Advancing evidence-informed, multi-criteria food selection in nutrition sensitive programme design: Methods and tools for decision-support

Malnutrition affects one in three people globally, with low-quality diets underpinning its multiple forms. Many nutrition-sensitive programmes (NSPs) aim to improve diet quality by strengthening the supply, access, and demand for nutrient-rich foods through actions across food, agriculture, and social protection sectors. Designing such programmes can require choosing which foods to produce, distribute, or promote, using multiple, sometimes competing, criteria related to nutrition, feasibility, acceptability, livelihoods, and sustainability. However, food selection decision processes and the use of analytical tools to support them remain poorly documented. This thesis aimed to understand and respond to the need for simple, practical, and context-appropriate decision-support for food selection through five interlinked studies. First, mixed-methods research explored decision processes during NSP design to identify food selection criteria and stakeholder priorities. Second, a scoping review mapped existing tools to assess their methods, data requirements and capacity to incorporate multisectoral decision criteria. Third, a qualitative study examined how nutrition modelling tools influence policy and programme decisions and the conditions under which analytical evidence is taken up in practice. Fourth, a diet modelling analysis tested whether household level consumption data could be used as a pragmatic alternative to individual dietary data to expand analytical feasibility in low-resource settings. Finally, a multi-criteria decision analysis (MCDA) tool integrating diet modelling with stakeholder-weighted criteria was developed and demonstrated through case studies in Mozambique, Indonesia, and Malawi. Food selection emerged as a critical, but inconsistently supported NSP design decision shaped by diverse and sometimes conflicting sectoral priorities. Existing tools reflect their originating sectors and were often too narrow in scope, data-intensive or technically complex to support cross-sectoral decision making. Uptake of modelling tools and their outputs depends less on technical sophistication than transparency, local ownership, and institutional capacity. Dietary data scarcity barriers to the use of diet modelling approaches could be overcome by transforming existing household-level consumption data. The final MCDA tool offers a rapid and adaptable approach for identifying trade-offs and supporting participatory and evidence-informed food selection. Together, this work advances understanding of how food selection decisions are made in nutrition sensitive programming and provides a practical framework for integrating multisectoral priorities, accessible data, and stakeholder engagement into decision support, strengthening the foundations for more collaborative and effective action to improve diets.