. Oxford Univ Press. Valente, T. W., Fujimoto, K., Chou, C.-‐P., & Spruijt-‐Metz, D. (2009). Adolescent Affiliations and Adiposity: A Social Network Analysis of Friendships and Obesity. Journal of Adolescent Health, 45(2), 202–204. http://doi.org/10.1016/j.jadohealth.2009.01.007 Van Cauter, E., & Knutson, K. L. (2008). Sleep and the epidemic of obesity in children and adults. European Journal of Endocrinology, 159(suppl_1), S59–S66. http://doi.org/10.1530/EJE-‐08-‐0298 van den Berg, A. E., & Custers, M. H. G. (2011). Gardening Promotes Neuroendocrine and Affective Restoration from Stress. Journal of Health Psychology, 16(1), 3–11. http://doi.org/10.1177/1359105310365577 van den Berg, A. E., Hartig, T., & Staats, H. (2007). Preference for nature in urbanized societies: stress, restoration, and the pursuit of sustainability. Journal of Social Issues, 63(1), 79–96. van der Horst, K., Oenema, A., Ferreira, I., Wendel-‐Vos, W., Giskes, K., van Lenthe, F., & Brug, J. (2007). A systematic review of environmental correlates of obesity-‐related dietary behaviors in youth. Health Education Research, 22(2), 203–226. http://doi.org/10.1093/her/cyl069 van Dillen, S. M. E., de Vries, S., Groenewegen, P. P., & Spreeuwenberg, P. (2012). Greenspace in urban neighbourhoods and residents' health: adding quality to quantity. Journal of Epidemiology & Community Health, 66(6), e8–e8. http://doi.org/10.1136/jech.2009.104695 van Duyn, M. A. S., & Pivonka, E. (2000). Overview of the Health Benefits of Fruit and Vegetable Consumption for the Dietetics Professional. Journal of the American Dietetic Association, 100(12), 1511–1521. http://doi.org/10.1016/S0002-‐8223(00)00420-‐X Vartanian, L. R., Herman, C. P., & Polivy, J. (2007). Consumption stereotypes and impression management: How you are what you eat. Appetite, 48(3), 265–277. http://doi.org/10.1016/j.appet.2006.10.008 Ventura, A. K., Loken, E., Mitchell, D. C., Smiciklas-‐Wright, H., & Birch, L. L. (2006). Understanding reporting bias in the dietary recall data of 11-‐year-‐old girls. Obesity (Silver Spring, Md.), 14(6), 1073–1084. http://doi.org/10.1038/oby.2006.123 Ventura, E. E., Davis, J. N., Alexander, K. E., Shaibi, G. Q., Lee, W., Byrd-‐Williams, C. E., et al. (2008). Dietary Intake and the Metabolic Syndrome in Overweight Latino Children. Journal of the American Dietetic Association, 108(8), 1355–1359. http://doi.org/10.1016/j.jada.2008.05.006 Verheij, R. A., Maas, J., & Groenewegen, P. P. (2008). Urban-‐-‐Rural Health Differences and the Availability of Green Space. European Urban and Regional Studies, 15(4), 307–316. http://doi.org/10.1177/0969776408095107 Vgontzas, A. N., Liao, D., Pejovic, S., Calhoun, S., Karataraki, M., & Bixler, E. O. (2009). Insomnia With Objective Short Sleep Duration Is Associated With Type 2 Diabetes: A population-‐ based study. Diabetes Care, 32(11), 1980–1985. http://doi.org/10.2337/dc09-‐0284 105 Voorhees, C. C., Murray, D., Welk, G., Birnbaum, A., Ribisl, K. M., Johnson, C. C., et al. (2005). The role of peer social network factors and physical activity in adolescent girls. American Journal of Health Behavior, 29(2), 183–190. Wang, M. C., Rauzon, S., Studer, N., Martin, A. C., Craig, L., Merlo, C., et al. (2010). Exposure to a Comprehensive School Intervention Increases Vegetable Consumption. Journal of Adolescent Health, 47(1), 74–82. http://doi.org/10.1016/j.jadohealth.2009.12.014 Wang, Y. (2001). Cross-‐national comparison of childhood obesity: the epidemic and the relationship between obesity and socioeconomic status. International Journal of Epidemiology, 30(5), 1129–1136. Waters, E., de Silva-‐Sanigorski, A., Burford, B. J., Brown, T., Campbell, K. J., Gao, Y., et al. (2011). Interventions for preventing obesity in children (Review). Chichester, UK: The Cochrane Library. http://doi.org/10.1002/14651858.CD001871.pub3 Weickert, M. O., & Pfeiffer, A. F. H. (2008). Metabolic effects of dietary fiber consumption and prevention of diabetes. The Journal of Nutrition, 138(3), 439–442. Willett, W. C., Sampson, L., Stampfer, M. J., Rosner, B., Bain, C., Witschi, J., et al. (1985). Reproducibility and validity of a semiquantitative food frequency questionnaire. American Journal of Epidemiology, 122(1), 51–65. Willi, C., Bodenmann, P., Ghali, W. A., Faris, P. D., & Cornuz, J. (2007). Active smoking and the risk of type 2 diabetes: a systematic review and meta-‐analysis. JAMA: the Journal of the American Medical Association, 298(22), 2654–2664. Williams, G. C., Grow, V. M., Freedman, Z. R., Ryan, R. M., & Deci, E. L. (1996). Motivational predictors of weight loss and weight-‐loss maintenance. Journal of Personality and Social Psychology, 70(1), 115–126. Williamson, G. (2012). Possible effects of dietary polyphenols on sugar absorption and digestion. Molecular Nutrition & Food Research, 57(1), 48–57. http://doi.org/10.1002/mnfr.201200511 Wouters, E. J., Larsen, J. K., Kremers, S. P., Dagnelie, P. C., & Geenen, R. (2010). Peer influence on snacking behavior in adolescence. Appetite, 55(1), 11–17. http://doi.org/10.1016/j.appet.2010.03.002 Wright, W., & Rowell, L. (2010). Examining the effect of gardening on vegetable consumption among youth in kindergarten through fifth grade. WMJ : Official Publication of the State Medical Society of Wisconsin, 109(3), 125–129. Yang, F., Bao, Z. Y., & Zhu, Z. J. (2011). An Assessment of Psychological Noise Reduction by Landscape Plants. Int. J. Environ. Res. Public Health, 8(12), 1032–1048. http://doi.org/10.3390/ijerph8041032 Zenzen, W., & Kridli, S. (2009). Integrative Reviewof School-‐based Childhood Obesity Prevention Programs. Journal of Pediatric Health Care, 23(4), 242–258. http://doi.org/10.1016/j.pedhc.2008.04.008 Zimring, C. (1982). The built environment as a source of psychological stress: Impacts of buildings and cities on satisfaction and behavior. In Environmental Stress (pp. 151–198). New York, NY: Cambridge University Press.

Abstract (if available)

Abstract The high rates of childhood obesity in the US make it imperative that factors contributing to the onset of obesity are better understood. The goal of this dissertation is to better understand how the environment contributes to obesity-related behaviors at the social, structured learning, and physical levels. The objectives for this dissertation are the following: 1) better understand similarities among adolescents and their friends in intake of fruit, vegetables (FV) and soda, 2) evaluate the impact of a school-based nutrition and gardening intervention on FV intake as a mediator of dietary determinants and BMI, and 3) determine the impact of green space on perceived stress among youth in their everyday living environments. ❧ Study 1 utilized cross-sectional data from 617 middle school students in 17 classrooms in the Get Moving study. Social networks were constructed by asking students to name their five best friends in their classroom, and dietary frequency of FV and soda were obtained. Exponential random graph models were used to measure homophily on dietary intake among friends. Study 2 utilized data from the LA Sprouts study, a randomized, controlled, school-based gardening and nutrition intervention with 363 elementary school students. At baseline and 12-weeks, BMI was measured and participants completed questionnaires on dietary determinants and behavior. Path modeling was used to examine co-occurring changes in these measures. Study 3 utilized data on 63 3rd-8th grade students who participated in the Mobile Healthy PLACES study. Children completed Ecological Momentary Assessment (EMA) questions on stress, and GPS-derived variables Open Recreational Space (ORS) and Normalized Difference Vegetation Index (NDVI) measured green space. Analyses to examine the effect of green space on stress were completed using multilevel modeling and partial correlations. ❧ In Study 1, there was no significant association between friends and their intake of FV, or soda (p>0.9). In Study 2, change in controlled motivation to eat FV, willingness to try FV, and vegetable preferences was associated with change in vegetable intake (0.21 ± 0.6, p<0.001

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Motivation and the meanings of health behavior as factors associated with eating behavior in Latino youth

PDF

Psychosocial and behavioral ractors associated with emotional eating in adolescents

PDF

Objectively measured physical activity and related factors in minority youth

PDF

The vicious cycle of inactivity, obesity, and metabolic health consequences in at-risk pediatric populations

PDF

Effects of parent stress on weight-related parenting practices and child obesity risk

PDF

Investigating factors that influence peer relationships and obesity during middle childhood

PDF

Effects of sugar and fiber consumption in minority adolescents and self-tracking as a potential dietary intervention tool

PDF

A network analysis of online and offline social influence processes in relation to adolescent smoking and alcohol use

PDF

System dynamics of in-home family eating behavior: insights from intensive longitudinal data using Ecological Momentary Assessment and wearable sensors

PDF

Adolescent cigarette use and psychological distress: relation to adult obesity risk

PDF

Effects of near-roadway air pollution exposure on obesity, obesity-related behavior, and neurobehavioral deficits during peripuberty

PDF

The acute relationship between affective states and physiological stress response, and the moderating role of moderate-to-vigorous physical activity

PDF

Mixed methods investigation of user engagement with a smoking cessation app

PDF

Cultural risk and protective factors for tobacco use behaviors and depressive symptoms among American Indian adolescents in California

PDF

Applications of mindfulness toward promoting psychosocial health in diverse communities

PDF

Social networks, neighborhood environments, dietary patterns and obesity risk among adults in Los Angeles County

PDF

Childhood cancer survivorship: parental factors associated with survivor's follow-up care behavior and mental health

PDF

Understanding the dynamic relationships between physical activity and affective states using real-time data capture techniques

PDF

Marginalization in acculturation is related to objectively measured physical activity in Latina adolecents

PDF

Investigating daily effects of activity space-based built environment exposures on physical activity behaviors in Hispanic women during pregnancy and early postpartum