How to design a complex behaviour change intervention: experiences from a nutrition-sensitive agriculture trial in rural India.

Many public health interventions aim to promote healthful behaviours, with varying degrees of success. With a lack of existing empirical evidence on the optimal number or combination of behaviours to promote to achieve a given health outcome, a key challenge in intervention design lies in deciding what behaviours to prioritise, and how best to promote them. We describe how key behaviours were selected and promoted within a multisectoral nutrition-sensitive agriculture intervention that aimed to address maternal and child undernutrition in rural India. First, we formulated a Theory of Change, which outlined our hypothesised impact pathways. To do this, we used the following inputs: existing conceptual frameworks, published empirical evidence, a feasibility study, formative research and the intervention team's local knowledge. Then, we selected specific behaviours to address within each impact pathway, based on our formative research, behaviour change models, local knowledge and community feedback. As the intervention progressed, we mapped each of the behaviours against our impact pathways and the transtheoretical model of behaviour change, to monitor the balance of behaviours across pathways and along stages of behaviour change. By collectively agreeing on definitions of complex concepts and hypothesised impact pathways, implementing partners were able to communicate clearly between each other and with intervention participants. Our intervention was iteratively informed by continuous review, by monitoring implementation against targets and by integrating community feedback. Impact and process evaluations will reveal whether these approaches are effective for improving maternal and child nutrition, and what the effects are on each hypothesised impact pathway.

Upscaling Participatory Action and Videos for Agriculture and Nutrition (UPAVAN) trial comparing three variants of a nutrition-sensitive agricultural extension intervention to improve maternal and child nutritional outcomes in rural Odisha, India: study protocol for a cluster randomised controlled trial.

BACKGROUND: Maternal and child undernutrition have adverse consequences for pregnancy outcomes and child morbidity and mortality, and they are associated with low educational attainment, economic productivity as an adult, and human wellbeing. 'Nutrition-sensitive' agriculture programs could tackle the underlying causes of undernutrition. METHODS/DESIGN: This study is a four-arm cluster randomised controlled trial in Odisha, India. Interventions are as follows: (1) an agricultural extension platform of women's groups viewing and discussing videos on nutrition-sensitive agriculture (NSA) practices, and follow-up visits to women at home to encourage the adoption of new practices shown in the videos; (2) women's groups viewing and discussing videos on NSA and nutrition-specific practices, with follow-up visits; and (3) women's groups viewing and discussing videos on NSA and nutrition-specific practices combined with a cycle of Participatory Learning and Action meetings, with follow-up visits. All arms, including the control, receive basic nutrition training from government community frontline workers. Primary outcomes, assessed at baseline and 32 months after the start of the interventions, are (1) percentage of children aged 6-23 months consuming ≥ 4 out of 7 food groups per day and (2) mean body mass index (BMI) (kg/m2) of non-pregnant, non-postpartum (gave birth > 42 days ago) mothers or female primary caregivers of children aged 0-23 months. Secondary outcomes are percentage of mothers consuming ≥ 5 out of 10 food groups per day and percentage of children's weight-for-height z-score < -2 standard deviations (SD). The unit of randomisation is a cluster, defined as one or more villages with a combined minimum population of 800 residents. There are 37 clusters per arm, and outcomes will be assessed in an average of 32 eligible households per cluster. For randomisation, clusters are stratified by distance to nearest town (< 10 km or ≥ 10 km), and low (< 30%), medium (30-70%), or high (> 70%) proportion of Scheduled Tribe or Scheduled Caste (disadvantaged) households. A process evaluation will assess the quality of implementation and mechanisms behind the intervention effects. A cost-consequence analysis will compare incremental costs and outcomes of the interventions. DISCUSSION: This trial will contribute evidence on the impacts of NSA extension through participatory, low-cost, video-based approaches on maternal and child nutrition and on whether integration with nutrition-specific goals and enhanced participatory approaches can increase these impacts. TRIAL REGISTRATION: ISRCTN , ISRCTN65922679 . Registered on 21 December 2016.