International food trade contributes to dietary risks and mortality at global, regional and national levels

We adopt a demand-driven attributional perspective to link dietary risks to international food trade, enabling estimation of how imports contribute to the dietary health of importing countries and the tracing of responsibility for those impacts to exporting countries. Using bilateral trade data and an algorithm that links food consumption with primary production, we track the contribution of a country’s food exports to consumption in another country. We then apply established food-specific risk–disease relationships together with mortality rates and population numbers to estimate the proportion of diet-related diseases and mortality attributable to traded foods. The contribution of food trade to a country’s diet is understood as the portion of dietary intake derived from imported foods, adjusted for household-level food waste, with exports excluded from national consumption. This analysis differs from economic scenario modeling of trade liberalization or bans by focusing not on market feedbacks or counterfactual scenarios, but on attributing observed increases or decreases in dietary risks and diet-related diseases to imported foods. The approach parallels input–output assessments of environmental impacts embodied in trade by quantifying dietary risks and associated health impacts embodied in food trade. The aim is to highlight the importance of considering food trade’s role in dietary health regardless of specific policy measures, especially in the context of potential disruptions such as Brexit, climate-related disasters, and armed conflicts like the Russia–Ukraine war.

The paper situates its contribution within prior work that assessed correlations between broad health markers (for example, BMI, life expectancy) and trade openness or specific trade agreements, and studies attributing international trade’s role in calorie and nutrient distribution. It notes a gap in directly linking traded foods to specific dietary risks and associated disease outcomes. The discussion also relates the study to literature on environmental and social footprints embodied in trade, emphasizing that unlike environmental impacts, which affect exporting regions, the embodied health impacts primarily affect importing countries.

Data sources: The study used FAO bilateral trade data (FAOSTAT) compiled under International Merchandise Trade Statistics standards. To correct source-country attribution (which reflects the last value-added country), the authors applied a balancing algorithm based on input–output accounting that integrates primary production data, bilateral trade of primary and secondary products, and conversion factors (calorie-based and extraction rates) to express secondary products in primary equivalents. Food categories: Commodity-level data were aggregated into health-relevant groups: fruits, vegetables, legumes, nuts, and red meat (beef, lamb, goat, pork). Exposure estimation: Traded quantities were converted to per capita changes in consumption by dividing by country population and subtracting household-level food waste proportions. Exports were excluded from national consumption. Health impact assessment: A comparative risk assessment linked five dietary risks (low intake of fruits, vegetables, legumes, nuts; high intake of red meat) to four disease endpoints (coronary heart disease, stroke, colorectal cancer, and type 2 diabetes). Relative risks were taken from meta-analyses of prospective cohort studies. Mortality and population by age and country were sourced from the Global Burden of Disease project. Evidence certainty followed Bradford Hill-inspired criteria, World Cancer Research Fund assessments, and NutriGrade scoring (moderate to high certainty). Population and uncertainty: Analyses focused on adults aged 20+ years, with relative risks attenuated with age based on pooled analyses of metabolic risk factors. Uncertainty intervals were computed using 95% confidence intervals of relative risks and propagated using standard error methods. Reporting followed GATHER guidelines.

Trade volumes and diet contributions (2019): - Over 190 Mt of foods linked to dietary risks were traded internationally, representing 3–12% of their production: fruits 86 Mt (11%), vegetables 58 Mt (5%), red meat 25 Mt (11%), legumes 12 Mt (3%), nuts 8 Mt (12%). Major exporters included the Americas for fruits/legumes/nuts (notably Brazil, Argentina), Asia for vegetables (notably China), and Europe for red meat (notably Germany). - Average per-person daily intake supplied by imports across countries: fruits 31 g/d (14% of demand), vegetables 21 g/d (5%), red meat 9 g/d (11%), legumes 4 g/d (19%), nuts 3 g/d (21%). Regional reliance on imports ranged widely (for example, fruits: Africa 4 g/d (2%) vs Europe 145 g/d (64%)). Global health impacts: - Net change in diet-related mortality attributable to trade: −1.2 million deaths (95% CI 0.8–1.7 million). About 53% of avoided deaths were from coronary heart disease, 25% from stroke, 23% from cancer. - Positive contributions from imports: fruits −597,000; vegetables −380,000; nuts −300,000; legumes −98,000 deaths. - Negative contribution from imports of red meat: +147,000 deaths. - From the abstract/discussion: imports of fruits, vegetables, legumes, and nuts together associated with −1.4 million deaths globally; red meat imports associated with an increase of about +150,000 deaths. Regional patterns: - Avoided deaths associated with imports by importing region: Europe −675,000 (55%), Asia −301,000 (25%), Americas −209,000 (17%), Africa −33,000 (3%), Oceania −7,000 (1%). - Attributing avoided deaths to exporting regions: Americas −507,000 (41%), Asia −365,000 (30%), Europe −231,000 (19%), Africa −118,000 (10%), Oceania −5,000 (0.4%). Country-level highlights: - Largest health benefits from imports (avoided deaths): United States −140,000; Russia −134,000; Germany −107,000; China −89,000; United Kingdom −61,000. Papua New Guinea had a small net increase (+4 deaths) due to red meat imports outweighing benefits. - Largest exporter contributions to avoided deaths: China −117,000 (vegetables, nuts); United States −102,000 (nuts, legumes); Brazil −92,000; Spain −86,000 (vegetables, fruits); Turkey −69,000 (fruits). - Exporters contributing to increased mortality via red meat exports: Germany +10,000; Denmark +7,000; Ireland +3,500; Uruguay +2,000; Paraguay +1,400. Overall: Trade’s net health impact corresponds to about 19% of the total diet-related health burden associated with low intakes of fruits, vegetables, legumes, and nuts, and high intakes of red meat.

The study demonstrates that international food trade measurably influences dietary risks and non-communicable disease mortality, with substantial benefits from trade in fruits, vegetables, legumes, and nuts and harms from trade in red meat. By directly linking trade flows to final health outcomes, explicitly resolving bilateral patterns, and covering all trading countries, the analysis advances previous literature that focused on broader health markers or nutrient availability. The net trade-attributable health effect—roughly one-fifth of the global burden associated with the specified dietary risks—highlights the importance of integrating health considerations into trade and agricultural policy. Conceptually, the work adds a health dimension to embodied-impact literature, showing that unlike environmental and social footprints which primarily affect exporting regions, embodied health impacts occur in importing regions and vary with baseline consumption and mortality. The benefits accrue largely to developed, import-reliant regions (notably Europe), often supplied by less developed exporters. The results suggest that disruptions to trade (from climate events, policy shifts like Brexit, or conflicts such as Russia–Ukraine) could substantially alter diet-related disease burdens. Case analyses indicate Ukraine’s exports contributed to notable mortality reductions in importers, while Russia’s population benefits significantly from imports of health-promoting foods, implying sanctions on such foods could harm public health. Policy relevance includes reducing barriers to health-promoting food trade while curbing exports/imports of foods linked to increased mortality (for example, red meat), and considering producer incentives to diversify away from export-focused livestock in net exporting countries associated with harm.

International trade in key food groups substantially shapes dietary risks and mortality worldwide. Imports of fruits, vegetables, legumes, and nuts reduce non-communicable disease mortality, while red meat imports increase it; combined, trade yields a significant net reduction in deaths, concentrated in regions with higher import dependence. The findings underscore the need for health-sensitive trade and agricultural policies that facilitate access to health-promoting foods and constrain trade in foods linked to harm. Future research should extend to other health dimensions of trade, including impacts on overweight/obesity and ultra-processed foods, and apply the approach longitudinally to evaluate past and future trade agreements and disruptions.

- Scope of dietary risks: The analysis focused on five food-related risks and did not assess other health-relevant aspects of trade, such as effects on overweight/obesity or consumption of ultra-processed foods; thus, it cannot determine whether trade is overall beneficial or detrimental to health. - Causal inference: Health impact estimates rely on assumed causal relationships from observational meta-analyses; while supported by dose–response evidence, biological plausibility, and experimental data on intermediates, residual confounding cannot be excluded. - Transferability: Embodied health impacts are not conserved between exporting and importing regions due to differences in baseline intakes and mortality rates, which may affect attribution. - Data and modeling: Trade data corrections and conversion factors, waste adjustments, age-attenuation, and relative-risk estimates introduce uncertainties; results are subject to the underlying data quality and model assumptions, though uncertainty was quantified via 95% CIs.