A young man living in a city may have very different food options, habits and needs than a middle-aged woman living in a rural community. Yet national nutrition campaigns often speak to them as though they eat the same foods under the same circumstances.
A new global modeling study suggests those differences matter. Researchers combined dietary, demographic and food-supply data from 165 countries to examine how diet quality and environmental effects vary by age, sex, education and urban or rural residence. They found that the highest-impact demographic group produced nearly twice the diet-related greenhouse gas emissions of the lowest-impact group. Healthier dietary patterns were also generally associated with lower emissions, land use and water use.
The findings, published in Nature Food, do not show that changing one person’s diet will produce a specific environmental benefit or improve that person’s health. The study did not ask people to change what they ate or follow them over time. Instead, it used a new modeling framework called MATILDA to combine estimated food intake with global supply-chain and population data.
The model evaluated 72 demographic groups within each country rather than relying only on national averages. That allowed researchers to identify differences that can disappear when an entire country is assigned one typical diet.
“There is no one-size-fits all when it comes to sustainable and healthy diets,” said Hongyi Cai of Leiden University, who helped develop MATILDA. “When we benchmark the population against the actual diets they follow, food system impacts become more accountable, health inequalities become more explicit, and differentiated needs become more clear.”
Across the countries studied, highly educated urban men ages 15 to 34 had the highest median diet-related greenhouse gas emissions, at 5.66 kilograms of carbon dioxide equivalent per day. Rural women ages 35 to 54 with lower levels of education had the lowest, at 2.85 kilograms per day.
Those findings describe median patterns across 165 countries. They do not mean every young urban man has a high-impact diet or every rural woman has a low-impact one.
They also should not be interpreted as a ranking of who eats responsibly.
A lower environmental footprint may reflect eating less food, having fewer choices or being unable to afford certain products. It does not necessarily indicate that a person is eating enough, meeting nutritional needs or voluntarily choosing a sustainable diet.
The two groups’ diets were also not separated by a simple plant-versus-animal divide. Plant-sourced foods accounted for 62% of the diet in the highest-impact group and 70% in the lowest-impact group. The remaining difference may involve the types and quantities of foods consumed, where those foods were produced and how they moved through global supply chains.
The researchers assessed diet quality using the Alternative Healthy Eating Index, or AHEI. The score rewards higher consumption of foods such as vegetables, fruits, whole grains, nuts and legumes while considering intake of red and processed meat, sugar-sweetened drinks, sodium and other dietary components.
For every 5% increase in the diet-quality score, the model found a 7.7% reduction in diet-related greenhouse gas emissions, a 4.4% reduction in water use and a 7.2% reduction in land use.
Those numbers represent modeled associations across dietary patterns. They do not mean an individual who raises a diet score by 5% is guaranteed to reduce emissions by exactly 7.7%. The study also did not measure whether people with higher scores experienced less heart disease, cancer or other health outcomes.
Still, the overlap suggests that policies supporting foods associated with better diet quality may sometimes offer environmental benefits as well.
That does not make every plant food equally nutritious or environmentally preferable. The effects of a food can vary according to how and where it was grown, how much land and water production required and how far it traveled through supply chains. Diet quality also depends on the overall pattern rather than whether a product carries a plant-based label.
“People don't make food choices in isolation,” said Paul Behrens, a study author affiliated with the University of Oxford and Leiden University. “What we eat is shaped by where we live, what we can afford and what is available to us.”
That point is central to the researchers’ policy argument.
A public campaign encouraging people to eat more fruits, vegetables or legumes may have limited effect in a community where those foods are expensive, difficult to reach or rarely stocked. A supermarket discount may help some households but do little for people without reliable transportation or adequate cooking facilities.
Other communities may already eat relatively low-impact diets but still face food insecurity or nutritional deficiencies. Asking those populations to reduce consumption further could deepen inequities rather than improve health.
The researchers argue that policies should begin with what different groups already eat and the barriers they face. Options could include supermarket incentives, changes in food availability, public information campaigns, pricing policies or regulations, but the most appropriate approach would vary by population and country.
“Instead of treating entire populations as though they eat the same foods, we can now identify where interventions are likely to have the greatest benefit while making sure vulnerable groups aren't left behind,” Behrens said.
The model itself has limitations. It combines information from several large datasets, each with its own uncertainty and gaps. Dietary surveys can be affected by inaccurate recall or reporting. Some countries and demographic groups have stronger underlying data than others.
The analysis also placed people into broad categories based on age, sex, education and urban or rural location. It could not fully capture income, cultural identity, household structure, food insecurity, disability or the many personal circumstances that shape eating.
The Alternative Healthy Eating Index was also developed from nutrition research and priorities that may not capture the full cultural and nutritional context of every country in the analysis.
MATILDA therefore cannot prescribe a perfect diet or policy for any individual community. Its value lies in showing why national averages are often insufficient.
Food advice can sound simple when reduced to phrases such as “eat more plants” or “choose healthier options.” In practice, those choices depend on whether suitable foods are accessible, affordable, familiar and realistic within people’s lives.
The study suggests that healthier and lower-impact diets may often overlap. Reaching them, however, will likely require more than asking everyone to make the same change.
Hongyi Cai received support from a starter grant from the Dutch government. Arthur Schoorlemmer received support from Leiden University’s Kiem seed funding and the Dutch Research Council through the National Science Agenda project “Resilient and Efficient Agricultural Proteins to Sustain Our World.” Paul Behrens received support through a British Academy Global Professorship and a REAPRA Senior Fellowship.