Evaluating OzHarvest’s primary-school Food Education and Sustainability Training (FEAST) program in 10–12-year-old children in Australia: protocol for a pragmatic cluster non-randomized controlled trial

Background The promotion of healthy eating is a public health priority. Poor dietary behaviours, including low fruit and vegetable (F&V) consumption are of particular concern among children. Novel nutrition promotion strategies are needed to improve F&V consumption. Sustainability education could be used to support nutrition education within the school context. The purpose of this paper is to report the protocol for impact and process evaluation of the school-based Food Education and Sustainability Training (FEAST) program, designed to educate children about sustainability, food waste and nutrition, using hands-on cooking activities. Methods A pragmatic, parallel, cluster non-randomized controlled trial with pre- and post-measures, will be implemented among 20 primary schools (10 intervention vs 10 wait-list-control) within NSW, Australia, involving children in Grades 5–6. FEAST is a curriculum-aligned program, delivered as a 1.5-h lesson/week, for a 10-week unit of inquiry, incorporating theory and cooking. FEAST was developed using theoretical frameworks which included Social Cognitive Theory and the Precede-Proceed Planning model. Primary outcomes include children’s self-reported F&V intakes (serves/day). Food literacy constructs such as: nutrition knowledge, food preparation and cooking skills, self-efficacy and behaviours, food waste knowledge and behaviours and food production knowledge, will be assessed as secondary outcomes. Process evaluation will assess program reach, adoption, implementation, maintenance, satisfaction and perceived benefits by teachers and students. An online survey (including quantitative and qualitative questions) was developed for administration at baseline (impact evaluation) and immediately post-intervention (impact and process evaluation). Intervention effects on quantitative study outcomes will be estimated with ​generalised linear mixed models, including random effects and will follow the intention-to-treat