The ongoing nutrition transition thwarts long-term targets for food security, public health and environmental protection

Global dietary patterns are shifting from scarce, plant-based diets with fresh and minimally processed foods to more affluent diets high in sugar, fat, and animal-source foods, often as highly processed products. While undernourishment persists in absolute numbers, there is a shift in public health burdens from undernutrition-related infectious and neonatal disorders towards overconsumption-related chronic diseases (e.g., diabetes, cardiovascular diseases). Suboptimal diets are currently the leading global health risk, with large attributable DALY burdens, and healthier diets could avoid millions of premature adult deaths annually. At the same time, food demand is driven by population growth, demographic aging, reduced physical activity, and increasing household food waste, amplifying agriculture’s environmental footprint (land use, water withdrawals, GHG emissions, nutrient pollution, biodiversity loss). Recognizing the “Global Syndemic” of obesity, undernutrition, and climate change, the study aims to integrate these dimensions within one framework. Central research question: How have food-consumption-related symptoms of the Global Syndemic evolved worldwide over recent decades, and what are the outcomes if the observed nutrition transition continues into the future? The study assesses underweight, overweight and obesity, body height (as a proxy for stunting), caloric intake, household food waste, dietary composition, and total demand for food and animal-source foods, enabling integrated analysis across health, food systems, and environmental change.

The paper builds on the concept of the nutrition transition (Popkin) and the Lancet Commission’s framing of the Global Syndemic linking obesity, undernutrition, and climate change. It references evidence that suboptimal diets are the leading global health risk and situates rising food demand as a driver of agricultural expansion and environmental pressures. Prior approaches often estimated food demand without explicitly incorporating anthropometric dynamics (BMI, height, age-sex structure, PAL) or relied on uniform regional food waste shares. Earlier studies provided demand projections but used shared elasticity parameters or static BMI assumptions, and often lacked future trajectories of food intake separate from demand. The authors position their work as addressing these gaps by integrating anthropometric and demographic dynamics, providing country-level projections of BMI distributions by age and sex, projecting body height, and estimating intake and waste consistently with dietary composition using a nested demand model. Comparisons with FAO and other projections indicate their estimates fall within established ranges for total and livestock calorie demand growth, offering validation while adding methodological innovations.

An open-source food demand model simulates long-term scenarios (1965–2100) in 5-year steps across 249 ISO 3166-1 countries/territories. It distinguishes subpopulations by sex (male/female), age (20 five-year cohorts from 0 to 100+), and BMI groups (six adult BMI ranges; five child BMI SD groups). Drivers are population (by age/sex) and per-capita income from the Shared Socioeconomic Pathways (SSPs). The model uses regression-based functional relationships (estimated on historical data) to project: (1) body height (male/female 15–19) as a function of lagged per-capita demand for animal-source foods, pulses, and oils; children’s heights are scaled from WHO growth standards; older cohorts retain their cohort-specific adult height. (2) BMI distributions by age and sex as functions of per-capita income, yielding shares in underweight/normal/overweight/obese (adults) and SD bands (children). (3) Physical activity levels (PAL) by age-sex group inferred from observed global inactivity data (WHO) complemented by rules tied to income, age, and sex. (4) Basal metabolic rate (BMR) via Schofield equations by age, sex, and body weight (also tested variants including height and climatic temperature), then total energy expenditure as BMR × PAL. (5) Food intake is computed for each BMI-age-sex cell, aggregated to country totals using demographic projections and adding energy for pregnancy/lactation (based on newborns). (6) Household food waste is estimated top-down by regressing the demand-to-intake ratio on per-capita income, with demand defined as FAOSTAT dietary energy availability; waste equals demand minus intake. (7) Dietary composition is estimated with a nested demand tree consistent with total calories: first animal vs plant calories; then empty calories (oils, sugar, alcohol) vs nutritious plant calories; then vegetables/fruits/nuts vs staples, each split as an income-dependent share. The model produces per-capita and total demand (intake and waste), BMI/obesity/underweight prevalence, body height, and dietary composition. Scenarios: SSP1–SSP5 trajectories for income and population. Calibration: For 1965–2010, projections are bias-corrected to match reported data for body height, BMI, per-capita demand, and dietary composition using additive calibration factors held constant post-2010 (bounds enforced to maintain plausible ranges). Validation: Five-fold cross-validation shows uncalibrated out-of-sample R² of ~0.418–0.722 (2010), improving to 0.542–0.897 when calibrating on 1975 for predicting 2010. The model compares favorably with previous versions without time-dependent parameters and aligns with external projections for demand growth. Code and data availability: GitHub (MAGPIE framework), Zenodo archives for model, outputs, and analysis scripts; R libraries mrregression and moinput used for regressions and data processing.

- By 2050, 45% (39–52%) of the world’s population will be overweight and 16% (13–20%) obese (2010: 29% overweight, 9% obese). By 2100: 56% overweight, 23% obese. - Underweight prevalence approximately halves, but absolute numbers remain 0.4–0.7 billion; in SSP2, underweight declines from 744 million (11%) in 2010 to 528 million (6%) in 2050 and 394 million (4%) by 2100. Across SSPs: 383–741 million (4–7%) in 2050; 330–733 million (4–6%) in 2100. - Children’s obesity rises from 1% (1965) to 6% (2010), to 9% (2050) and 13% (2100). Among children, overweight/obesity is higher in males; among adults, overweight higher in working-age men, obesity similar by sex but higher among women 60+. - Average adult height increased globally from 167→169 cm (men) and 155→157 cm (women) between 1965–2010; projected 171 cm (men) and 158 cm (women) in 2050, with substantial cross-country variation. - Global total food demand rose from 12 EJ (1965) to 30 EJ (2010) and is projected to reach 45 EJ (43–47) by 2050 and 48 EJ (36–62) by 2100. High population growth scenarios (SSP3) yield the highest demand; India and Africa drive future increases. - Dietary shift: animal-source calorie share rises from 18% (2010) to 24% (22–27%) by 2050; empty calories increase disproportionately; vegetables, fruits, and nuts increase insufficiently; staple consumption stabilizes. - Decomposition of per-capita demand growth shows similar contributions from rising BMI (overconsumption), increasing food waste, and higher food energy requirements due to demographic change and height; declining PAL has negligible effects. - Animal-source food demand roughly doubles by 2050 (vs 2005), driven mainly by population growth and higher animal shares in diets; smaller roles for waste, BMI, PAL, or energy requirement changes. - Food waste: demand exceeds intake by ~25% (2010) and ~33% (2050, SSP2). Hypothetical global impacts (Table 1): • Eliminating underweight: +1.4% (2010) and +0.7% (2050) demand. • Normalizing BMI for all overweight/obese: −4.4% (2010) and −6.8% (2050). • Raising all inactive to moderate PAL: +5.2% (2010) and +5.5% (2050). • Eliminating household food waste: −24.9% (2010) and −33.2% (2050). - Regional dynamics: historical demand growth largest in Asia and North Africa; future increases largest in India and Africa. - Model performance: Out-of-sample R² up to ~0.897 (with calibration); demand growth +65% to 2050 (vs 2005) aligns with FAO (+54%) and other studies (+70%); livestock calories +108% (vs FAO +76%, similar to +110% by Bijl et al.).

Findings indicate a persistent and advancing nutrition transition: underweight declines relatively but remains substantial in absolute terms, while overweight and obesity continue to rise to pandemic levels, undermining SDG2 targets to end all forms of malnutrition. The results support Popkin’s transition from “famine” to “degenerative diseases,” but do not yet show a shift toward healthier patterns via behavioral change; instead, in high-income countries, obesity still increases and the share with healthy BMI declines. Middle-income countries experience coexistence of under- and overnutrition, and diets shift toward high animal-source and empty calories, with inadequate growth in vegetables, fruits, and nuts. These trends, together with demographic change and wasteful consumption, drive strong increases in total and animal-source food demand, implying escalating environmental pressures (land, water, nitrogen, GHGs, biodiversity). Decomposition clarifies how population, aging/height, BMI-related overconsumption, and household waste each contribute to rising demand, improving the attribution of potential mitigation levers and cautioning against misattributing undernourishment when back-calculating from aggregate demand. The analysis underscores that without a paradigm shift in food policy and broader food environments, the Global Syndemic will intensify, and planetary boundaries will be further exceeded. Integrated strategies that address consumption behaviors—particularly reducing animal-source foods and household food waste—offer strong synergies for health and environmental goals, while eliminating underweight would have minimal impact on global food demand. Forward-looking national dietary and anthropometric assessments are needed to anticipate rapid changes and to design context-specific policy interventions; cross-country patterns documented here can guide nations extrapolating beyond their historical experience.

The study provides an integrated, forward-looking global assessment of the nutrition transition, projecting BMI distributions by age and sex, body height, intake, household food waste, dietary composition, and total food and animal-source demand for 1965–2100 across SSPs. It shows that current trajectories are incompatible with achieving zero hunger, healthy diets, and environmental sustainability: underweight persists in absolute numbers, while overweight/obesity rise markedly; total and animal-source food demand increase substantially, amplifying environmental pressures. The open dataset and model furnish interfaces for integrated assessments spanning public health, food systems, and environmental change. Policy should proactively reshape food environments and behaviors, prioritizing reductions in household food waste and animal-source food consumption, while addressing all forms of malnutrition. Future research should: (1) develop national, forward-looking dietary and anthropometric scenarios; (2) extend models to include additional drivers (e.g., urbanization, prices, inequality, globalization, climate variables) and time lags; (3) update parametrizations as new high-income data become available; and (4) evaluate policy packages (taxes, labeling, marketing restrictions, public procurement, education) for synergistic health and environmental benefits.

- Intake vs FAOSTAT mismatch in some low-income countries: Anthropometric intake estimates exceed FAOSTAT dietary energy availability for certain settings, not resolved by alternative metabolic equations (including height or temperature). Possible explanations are discussed in supplementary materials. - High-income uncertainty: Limited observations at very high income levels (> USD 50,000 per capita) introduce uncertainty in projecting dietary composition and obesity saturation. The model assumes saturation near current maxima, potentially underestimating future obesity in rich countries. It is better suited for SSP2/3/4/5 continuation of materialistic lifestyles than for SSP1 with preference shifts. - Socioeconomic drivers simplified: Per-capita income proxies aggregate development; other determinants (inequality, education, urbanization, globalization, food prices, policy, industry power) are not explicitly modeled due to collinearity and data/projection limitations, which may miss time trends (e.g., global rise in obesity independent of income) and lead to conservative obesity projections. - Social dynamics uncertainty: Correlations between income and BMI, waste, and diet could change due to technological, social, or policy shifts; adding robust projections for variables like urbanization and prices would improve realism. - Calibration and historical data gaps: Historical BMI (1965–1970) and dietary composition for some countries required model-based gap-filling; calibration factors held constant post-2010 may not capture future structural breaks.