The nature of protein intake as a discriminating factor of diet sustainability: a multi-criteria approach

Abstract

Animal production is responsible for 56-58% of the GHG emissions and limiting meat consumption would strongly contribute to reducing human health risks in Western countries. This study aimed to investigate the nature of protein intake as a discriminating factor for diets' sustainability. Using data from 29,210 French adults involved in the NutriNet-Santé cohort, we identified clusters according to 23 protein sources. A multicriteria (environmental, economic, nutritional and health) sustainability analysis was then conducted on the identified clusters. The economic analysis focused on both food and protein expenditure structures, using a budget coefficient approach. Relative values of clusters compared to the whole sample were calculated. We identified five clusters: milk-based, meat-based, fast food-based, healthy-fish-based, and healthy-plant-based. We found that the healthy-plant-based and healthy-fish-based clusters were the most sustainable, conciliating the compromise between human health (0.25 and 0.53 respectively for the Health Risk Score) and the protection of the environment (- 62% and - 19% respectively for the pReCiPe indicator). Conversely, the highest environmental impacts (+ 33% for the pReCiPe indicator) and the highest health risk (0.95 for the HRS) were observed for the meat-based cluster, which was associated with the lowest nutritional scores (- 61% for the PNNS-GS2 score). The economic analysis showed that the healthy-plant-based cluster was the one with the highest food budget coefficient (+ 46%), followed by the healthy-fish-based cluster (+ 8%), partly explained by a strong share of organic food in the diet. However, the meat-based cluster spent more of their food budget on their protein intake (+ 13%), while the healthy-plant-based cluster exhibited the lowest expenditure for this intake (- 41%). Our results demonstrate that the nature of protein intake is a discriminating factor in diet sustainability. Also, reducing animal protein consumption would generate co-benefits beyond environmental impacts, by being favorable for health, while reducing the monetary cost associated with protein intake.

Figure 1

Protein contributions per food group across clusters. Values are energy-adjusted means of protein contributions of food groups (%/day) computed using ANCOVA model. Food groups are formed as follows: Vegetables include all vegetables and soups; Fruits include fresh fruits, fruits in syrup and compote, dried fruits and seeds; Beverages include all non-alcoholic beverages that are fruit nectar, syrup, soda (with or without sugar), plant-based beverages (except soya-based), milk consumed with tea/coffee; Dairy products include yogurts, fresh cheese and cheese; Potatoes include other tubers; Cereals include breakfast cereal low in sugar, bread semolina, rice and pasta; SFF (sweet and fat foods) include croissants, pastries, chocolate, biscuits, milky dessert, ice cream, honey and marmalade, cakes, chips, salted oilseeds, salted biscuits; Fast-food include sandwich, prepared foods such as pizza, hamburger, ravioli, panini, salted pancake, etc.; Soya-based food includes tofu, soya-based meat substitute and vegetable patties, soya-based yogurt, soya-based milk; Fat includes animal fat (butter and lards); Dressing includes ready-to-use salad dressing, mayonnaise or cream-based sauces, sour cream and butter and all fat-based sauces; Oil includes plant-based oils; Meat includes beef and pork.

Figure 2

Sustainability indicators across clusters. The energy-adjusted means of indicators, computed using ANCOVA model, were rescaled to the same scale by equalizing the maximum value to 1 and the minimum value to 0 for each indicator. For the environmental indicators and the HRS, higher values denote higher impacts/risk. For nutritional quality indicators, higher values denote higher nutritional quality. HRS heath risk score, GHG greenhouse gas (kg CO2 eq/d), CED cumulative energy demand (MJ/d); Land Use (m²/d), pRecipe partial Recipe, PANDiet diet quality index based on the probability of adequate nutrient intake, cDQI comprehensive diet quality index, PNNS-GS2 Programme National Nutrition Santé-Guidelines Score 2.

Figure 3

Food budget coefficients across clusters according to production mode (A); protein budget coefficients across clusters (B). Values are energy-adjusted means of budget coefficients computed using ANCOVA model. Panel A refers to food budget coefficients across clusters according to production mode (organic/conventional). Panel B refers to protein budget coefficients across clusters.