Nutrition and Enjoyable Activity for Teen Girls (NEAT Girls) group randomized controlled trial: evaluation of a school-based obesity prevention program for adolescent girls from low-income communities

Research Doctorate - Doctor of Philosophy (PhD)Current rates of obesity among youth are high and prevalence continues to increase most rapidly among those of low socio-economic position (SEP). Overweight and obesity are a major cause of preventable morbidity and mortality, and given the high likelihood of paediatric obesity persisting into adulthood, effective behavioural interventions that target high-risk groups are urgently needed. Adolescent girls of low-SEP require priority attention because they are particularly susceptible to demonstrating poor health behaviours implicated in the development of obesity. The primary aim of this thesis was to evaluate a 12-month, school-based obesity prevention intervention targeting adolescent girls living in low-income communities. The Nutrition and Enjoyable Activity for Teen Girls (NEAT Girls) intervention was evaluated using a group randomised controlled trial (RCT) that involved 357 adolescent girls (M = 13.2 years, SD = 0.5) from six intervention and six control schools located in the Newcastle, Central Coast and Hunter areas of New South Wales, Australia. The multi-component intervention was guided by Social Cognitive Theory (SCT) and focused on promoting physical activity, healthy eating and reducing sedentary behaviours. The program components included enhanced school sport and lunchtime physical activity sessions, nutrition workshops, interactive educational seminars, student handbooks, pedometers for self-monitoring, text messages to encourage the targeted health behaviours, and newsletters for parents. Assessments were conducted at baseline, 12- (post-test) and 24- months (follow-up). The primary outcome measure was body mass index (BMI), and secondary outcomes included BMI z-score, percentage body fat (bioelectrical impedance analysis), physical activity (accelerometers), dietary intake and sedentary behaviour (self-report), muscular fitness (sit-up and prone hold tests), physical self-perceptions, self-esteem, and hypothesised social-cognitive mediators of physical activity and dietary behaviour change (questionnaires). After 12- and 24- months, relative to the control group, the intervention group demonstrated improvements in all body composition outcomes, although there was no statistically significant effect on the primary outcome (BMI). Only changes in percentage body fat were statistically significant after 24-month follow-up (adjusted mean difference [95% confidence intervals] = –2.0% [–3.0 to –0.9], p = .006). After 12-months, girls in the intervention group reported 31 mins/day less [–62 to –1], p = .024) screen time than their control group peers. There were no significant group-by-time effects for any of the remaining outcomes. These findings provide some evidence for the benefit of a school-based intervention to prevent unhealthy weight gain in adolescent girls living in low-income communities. Considering the range of adverse health consequences associated with excessive time spent in sedentary behaviours, the large reductions in screen time observed in the current study may have important implications for adolescent girls. Secondary Aims: few studies have examined the psychosocial mediators of physical activity and dietary behaviour change in adolescents and the poor quality of existing measures has compounded the lack of available evidence. Therefore, a secondary aim of this thesis was to develop and evaluate social-cognitive measures related to adolescent physical activity and dietary behaviours. Two questionnaires, each relating to physical activity or dietary behaviour, based on the following constructs from SCT were developed: self-efficacy, intention (proximal goals), situation (perceived barriers/facilitators of the physical environment), social support, behavioural strategies, and outcome expectations and expectancies. In the first stage of development, the scales were reviewed by an expert panel and then further refined during a focus group conducted with adolescents (n = 12, 14.1 ± 0.6 years). In the second stage, the scales were pilot tested in a sample of secondary school students (n = 173, 13.7 ± 1.2 years) that was independent of the NEAT Girls study sample. Reliability of the measures was assessed using a two-week test-retest (intra-class correlation coefficients [ICC] and internal consistency [Cronbach’s alpha]). Construct validity was also assessed by confirmatory factor analyses (CFA) to establish model fit of each scale by using multiple fit indices. All scales demonstrated sound reliability (ICC = 0.81 to 0.91;  = 0.63 to 0.79) and fit indices indicated each model to be an adequate-to-exact fit to the data. As such, these scales were used to measure hypothesised social-cognitive determinants of physical activity and dietary behaviour in the NEAT Girls study, and may provide suitable use for future research investigating social-cognitive correlates and mediators of physical activity and dietary behaviour, and the evaluation of theoretical models based on the SCT. An additional secondary aim involved performing a longitudinal test of the hypothesised paths of influence in Bandura’s SCT model to explain change in physical activity behaviour (moderate-to-vigorous physical activity [MVPA] minutes using accelerometry) following the NEAT Girls 12-month intervention. In the model, it was hypothesised that self-efficacy, outcome expectations and intention (proximal goals) related to physical activity would predict change in physical activity, and that self-efficacy, outcome expectations and parental support for being physically active would predict physical activity intention. Structural equation modelling was used to test the model. Although model-fit indices indicated the model was an adequate fit to the data, the amount of variance explained for physical activity (28%) and intention (34%) was small. Further, the proposed pathways linking the social-cognitive influences to behaviour change were not well supported. Only self-efficacy predicted change in physical activity. Support for the SCT model to predict change in physical activity behaviour was not strong. Future model testing should consider augmentation or integration of theory, including ecological components, to advance our understanding of health behaviours in adolescent girls of low-SEP