This study aimed to identify diets with improved nutrient quality and environmental impact within the boundaries of dietary practices.

We used Data Envelopment Analysis to benchmark diets for improved adherence to food-based dietary guidelines (FBDG). We then optimised these diets for dietary preferences, nutrient quality and environmental impact. Diets were evaluated using the Nutrient Rich Diet score (NRD15.3), diet-related greenhouse gas emission (GHGE) and a diet similarity index that quantified the proportion of food intake that remained similar as compared with the observed diet.

National dietary surveys of four European countries (Denmark, Czech Republic, Italy and France).

Approximately 6500 adults, aged 18–64 years.

When dietary preferences were prioritised, NRD15·3 was ~6 % higher, GHGE was ~4 % lower and ~85 % of food intake remained similar. This diet had higher amounts of fruit, vegetables and whole grains than the observed diet. When nutrient quality was prioritised, NRD15·3 was ~16 % higher, GHGE was ~3 % lower and ~72 % of food intake remained similar. This diet had higher amounts of legumes and fish and lower amounts of sweetened and alcoholic beverages. Finally, when environmental impact was prioritised, NRD15·3 was ~9 % higher, GHGE was ~21 % lower and ~73 % of food intake remained similar. In this diet, red and processed meat partly shifted to either eggs, poultry, fish or dairy.

Benchmark modelling can generate diets with improved adherence to FBDG within the boundaries of dietary practices, but fully maximising health and minimising GHGE cannot be achieved simultaneously.

The objective of this research is to propose methodology that can be used to benchmark current diets based on their nutrient intakes and to provide guidelines for improving less healthy diets in a way that is acceptable for the studied population.

We discuss important limitations of current diet models that use optimisation techniques to design healthier and acceptable diets. We illustrate how data envelopment analysis could be used to overcome such limitations, and we describe mathematical models that can be used to calculate not only healthier but also acceptable diets.

We used data from the Nutrition Questionnaires plus dataset of habitual diets of a general population of adult men and women in The Netherlands (n 1735).

Adult population.

We calculated healthier diets with substantial higher intakes of protein, fibre, Fe, Ca, K, Mg and vitamins, and substantially lower intakes of Na, saturated fats and added sugars. The calculated diets are combinations of current diets of individuals that belong to the same age/gender group and comprise of food item intakes in proportions observed in the sample.

The proposed methodology enables the benchmarking of existing diets and provides a framework for proposing healthier alternative diets that resemble the current diet in terms of foods intake as much as possible.