Adolescent Social Networks, Physical Activity and Other Health Behaviors John R Sirard 10.20.2011 Overview 1. Background 2. Social factors related to physical activity (PA) 3. Social Networks, PA and Screen Time (crosssectional results) 4. Future Directions (Social Network Analysis and Health – NIH FOA) 1. Background • Physical Activity: Any bodily movement produced by the contraction of skeletal muscle that increases energy expenditure above a basal level. – Not just sports and planned exercise – A behavior (not fitness or other physiological outcomes) % meeting PA recommendation: > 60 min of moderatevigorous physical activity per day % meeting recommendation 30 25 20 15 Males Females 10 5 0 9th 10th 11th Grade 12th PA Prevalence - A New Challenge (Troiano et al. 2008) • 2009 YRBSS : 18.4% (Males: 24.8%; Females: 11.4%) • Measured by accelerometer… • % meeting PA recommendation: > 60 min of moderatevigorous physical activity per day PA and Health in Youth • • • • Lipid profile Blood Pressure Overweight and obesity Metabolic Syndrome • CVD risk factors • Bone health • Depression, anxiety, mood, self-esteem • QOL and fatigue of cancer survivors PA and Other Health Behaviors Trost SG. Physical Education, Physical Activity and Academic Performance. Fall 2007 Research Brief. Active Living Research (RWJF) • “Students whose time in PE or school-based physical activity was increased maintained or improved their grades and scores on standardized achievement tests, even though they received less classroom instructional time than students in control groups”. 2. Social Factors Related to PA Attributes of Self Genetics. Self Efficacy impacts motivation and attempts to alter, modify, and maintain health behaviors. Policy Systemic & Policy Factors Unhealthy foods easily accessible, heavily marketed to children Healthy foods are more costly and difficult to obtain. Urban sprawl, lack of pedestrian infrastructure Diet Unhealthy dietary behaviors; low intake of fruits and vegetables Physical Activity Lack of PA and excess sedentary activities Core Relationships Family, peers, authority figures offering different types of support for healthy behaviors Intra-personal & Behaviors Inter-personal Environment Extreme climates, walking or biking trails, public gyms and pools, public transportation, School/work environments Behavior settings Technology Labor saving devices, automobiles, communication technology Cultural, Religious & Family Values Reasons for eating; the quantity, quality, and types of foods eaten; gender roles Social Factors and PA • Social Support for Physical Activity – Emotional Support (encourages you to do physical activity) – Informational Support – Tangible Support (do physical activities with you, pay for equipment / fees) • Perceived support for PA from family and friends positively associated with adolescent PA – typically Social Networks Social Networks and Obesity • Christakis and Fowler (NEJM 2007) – Framingham Cohort Study with 30 years of followup • “…obesity appears to spread through social ties”. – National Longitudinal Study of Adolescent Health • Strauss 2003, Trogdon 2008 • Challenges to analysis and interpretation (Lyons 2011) 3. Social Networks, PA and Screen Time (cross-sectional results) Project EAT 2010 • Eating and Activity among Teens – New cohort of 2,793 middle and high school students from 20 Mpls/St. Paul middle schools and high schools (2009-2010) – Weight-related behaviors (PA, screen time, diet, weight control practices, disordered eating) and – Related multi-level correlates (individual, peer, family, school, neighborhood) – Main “outcome”: weight status Social Networks and PA in Project EAT 2010 • Purpose: To examine how friends’ PA and screen time is related to an individual adolescent’s PA and screen time by using data from nominated friends Best Male friend Male friend 2 Male friend 3 • Hyp: Associations between ego and friend PA and screen time would be strongest between same gender and weakest for opposite gender associations. • Hyp: Associations between ego and friend behaviors would be stronger for high school versus middle school students. Best Female friend Female friend 2 Female friend 3 Example of network connections at one middle school. Triangles indicate Girls, Circles indicate Boys, the numbers in the symbol represent the grade level. How many friends are enough? • Sensitivity Analyses 1. Data from all 6 nominated friends (n=251) • From smaller, lower income, more ethnically diverse schools – not representative of larger Project EAT sample 2. Data from all but one nominated friend (n=585) • Also not representative of larger sample 3. Data from at least 1/3 of nominated friends (n=1655) 4. Data from at least one nominated friend (n=2126) Table 1: Sample characteristics for n= 2126 Egos with any friends in dataset. Values are mean+SD or n (percent) Age (years) Grade Middle School (n=1114) High School (n=1012) Ethnicity group White (n=426) African American/Black (n=567) Latino/Hispanic (n=373) Asian American (n=409) Native American (n=87) Mixed/Other (n=264) Socioeconomic Status Missing (n= 64) Low (n=611) Low-middle(n=526) Middle(n=717) Upper-Middle + High (n=208) Males (n=983) 14.29±2.01 Females (n=1143) p-value 14.11±1.88 0.03 504 (51.3) 479 (48.7) 610 (53.4) 533 (46.6) 0.34 224 (22.8) 260 (26.5) 164 (16.7) 195 (19.8) 40 (4.1) 100 (10.2) 202 (17.7) 307 (26.9) 209 (18.3) 214 (18.7) 47 (4.1) 164 (14.4) 0.01 31 (3.2) 286 (29.1) 249 (25.3) 320 (32.6) 33 (2.9) 325 (28.4) 277 (24.2) 397 (34.7) 97 (9.8) 111 (9.7) 0.61 Table 1 (cont’d): Sample characteristics for n= 2126 Egos with any friends in dataset. Values are mean+SD or n (percent) Males (n=983) Weight Status * Not overweight (n=1272) Overweight (n=854) Physical Activity (hrs/wk) Screen Time (hrs/wk) Females (n=1143) p-values 566 (57.6) 417 (42.4) 706 (61.8) 437 (38.2) 0.67 6.83±4.84 45.15±28.85 4.98±4.40 36.51±23.92 <.001 <.001 Number of friends 2.64+sd 2.62+sd 0.XX Number of Male friends 1.49+sd 1.01+sd 0.XX Number of Female friends 1.15+sd 1.61+sd 0.XX Overweight status is defined as > 85th percentile on the Centers for Disease Control and Prevention’s age and gender specific Body Mass Index curves BMI = kg / m2 Table 2. Estimated change in male’s physical activity or screen time for every 1 hour increase in friends’ mean (or best friend’s) physical activity or screen time Male Ego Physical Activity p-value Screen Time p-value Male Friends Female Friends Male Best Friend Female Best Friend n=838 n=668 n=483 n=389 0.059 (.037) 0.114 (.042) 0.111 0.007 * ? 0.061 (.042) 0.034 (.045) 0.149 0.448 0.024 (.042) 0.111 (.054) -0.025 (.043) 0.110 (.072) 0.559 0.038 0.558 0.129 Adjusted for Ego’s age, SES, race/ethnicity, BMI, and number of friends available * Significant interaction with school level (middle vs. high school) Table 3. Estimated change in female’s physical activity or screen time for every 1 hour increase in friends’ mean (or best friend’s) physical activity or screen time Female Ego Physical Activity p-value Screen Time p-value Male Friends n=723 Female Friends n=1007 0.122 (.025) .182 (.0439) Male Best Friend n=404 Female Best Friend n=612 .086 (.026) .128 (.052) <.0001 <.0001 0.0009 0.0142 * .097 (.039) .091 (.043) -.010 (.032) .022 (.056) 0.012 0.032 0.75 0.696 Adjusted for Ego’s age, SES, race/ethnicity, BMI, and number of friends available * Significant interaction with school level (middle vs. high school) Strengths and Limitations • Strengths – Large, diverse sample – Novel method and analyses • Limitations – Cross-sectional – Incomplete network data Conclusions • Several associations between adolescent and friend PA and screen time • Consistent associations for female PA support previous research on youth PA and sports participation • Limited associations between Male’s PA and screen time vs. friends’ PA and screen time 4. Future Directions NIH Funding Opportunity Announcement • PAR-10-145, “Social Network Analysis and Health” • Mechanism: R01 • Due: May 11, 2012 • Start: April 2013 Overall Research Goal • Do changes in friend networks from 8th through 10th grade predict changes in weight-related behaviors of individuals as well as sub-networks (cliques) within a school? – Longitudinal, multiple time points – More complete network data • Include multi-level predictors of PA and screen time as mediators of the hypothesized behavior changes – individual-level psychosocial factors, family/home environment, school, social and physical environments of neighborhood. Possible Research Questions • How are these weight-related behaviors transmitted within schoolbased peer networks? – If an active person in 8th grade becomes friends with a group of inactive individuals in 9th grade, does this individual become less active themselves or do they maintain their activity level? – Does the network become more like the new friend or is there no effect? – Why? – What role do PA, sport, TV viewing, video games play in selecting new friend(s) – What are the characteristics of the individual and those in the network and how do those factors mediate behavior change of the individual or the network? poorly written Specific Aims 1. To determine if changes in an individual’s social network (to a network with different weight-related behaviors) results in changes in the individual’s weight-related behavior that is similar to the network’s behavior 2. To determine if changes in the weight-related behaviors of an individual(s) within a clique influences the behaviors of others in the clique. 3. To identify the… A. B. Individual and network characteristics that promote, maintain, or increase PA within the network Individual and network characteristics that promote, maintain, or increase screen time within the network Additional Research Questions (School-level Analysis) • Does the “spread” of student athletes within non-sports clubs/organizations and across sub-networks (cliques) affect the overall physical activity level of the school? • Collect rosters for school-based sports teams and selected formal non-sports clubs (e.g., student government, newspaper, yearbook, band). – Hyp: School-level physical activity will be greater in schools with a greater “spread” of student athletes within non-sport groups/clubs, and across cliques due to the added exposure of other students to those student athletes, compared to schools where student athletes stay within limited insular cliques.