Back

Decoupled but reconcilable: a nutrition-feasible global reallocation lowers dietary inflammation and environmental footprints simultaneously in every country

Huang, S.; Wang, X.

2026-09-01 nutrition
10.64898/2026.08.30.26361755 medRxiv
Show abstract

Background: Pro-inflammatory and high-environmental-impact diets both threaten population and planetary health, but whether the two objectives align or conflict across countries is unresolved. We tested whether a supply-based dietary inflammatory index (sDII) is coupled to greenhouse-gas (GHG), land and freshwater footprints, and whether a nutrition-feasible reallocation can lower both simultaneously. Methods: From FAO Food Balance Sheets we built sDII (12 inflammatory-weighted components; construct validity r=0.9999) and five per-capita footprints using three independent life-cycle inventories for 182 national food-supply series. For each country, constrained optimisation reallocated 13 food-group supplies under isoenergetic, protein-preserving and food-group-bound constraints, minimising sDII and GHG jointly (Pareto frontier). Health burden was estimated via pooled relative-risk meta-analysis and 2023 World Bank population data. Results: sDII was only weakly associated with GHG (Spearman rho=0.14), land (rho=0.13) and freshwater (rho=0.28) in 2023. The balanced-Pareto reallocation lowered both sDII and GHG in 182/182 series (100% synergy): population-weighted delta sDII=-0.235, GHG -37.5%, land -49.3%, water -17.2%, i.e. 4.28 Gt CO2e/yr avoided. The associated reduction in metabolic-syndrome burden was directionally consistent but modest (~1.1% of the prevalent pool, ~2.84 million cases). Results were robust to three life-cycle inventories and three feasibility-bound regimes. Conclusions: Anti-inflammatory and low-carbon goals are decoupled rather than conflicting, and an isoenergetic, protein-preserving reallocation reconciles them in every country. Environmental gains are large and robust; health gains are directionally consistent but modest--triangulation, not a causal claim.

Matching journals

The top 3 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.