Mechanisms of scaling up: combining a realist perspective and systems analysis to understand successfully scaled interventions

Sustainable shifts in population behaviours require system-level implementation and embeddedness of large-scale health interventions. This paper aims to understand how different contexts of scaling up interventions affect mechanisms to produce intended and unintended scale up outcomes. A mixed method study combining a realist perspective and systems analysis (causal loop diagrams) of scaled-up physical activity and/or nutrition interventions implemented at a state/national level in Australia (2010-18). The study involved four distinct phases: Phase 1 expert consultation, database and grey literature searches to identify scaled-up interventions; Phase 2 generating initial Context-Mechanism-Outcome configurations (CMOs) from the WHO ExpandNet framework for scaling up; Phase 3 testing and refining CMOs via online surveys and realist interviews with academics, government and non-government organisations (NGOs) involved in scale up of selected interventions (Phase 1); and Phase 4 generating cross-case mid-range theories represented in systems models of scaling up; validated by member checking. Descriptive statistics were reported for online survey data and realist analysis for interview data. Seven interventions were analysed, targeting nutrition (n = 1), physical activity (n = 1), or a combination (n = 5). Twenty-six participants completed surveys; 19 completed interviews. Sixty-three CMO pathways underpinned successful scale up, reflecting 36 scale up contexts, 8 key outcomes; linked via 53 commonly occurring mechanisms. All five WHO framework domains were represented in the systems models. Most CMO pathways included 'intervention attributes' and led to outcomes 'community sustainability/embeddedness' and 'stakeholder buy-in/perceived value'. Irrespective of interventions being scaled in similar contexts (e.g., having political favourability); mechanisms still led to both intended and unintended scale up outcomes (e.g., increased or reduced sustainability). This paper provides the first evidence for mechanisms underpinning outcomes required for successful scale up of state or nationally delivered interventions. Our findings challenge current prerequisites for effective scaling suggesting other conditions may be necessary. Future scale up approaches that plan for complexity and encourage iterative adaptation throughout, may enhance scale up outcomes. Current linear, context-to-outcome depictions of scale up oversimplify what is a clearly a complex interaction