Highlights
– Multi-objective optimization of diets in 368,733 EPIC participants found synergistic benefits when the EAT‑Lancet Healthy Reference Diet (HRD) score, plant species richness, and lower intake of ultra‑processed foods (UPFs) were improved concurrently.
– Optimal diets increased mean HRD scores by ~13.9 points and plant dietary species richness by ~1.36 species, reduced UPF share by ~12.4 percentage points, improved PANDiet nutrient adequacy by ~4.1 percentage points, and lowered dietary greenhouse gas emissions (GHGe) by ~1.07 kg CO2‑eq/day and land use by ~1.43 m2/day.
Background
Global dietary patterns increasingly favor a narrow set of foods and a growing share of ultra‑processed foods (UPFs). These shifts are implicated in rising noncommunicable disease burdens and contribute substantially to environmental pressures including greenhouse gas emissions (GHGe) and land use. Sustainable diet frameworks, notably the EAT‑Lancet Healthy Reference Diet (HRD), aim to reconcile nutrition and planetary health by promoting plant-forward, diverse diets with moderated animal-sourced foods. Two additional, less-studied dimensions are food biodiversity — typically operationalized as dietary species richness (DSR) — and food processing level (for example, Nova classification). Though each has been linked independently to nutritional quality or environmental metrics, their combined impact on nutrient adequacy and environmental sustainability across large populations has not been comprehensively evaluated.
Study design and methods
The study analyzed dietary data from 368,733 adult participants in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Dietary intake was characterized by:
- Healthy Reference Diet (HRD) adherence using a 0–140 point HRD score reflecting concordance with EAT‑Lancet targets.
- Dietary species richness (DSR), disaggregated into DSRPlant and DSRAnimal, representing the number of distinct species consumed.
- Food processing level quantified with the Nova classification expressed as percentage of total grams per day (% g/day) attributable to UPFs.
Primary outcomes were the PANDiet score (Probability of Adequate Nutrient Intake Diet), dietary greenhouse gas emissions (kg CO2‑eq/day), and dietary land use (m2/day). Regression models assessed associations and interactions among HRD, DSR, and Nova-derived UPF intake with outcomes. These regression results then informed a multi‑objective optimization (MOO) approach to identify counterfactual diets that balance improvements in nutrient adequacy and reductions in environmental impacts while remaining feasible relative to observed consumption patterns. Reported changes compare optimized diets to observed diets, and uncertainty is presented as 95% confidence intervals where available.
Key findings
The multi‑objective optimization returned dietary scenarios showing that moderate, simultaneous adjustments across the three dietary dimensions yielded meaningful improvements in both nutrition and environmental impact.
Main quantitative results
- HRD adherence: mean HRD score increased by 13.91 points (95% CI: 13.89 to 13.93) relative to observed diets.
- Food biodiversity: plant dietary species richness (DSRPlant) increased by 1.36 species (95% CI: 1.35 to 1.37).
- Food processing: UPFs were substituted with unprocessed or minimally processed foods by 12.44 percentage points (95% CI: 12.40 to 12.49) of total grams/day.
- Nutrient adequacy: PANDiet score increased on average by 4.12 percentage points (95% CI: 4.10 to 4.14).
- Environmental outcomes: GHGe decreased by 1.07 kg CO2‑eq/day (95% CI: 1.05 to 1.09) and land use decreased by 1.43 m2/day (95% CI: 1.41 to 1.45).
Interpretation of effect sizes
Although absolute changes in single dimensions were moderate (e.g., an increase of ≈1.4 plant species and a ≈12 percentage point reduction in UPF grams), combined they produced clinically and policy‑relevant improvements. A ~4.1 percentage point mean increase in PANDiet indicates better overall nutrient adequacy at the population level, which may translate to fewer micronutrient shortfalls. The GHGe reduction of ~1.07 kg CO2‑eq/day per person is substantial: scaled to populations and across time, such per‑person reductions would meaningfully contribute to climate mitigation targets when implemented widely.
Secondary and subgroup findings
The paper reports that improvements were broadly consistent across EPIC countries and cohorts but that baseline diets and the scope for improvement varied regionally. Interactions suggested that gains from increased biodiversity were particularly valuable in diets with higher baseline UPF shares and lower HRD adherence — implying diminishing returns where diets were already plant‑diverse and minimally processed.
Expert commentary and contextualization
This study integrates three complementary aspects of diet quality — nutrient targets (EAT‑Lancet HRD), biodiversity (DSR), and processing level (Nova) — within a large and geographically diverse European cohort, using state‑of‑the‑art multi‑objective optimization. The approach is methodologically rigorous and pragmatically oriented: rather than prescribing radical dietary overhaul, optimization sought feasible changes that could realistically be adopted by populations. The result — that moderate, multi‑dimensional shifts generate synergistic improvements — aligns with mechanistic and empirical evidence linking higher plant diversity to better micronutrient profiles, and lower UPF consumption to improved dietary quality and reduced environmental footprints.
Strengths include the very large sample, use of validated cohort dietary data, explicit modeling of species‑level biodiversity, and an optimization framework that balances competing objectives rather than focusing on single outcomes. The reported CIs are narrow, reflecting precision from large sample size and repeated modeling.
Limitations deserve consideration. Dietary assessment in EPIC relies largely on self‑reported instruments (food frequency questionnaires or diet histories), subject to recall and misclassification biases; disaggregation to species level is challenging and may undercount biodiversity, particularly for mixed dishes. Life‑cycle assessment (LCA) data used to derive GHGe and land use estimates carry uncertainty and vary by production system, geographical origin, and food processing methods; any environmental gains estimated are contingent on the accuracy and representativeness of those LCA inputs. The optimization focused on nutritional adequacy and two environmental metrics; other important outcomes such as water use, biodiversity impacts at farm scale, cultural acceptability, affordability, and food safety were not jointly optimized. Finally, cross‑sectional optimization models cannot establish causal impacts of dietary change on health outcomes, and behavior change strategies to achieve suggested substitutions were not tested.
Clinical and public health implications
For clinicians, dietitians, and public health policymakers, the findings reinforce several actionable messages:
- Incremental, combined improvements across multiple dietary dimensions can yield substantial joint gains in nutrition and sustainability; interventions need not be all‑or‑nothing.
- Encouraging greater plant species variety (not just volume) in diets — for example, rotating pulses, whole grains, vegetables, fruits, nuts, and herbs — can boost micronutrient adequacy.
- Reducing UPF consumption and replacing those calories gram‑for‑gram with minimally processed plant‑based foods and modest amounts of animal‑sourced foods aligned with EAT‑Lancet targets is feasible and beneficial.
- Policy levers (food procurement, agricultural incentives, labeling, reformulation restrictions, fiscal measures) that incentivize biodiversity in production and lower UPF availability could produce synergistic health and climate co‑benefits.
Research and implementation gaps
Key next steps include:
- Intervention studies testing the feasibility, acceptability, and health outcomes of diet patterns designed to simultaneously increase species richness, decrease UPFs, and align with HRD targets.
- Refinement of biodiversity metrics that capture the nutritional contribution of species within mixed dishes and processed products, and assessment of agricultural biodiversity’s role in supply chains.
- Extended environmental assessments incorporating water use, nutrient runoff, and biodiversity loss metrics, and exploring trade‑offs across regions and production systems.
- Equity analyses: understanding affordability, cultural fit, and access barriers for lower‑income groups and diverse cultural diets across regions.
Conclusion
This large, methodologically rigorous study demonstrates that moderate, coordinated improvements across the EAT‑Lancet HRD adherence, plant species richness, and reduced ultra‑processed food intake have the potential to produce meaningful gains in nutrient adequacy while lowering GHGe and land use. The findings support a multidimensional approach to dietary guidance and policy that values food biodiversity and processing level as complements to nutrient‑based and climate‑focused recommendations. Implementing such strategies will require coordinated efforts across agriculture, food systems, clinical practice, and public policy, and should be guided by attention to cultural context, affordability, and robust environmental accounting.
Funding and trial registration
Funding details are reported in the original publication (Berden et al., Am J Clin Nutr. 2025). This analysis used observational cohort data from EPIC and applied computational optimization; it is not a registered interventional trial.
Selected references and further reading
Key contextual documents include the EAT‑Lancet Commission report on healthy diets from sustainable food systems (2019), foundational literature on the Nova classification and ultra‑processed foods, and methodological papers describing the PANDiet nutrient adequacy metric and methods for dietary life‑cycle assessment. The primary study summarized here: Berden J et al. SYNERGIES BETWEEN FOOD BIODIVERSITY, PROCESSING LEVELS, AND THE EAT‑LANCET DIET FOR NUTRIENT ADEQUACY AND ENVIRONMENTAL SUSTAINABILITY: A MULTI‑OBJECTIVE OPTIMIZATION USING THE EPIC COHORT. Am J Clin Nutr. 2025 Nov 17. doi:10.1016/j.ajcnut.2025.11.006.
Practical takeaway
Small, realistic shifts — eat a wider variety of plant species across meals, choose fewer ultra‑processed foods, and move toward the EAT‑Lancet plate — can together make diets both healthier and more sustainable. These co‑benefits should inform dietary guidance, procurement policies, and public health campaigns seeking win‑win solutions for human and planetary health.
