Enviroscore: normalization, weighting, and categorization algorithm to evaluate the relative environmental impact of food and drink products

The global food system faces significant environmental challenges, contributing substantially to climate change, water scarcity, deforestation, and ecosystem pollution. Food production in the European Union alone accounts for 20-30% of the overall human impact. The heterogeneity within the food sector, in terms of production practices, company size, and seasonality, results in varying environmental performances for the same product depending on its origin and processes. Life Cycle Assessment (LCA) offers a robust methodology for evaluating environmental impacts and identifying improvement strategies. However, current approaches have limitations in communicating these results to consumers: they often focus on single environmental impact categories (typically climate change), ignoring potential shifts in other impacts; most focus on specific products, hindering comparisons; and existing systems lack robust scientific backing and consumer trust. These limitations necessitate the development of standardized normalization and weighting methods for a wider range of environmental impact categories to create a single, easily understandable index. This study aims to develop such an index, the EFSI, and a corresponding 5-scale score, the Enviroscore, to improve communication of food's environmental impact and encourage more sustainable consumption and production practices. The methodology involves defining normalization factors (NF) using the European Food Basket as a reference, selecting appropriate weighting factors, validating the EFSI's ability to capture variability between and within food products, and establishing and validating threshold values for the Enviroscore using the Delphi method.

Existing literature highlights the significant environmental burden of the food system, emphasizing the need for more effective communication of environmental impacts to consumers. Several studies have demonstrated the potential of carbon labeling to reduce the carbon footprint of food baskets. However, these studies often fail to capture the full picture, neglecting potential trade-offs between different impact categories. Furthermore, existing environmental impact information systems for food products lack a robust science-based method, and consumer trust in producers’ claims remains low. The limitations of existing methodologies emphasize the need for a more comprehensive and transparent approach, integrating various impact categories and providing a readily interpretable result for consumers.

A three-step approach was employed: **1. Development of the European Food Environmental Footprint Single Index (EFSI):** Normalization factors (NF) were defined using the environmental impact of the average European food basket as a reference. This basket comprised 23 representative food items selected from FAO Food Balance Sheets, with packed water added due to its high consumption. Life Cycle Assessment (LCA) was used to characterize the environmental impacts of the basket according to the Product Environmental Footprint (PEF) methodology and the ILCD methodology (13 impact categories). Normalization factors were calculated using Equation 1: NF_(f) = ΣC(i) x e(f_i)/_Population. Weighting factors from the EC were used to aggregate the 13 impact categories, resulting in the EFSI. **2. Relative Validation of the EFSI:** The EFSI's ability to capture variability between and within food products was evaluated by comparing its results to the EC Single Score. A dataset of 149 hypothetical food items (21 food products with variations in origin, production methods, and transport) was created, and their environmental impacts were characterized using LCA. Boxplots visualized the distribution of EFSI and EC Single Score results, while a correlation heatmap analyzed the influence of individual impacts and their correlation with each index. **3. Development and Validation of the Enviroscore:** Threshold values for the 5-scale Enviroscore were defined based on the distribution of EFSI values for 22 representative food items. The accuracy of these thresholds was validated by comparing the Enviroscore with expert categorization obtained using the Delphi method (three rounds with seven experts). The performance of the Enviroscore was tested by comparing the scores of representative food items with the scores of their corresponding hypothetical items, considering variations in production and transport.

The EFSI and Enviroscore effectively captured variability between and within food products. Plant-based products generally had lower EFSI and Enviroscore values than animal-based products. The EFSI showed better correlation with water scarcity than the EC Single Score (r=0.624 vs r=0.228). The Delphi method validation showed good agreement between the Enviroscore and expert categorization (weighted Kappa 0.642; p = 0.0025), with 100% accuracy for categories A and E. Performance validation revealed higher agreement for Enviroscore categories B and D. Deviations were mainly due to differences in production methods and transportation modes, particularly air transport. The study found that transport mode significantly impacted scores, with air transport resulting in higher scores, especially for plant-based products. The correlation heatmap showed that climate change was highly correlated with most impact categories, while water scarcity showed low correlation with others. Both EFSI and EC Single Score were highly correlated with most impacts, except for water scarcity, where EFSI showed a stronger correlation.

The EFSI and Enviroscore provide a science-based, transparent, and easily understandable method for communicating the relative environmental impact of food products. The methodology's strength lies in its basis on the established PEF methodology and its ability to reflect variability between and within food products. The good agreement between the Enviroscore and expert categorization, despite the inherent variability of food products and lack of a gold standard, supports the validity of the approach. The sensitivity of the Enviroscore to differences in production methods and, especially, transport modes highlights its ability to incentivize improvements in sustainability throughout the supply chain. The EFSI's superior correlation with water scarcity underscores the importance of considering this critical resource constraint in food system assessments. The limitations of the study, such as the limited number of data points and food products, suggest a need for further research to expand the database and enhance the model's robustness.

This study successfully developed the EFSI and Enviroscore, providing a novel method for communicating food's relative environmental impact. The approach offers a transparent, science-based tool for consumers and food businesses to make more informed and sustainable choices. Further research should focus on expanding the dataset to enhance the robustness and generalizability of the findings and explore additional weighting schemes to reflect varied societal values.

The study's limitations include the relatively small number of food products and data points used in the analysis. The selection of representative food items and the assumptions made in the LCA could also affect the results. Further research with a larger, more diverse dataset is necessary to confirm the generalizability and robustness of the proposed Enviroscore and EFSI.