Developing a novel optimisation approach for keeping heterogeneous diets healthy and within planetary boundaries for climate change

Developing a novel optimisation approach for keeping heterogeneous diets healthy and within planetary boundaries for climate change

Play all audios:

Loading...

ABSTRACT BACKGROUND AND OBJECTIVES Current dietary habits have substantial negative impacts on the health of people and the planet. This study aimed to develop a novel approach for achieving


health-promoting and climate-friendly dietary recommendations for a broad range of consumers. SUBJECTS AND METHODS Hierarchical clustering analysis was combined with linear programming to


design nutritionally adequate, health-promoting, climate-friendly and culturally acceptable diets using Swedish national dietary data (_n_ = 1797). Diets were optimised for the average


consumption of the total population as well as for the dietary clusters. RESULTS Three dietary clusters were identified. All optimised diets had lower shares of animal-source foods and


contained higher amounts of plant-based foods. These dietary shifts reduced climate impacts by up to 53% while leaving much of the diet unchanged. The optimised diets of the three clusters


differed from the optimised diet of the total population. All optimised diets differed considerably from the food-group pattern of the EAT-Lancet diet. CONCLUSIONS The novel cluster-based


optimisation approach was able to generate alternatives that may be more acceptable and realistic for a sustainable diet across different groups in the population. SIMILAR CONTENT BEING


VIEWED BY OTHERS SUSTAINABLE HEALTHY DIET MODELING FOR A PLANT-BASED DIETARY TRANSITIONING IN THE UNITED STATES Article Open access 28 November 2023 DIETS CAN BE CONSISTENT WITH PLANETARY


LIMITS AND HEALTH TARGETS AT THE INDIVIDUAL LEVEL Article 21 March 2025 THE HEALTHY DIET BASKET IS A VALID GLOBAL STANDARD THAT HIGHLIGHTS LACK OF ACCESS TO HEALTHY AND SUSTAINABLE DIETS


Article Open access 27 May 2025 INTRODUCTION Contemporary diets in high and middle income countries are major contributors to the burden of chronic diseases as well as to the rapidly


accelerating climate crisis [1]. The global food system–from production to consumption—thus needs a revamp to meet the 2015 Paris Agreement on climate change [2] and the Sustainable


Development Goals. In a market economy, demand and supply of food are closely connected, making consumers’ eating behaviours one of the most important factors contributing to human and


environmental health [3]. The EAT-Lancet Commission has suggested a healthy reference diet that would also help keep the global food system within six environmental planetary boundaries [1].


It emphasises a ‘plant-forward’ diet dominated by whole grains, fruits, vegetables, nuts and legumes where meat and dairy constitute a small or negligible part. Despite this robust


evidence, there is currently no consensus on _how_ to operationalise these dietary targets and achieve acceptability among consumers in different population groups with diverse cultural


backgrounds [4]. For most high-income populations, adoption of the EAT-Lancet diet would imply a significantly higher share of plant-based foods while markedly reducing the intake of


animal-based products [5]. To account for both nutritional and environmental demands as well as affordability, holistic approaches such as optimisation analysis with linear programming (LP)


have been used for a wide range of settings [6, 7]. To also consider the cultural acceptability of optimised diets, the deviation from the reported average diet of the total population has


been minimised [6, 8,9,10,11]. However, delivering one “acceptable” solution based on the average consumption of different foods or food groups may imply minor dietary changes for some


individuals but larger and potentially unrealistic changes for several groups in the population [12,13,14]. For example, male individuals in European countries are likely to face larger


absolute and relative changes to their consumption of red/processed meat as compared to females given their different needs and baseline consumption levels [15]. Hence, developing any type


of food-based advice or guidance by optimisation of the average diet is likely to overlook the heterogeneity of diets within populations [16]. There is thus a need to explore if altering


current optimisation approaches could lead to solutions that better reflect the dietary variability in a given population. The primary aim of this study was to optimise the diet of groups in


the population with different eating patterns and to see if this provides a more realistic approach than optimising for the national average consumption. Diets were optimised to meet


nutritional requirements, food-based dietary guidelines (FBDGs) and a limit for food related greenhouse gas emissions (GHGE) of 1.57 kg/day as suggested by the Intergovernmental Panel on


Climate Change (IPCC) [17]. We also compared the optimised diets to the proposed EAT-Lancet diet [1]. MATERIALS AND METHODS STUDY DESIGN AND DIETARY DATA This was a modelling study combining


hierarchical clustering analysis with linear programming to design nutritionally adequate, health-promoting, climate-friendly and culturally acceptable diets. Self-selected diets were


derived from the nationally representative Swedish dietary survey Riksmaten Vuxna 2010–11 (Riksmaten Adults) [18]. The data, which were collected between May 2010 and May 2011 by the Swedish


Food Agency, is publicly available in fully anonymised form [19]. Briefly, a web-based 4-day diary was completed by 1797 adults aged 18–80, and all foods and drinks consumed over four


consecutive days were recorded. The participants were able to choose from more than 1900 different food items and dishes and several portion sizes. The study sample consisted of 56% females


and the mean age was 48 years. Information on income and other sociodemographic factors was also gathered. A more detailed description of the material and methods used for this study can be


found in the Supplementary Information. NUTRITIONAL COMPOSITION Energy and nutrient intakes of the edible parts of foods as eaten (e.g., cooked pasta) were automatically calculated through


linkage with the Swedish Food Agency’s Food composition database version Riksmaten Vuxna 2010–11. CLIMATE FOOTPRINTS The carbon dioxide equivalents (CO2eq) of foods were derived from the


Climate Database developed and maintained by the Research Institutes of Sweden (RISE) [20], which is linked to the Swedish Food Agency’s Food composition database. The database includes


CO2eq estimations for 2078 food items following life-cycle assessment standards [21, 22] taking into consideration Swedish production and consumption patterns [20]. The CO2eq estimations


consider the impact from carbon dioxide (CO2); methane (CH4); and nitrous oxide (N2O), which have been weighted in line with their respective global-warming potential over a 100 year period


using factors recommended by the IPCC [23]. The CO2eq data did not take into consideration the packaging, transportation from stores to households, meal preparation or food waste. COST OF


FOODS The webpage “Matpriskollen” [24], which compares the prices of foods among twelve of Sweden’s largest food retailers, was used to estimate the price of each food in the year 2020. An


average price was calculated for each food item based on varying available prices for a food item (including low price, conventional and organic varieties). GROUPING OF FOODS For analytical


and descriptive purposes, foods were grouped in 24 food categories, based on the categorisations used in the RISE Climate Database: Red meat (including red meat dishes); Processed meat (both


red meat and poultry); Poultry (including poultry based dishes); Seafood (including fish, mussels and crabs, and seafood dishes); Offal; Dairy (e.g., milk and cheese); Eggs; Pasta and rice


dishes with meat/fish (e.g., composite dishes like lasagne); Pasta and rice dishes with dairy/eggs (e.g., composite dishes like vegetarian lasagne); Vegetable oils; Vegetables (whole


vegetables and a few vegetable based dishes); Potatoes (including potato based dishes); Pulses (beans, lentils, peas and chickpeas); Fruits and berries (including smoothies); Nuts and seeds;


Meat alternatives (e.g., soy mince); Dairy alternatives (e.g., oat milk); Mixed/animal fats (added fats such as butter, margarine-butter mix); Cereals/grains (including e.g., breakfast


cereals and, pasta); Rice; Savoury snacks; Sugar and sweets (including chocolate); Drinks other than milk; and Other (e.g., seasonings and sauces). Further details on the categorisation can


be found elsewhere [20]. The foods in the baseline and optimised diets were additionally re-grouped in order to be comparable to the EAT-Lancet Commission’s food categorisation [1], namely:


Whole grains (rice, wheat, corn and other); Tubers or starchy vegetables (including potatoes); Vegetables; Fruits; Dairy foods (whole milk or equivalents, including butter); Beef, lamb and


pork; Chicken and other poultry; Eggs; Fish; Legumes; Nuts; Added fats (unsaturated oils and saturated oils); and Added sugars. This categorisation was either based on the most dominant


component or calculated based on the proportional shares, based on recipes. CLUSTER ANALYSIS Clusters analysis was performed to identify dominating eating patterns in the Swedish population.


Firstly, the R package clValid [25] was applied to the dietary data to simultaneously compare multiple clustering algorithms and clustering methods. By comparing the discriminatory power of


different calculation paths, clValid identified hierarchical clustering to be the best fitting clustering algorithm for our data. It also proposed using Canberra distances with Ward’s


method in a hierarchical clustering as this combination resulted in the highest value for Dunn’s Index (the ratio of the smallest distance between observations not in the same cluster to the


largest intra-cluster distance). Secondly, the NbClust package in R [26] (which uses 30 different indices to suggest the best clustering approach and number of clusters to choose based on


all combinations of self-organising clusters, distance measures, and clustering methods) was used to determine the optimal number of clusters when combining Canberra distances with Ward’s


method (results suggesting 2 or 3 clusters, visualised in Supplementary Fig. 1). Following on these initial exploratory analyses, data was scaled and hierarchical clustering using Ward’s


method and Canberra distances was applied to the dietary data. Based on the outputs from NbClust, three clusters were chosen for this analysis. Food groups that were consumed by less than


75% of the population were not included in the clustering to avoid bias emerging from missing data. Two exceptions were made for the food groups Pulses and Nuts and Seeds, since these food


groups are seen as indicators of both climate friendliness and healthy eating [1]. Hence, the following food groups were included in the clustering: Red meat, Processed meat, Vegetables,


Fruits and berries, Dairy, Pulses, Nuts and seeds, Seafood, Mixed animal fats, Sugar and sweets, Rice, Potatoes, Cereals/grains, Eggs, and Poultry. Whole grains were also included in the


clustering although not classified as a food group in the food consumption survey. For the clustering procedure, intakes of food groups were standardised for individual energy intake (g/MJ)


to account for heterogeneous energy intake. COMPARING THE CLUSTERS Clusters were compared post-hoc on the basis of the energy-adjusted intake of the food groups included in the cluster


analysis (g/MJ), age (y), income (SEK), sex (male/female), and CO2eq (g/MJ). Kruskal–Wallis test was used to statistically determine if significant differences between clusters existed with


regards to food groups, CO2eq and income since these variables were not normally distributed. Age was normally distributed and thus assessed with Analysis of Variance. Sex (categorical


variable) was assessed using Pearson’s chi-squared test. As for the non-normally distributed variables, the Dunn (1964) Kruskal–Wallis test for multiple comparison (alpha adjusted with the


Benjamini-Hochberg correction) was used as a post-hoc test to identify which clusters that differed significantly. Tukey’s honest significance test was applied as a post-hoc test for the


normally distributed variables. Statistical significance was set at _P_ ≤ 0.05. Both the cluster analysis and all statistical computations were performed in R version 4.1.1 [27]. The


healthiness of the three clusters was calculated in accordance with a previously developed healthy eating index relevant for the Swedish context – SHEIA15 [28]. The ratio between the


baseline intake and the recommended intake of nine different dietary components were accordingly calculated (Supplementary Table 1) and summed to a total score. Ratios <0 and >1 were


recoded to zero and one, respectively, resulting in a range of 0–9. As previously suggested [28], the summed ratios for the different dietary components were categorised into three defined


levels; low (<4 points), medium (4–7 points), and high (>7 points). OPTIMISATION The chosen optimisation method of LP has successfully been applied to optimise goal determinants of


diets while considering a multitude of (sometimes conflicting) constraints [6, 29]. Briefly, it is the application of an algorithm for either maximising or minimising a specific linear


objective function (the variable being optimised) which is subjected to a set of linear constraints (predetermined requirements that should be met) on a list of decision variables (in this


case, the absolute amount of each individual food item) [30]. A feasible solution is found when all constraints are met. If the selected constraints are too rigorous, the algorithm will not


be able to provide a solution, i.e., there will be no feasible solution to the mathematical problem. The constraints that determine the objective function’s capacity to be minimised or


maximised (i.e. those conditions fulfilled by 100% in relation to its predetermined limit) are considered “active constraints” [31]. Linear optimisation was performed with the CBC (COIN-OR


Branch and Cut) Solver algorithm, which is part of the Excel® 2016 software add-in OpenSolver, V. 2.9.0 [32]. We optimised the average diet of the total study sample (_n_ = 1797, i.e. the


“TotPop” diet) as well as the diet of the three clusters (Table 1), respectively. The relative deviation (RD) from the reported intake of each food item was calculated as RD (_w_opt –


_w_rep)/_w_rep, where wopt is the food weight in the optimised diet and _w_rep is the reported intake. As the objective function of all LP models, we chose the minimisation of the total


relative deviation (TRD) from the baseline diet [10, 11]. This objective function was implemented to maximise the similarity between the baseline and the optimised diet solutions. The


decision variables were the amounts of individual food items in the total study sample/each cluster. All optimisations applied dietary reference values (DRVs), covering the nutritional needs


of 97.5% of the population and based on the Nordic Nutrition Recommendations 2012 [33], as obligatory constraints (Supplementary Table 2). In cases where the DRVs differed depending on sex,


the nutritional constraints were weighted according to the DRVs and population size of the sex groups in the study sample. Total daily energy (kcal) was set to equal the baseline energy


intake within the total population/the three clusters in all models (Supplementary Table 2). All models were also constrained to meet the Swedish Food Based Dietary Guidelines (FBDGs) (Table


1) [34]. Individual food items were allowed to be reduced to 0 g; however, they were not allowed to increase by more than 200% relative to their respective baseline weight. This constraint


was applied to all foods except for the ones belonging to the food groups Pulses, Nuts and seeds, Dairy substitutes, Meat substitutes and Vegetable oils. Because of their plausible role in


making up a healthy and environmentally friendly diet and their partly recent appearance on the market, these foods/food groups were allowed to increase by any value. In a first set of


models, all aforementioned constraints, but no upper threshold for the associated GHGE, were applied. The second set of models also included a limit for total diet-related CO2eq. These


models were constrained to contain less than or equal to 1570 g of CO2eq per day. The cost of the baseline and optimised diets was calculated separately and was not included as a constraint


in the models. The average relative deviation (ARD) from the baseline food consumption (i.e., the TRD divided by the total number of food items included in the model) was calculated as an


output and used as a proxy of similarity between the baseline and the optimised food consumption and as an assumed indicator of cultural acceptability. Active nutrient constraints (those


meeting exactly 100% of the applied limit [31]) were identified for each solution. A more detailed description of the optimisation procedure can be found in the Supplementary Information.


RESULTS IDENTIFYING PREVALENT DIETARY CLUSTERS The cluster analysis resulted in three diet clusters roughly balanced in size (707, 534 and 556 individuals in clusters 1, 2 and 3


respectively). Supplementary Fig. 2 displays the hierarchical relationships between study participants. The three clusters differed significantly in their median daily consumption (g/MJ) of


all food groups part of the cluster analysis, median daily dietary CO2eq (g/MJ), median yearly income, mean age, and sex distribution (Supplementary Tables 3 and 4). Based on these observed


differences, the following classification of the clusters was made: * Cluster 1 – “the Classic Baseline diet”: High inclusion of foods of a typical Swedish diet (red and processed meat, and


potatoes), low inclusion of fruits and vegetables, high CO2eq emission, medium SHEIA15 (Swedish healthy eating index) * Cluster 2 – “the NutRich Baseline diet”: High inclusion of nutrient


dense animal products, nuts and vegetables, highest CO2eq emission, high SHEIA15 * Cluster 3 – “the LowClim Baseline diet”: High inclusion of low GHGE-foods with favourable nutritional


properties (vegetables, pulses) and, to some extent, less favourable (sugar and sweets), lowest CO2eq emission, high SHEIA15 BASELINE DIETS The CO2eq emissions of the baseline diets ranged


between 2770 (LowClim Baseline) and 3361 (Classic Baseline) g/day (Table 2). All baseline diets contained lower than recommended amounts of carbohydrates, dietary fibre, and iron


(Supplementary Table 5). They were also lower than recommended with respect to the DRV for vitamin D, except for the LowClim Baseline diet which met this DRV by 100%. All baseline diets


exceeded the recommended amounts of saturated fatty acids and sodium (Supplementary Table 5). The cost of the four baseline diets ranged between SEK 65 and 68 (approximately 6.5 


USD/person/day) (Table 1). OPTIMISED DIETS In the optimised isocaloric diets meeting DRVs and the Swedish FBDGs only (TotPop, Classic, NutRich and LowClim models), GHGE were reduced by 8–24%


compared with the baseline diets (Table 2). The cost increased slightly (~1–3%), and average relative deviations (ARDs) were low (~4%) for most of these diets. The exception was the Classic


diet, which had a marginally lower (−1%) cost and an ARD of about 20%. The number of foods removed, reduced or increased was fairly similar across the optimised diets. However, more foods


in the Classic diet were changed compared to the other ones. Adding the upper CO2 constraint of 1.57 kg CO2eq/person/day [17] (TotPop+, Classic+, NutRich+ and LowClim+ models) reduced


diet-related GHGE by 43–53% (Table 2). Compared to baseline, the diet cost was reduced approximately by 8–13% in all these optimised diets (Table 2). The inclusion of the CO2eq constraint


increased the ARDs only slightly for all diets, ranging from 5.8 % in the LowClim+ diet to 22.8% in the Classic+ diet. All optimised diets constrained to meet nutritional, FBDG and CO2eq


targets had lower shares of animal-based foods (Fig. 1). The Classic+ diet contained 82% less Red meat, 81% less Processed meat, 62% less Poultry, and only about one third of the Dairy


compared to its baseline amounts (Fig. 1). The TotPop+, NutRich+ and LowClim+ diets also contained considerably less Red/Processed meat. In contrast to the Classic+ diet, the other optimised


diets did not show increases in Seafood (Fig. 1). The optimised diets contained higher amounts of Vegetables (+6 to +159%), Potatoes (+106 to +131%), and Fruits and berries (+127 to +183%).


The greatest changes in Cereals/grains occurred in the Total+ diet (+56%) whereas the LowClim+ diet experienced only a moderate change ( + 8%) (Fig. 1). Rice was reduced by ~70% in all


optimised diets except for the LowClim+ diet, where this food group remained unchanged. A noticeable (15-fold) increase in Pulses was observed in the Classic+ diet only. A more detailed


presentation of each food group associated with the baseline and/or optimised clusters is found in Supplementary Tables 6–10. Iron and vitamin D were active lower-threshold constraints while


added sugars and sodium were active upper-threshold active constraints in almost all models (Supplementary Table 5). OPTIMISATION OF TOTAL DIET VS. CLUSTERING APPROACH Figure 2 was


developed to explore whether a diet optimised based on the average diet of the entire sample would result in a dietary pattern equal to the diets of the optimised clusters. Figure 2


illustrates how much each of the optimised cluster diets (Classic+, NutRich+, and LowClim+) differ from the diet optimised based on the average intake of the total population (TotPop +).


Values indicate the absolute difference between the baseline vs. optimised energy-adjusted intake (g/MJ/day) of different food groups—i.e., the dietary change resulting from optimisation—in


the TotPop+ model compared against the dietary change resulting from optimisation in each cluster. For example, the TotPop+ model requires an increase in cereal consumption of 10.5 g/MJ/day


whereas individuals belonging to the Classic cluster need to increase their Cereal intake by only 7.5 g/MJ/day. Hence, the resulting difference (−3 g/MJ/day) is shown in the graph. Overall,


the three cluster-specific diets face dietary shifts that differ from those demanded by the TotPop+ model. OPTIMISED DIETS VS. THE EAT-LANCET DIET Overall, the EAT-Lancet diet was higher in


Whole grain foods, Dairy, Poultry, Legumes, Nuts, and Added fats, but lower in Potatoes, Fruits, Red/processed meat, Eggs, Fish and Added sugars than that provided by the optimised diets and


expressed as a percentage of total energy intake (Fig. 3). However, all optimised diets matched the EAT-Lancet diet with regards to Vegetables. The NutRich+ diet was close to matching the


EAT-Lancet diet in terms of Added Sugars whereas the LowClim+ diet was closest with respect to Whole grains. The NutRich+ as well as LowClim+ diets also aligned well with the EAT-Lancet diet


in terms of Dairy foods. DISCUSSION In this study we demonstrated that the combination of cluster analysis with linear optimisation can provide guidance to nutritionally adequate,


health-promoting, affordable and climate-friendly diets for different self-selected dietary patterns for the Swedish Population. Our findings show that the three optimised cluster-specific


diets differed significantly from the model optimising the average diet of the total population. This novel modelling approach for a climate-friendly and healthy diet may therefore be


preferred as it is more consumer oriented. Optimising diets to meet nutritional recommendations and Swedish FBDGs reduced the GHGE by up to 24%. However, this reduction is not sufficient to


keep diets within planetary boundaries for climate change. To achieve this goal, the GHGE of the diets would have to be reduced by half compared to baseline. If extrapolating these


reductions to the entire adult population in Sweden (~10.4 million), our optimised diets could reduce domestic annual emissions from agricultural food production by roughly 33%, from 6.9 MT


[35] to about 4.6 MT. One important strength of our approach is that it leaves a considerable part of the baseline food consumption unchanged while at the same time also reducing cost. The


latter might be an additional argument to change diets in times of quickly rising food prices, for example as a result of the 2022 energy crisis. Similar to what others have found [8, 9,


36,37,38], the changes seen for all optimised diets were predominantly characterised by shifts from animal products such as red/processed meat, poultry and dairy to plant-based foods such as


fruits, vegetables and cereals/grains, albeit to varying degrees depending on the cluster. Particularly, the Classic Baseline pattern had to undergo the most pronounced changes compared to


the other two clusters to reach the proposed recommendations and requirements (Fig. 1 and Supplementary Tables 7–10). Besides differing between each other, our findings also show that the


three cluster-specific diets (Classic+, NutRich+ and LowClim+) would imply overall dietary shifts that differ from those demanded by the TotPop+ model (Fig. 2). Our results thus indicate


that a clustering-optimisation strategy is likely to better capture the dietary heterogeneity that may exist within a delimited context [39]. It is possible that individuals advised to


follow a diet that is based on their own specific cluster is more acceptable and thus realistic than a diet optimised on the basis of the national average diet. A similar approach to capture


dietary heterogeneity has been applied in the Netherlands [40] where linear programming was used to develop sustainable FBDGs for groups of individuals who consumed meat or not. As the


cluster-based optimisation approach considers group-specific preferences, it may make dietary behavioural change more efficient, e.g. by tailoring recommendations/advice to different


segments in the population. Naturally, these tailored recommendations should include EER values that may deviate from those calculated for the single clusters. Whether these findings could


increase the level of acceptance for climate-friendly diets tailored to different clusters/subgroups in the population remains to be investigated. The nutritious and health-promoting diets


in models TotPop, Classic, NutRich and LowClim were up to 24% lower in GHGE compared to baseline diets. The reduced climate impact from achieving nutritional and health goals aligns with


findings from previous research [10, 12, 41, 42]. Yet, our study also shows that switching to a diet meeting only DRVs and the current Swedish FBDGs is not sufficient to keep the climate


impact of Swedish diets below the IPCC-suggested CO2eq threshold. Such diets were only achievable if the defined GHGE constraint was added to the models (TotPop+, Classic+, NutRich+,


LowClim+). As a result, the cost decreased while our proxy for cultural acceptability (the ARD) changed only marginally compared to that observed in the models without a CO2eq constraint. In


fact, only 5–12% of the foods were changed (either increased/reduced/removed) in the CO2eq-constrained diets compared to the baseline diet, indicating that acceptance among consumers within


each dietary cluster could be high. In contrast to other studies from Brazil [43], the US [44], Denmark [29] and Ghana [45] where diets were optimised only to meet nutritional


recommendations and FBDGs, the cost of our climate-optimised diets dropped below that of the baseline diet, contradicting assumptions that a healthy, climate-friendly diet is more costly


than prevailing food patterns [46] and confirming previous modelling studies indicating lower cost of sustainable nutrition in high-income countries [47]. Our findings reveal that the


optimised diets did not align very well with the EAT-Lancet Commission’s dietary recommendation on a sustainable diet. These discrepancies may have several explanations. Firstly, our


LP-modelling approach addresses aspects such as a nutrient adequacy (by ensuring the fulfilment of 27 DRVs and the Swedish FBDGs), a shortcoming of the EAT-Lancet diet that already


previously has been addressed [48]. Secondly, we implemented dimensions of cultural acceptability (by minimising the TRD and constraining the RD of individual food items) as well as


affordability. These aspects are not reported to have been addressed during the design process of the EAT-Lancet diet. Secondly, the food categorisation in the Riksmaten survey includes


mixed dishes (wherein e.g. added fats can be “hidden”) whereas the EAT-Lancet diet is composed of “basic” food groups. Hence, the food groups used in Riksmaten are not fully comparable with


the EAT-Lancet reference diet’s food groups. Thirdly, in contrast to the optimised diets at hand, the EAT-Lancet diet was developed aiming at health promotion and evaluated against other


environmental factors besides GHGE such as water footprint, land use change, and biodiversity. Lastly, the EAT-Lancet diet was developed as a global reference diet and was thus not tailored


to a specific national or cultural context. In fact, the authors behind this diet call for cultural and regional adaptations of the dietary recommendations [1]. Hence, the modelling strategy


suggested here may be seen as a novel and complementary approach to achieve a cultural tailoring of the EAT-Lancet diet to several distinct subgroups of dietary patterns within a


population. This study assessed the environmental impact of the Swedish diets only on the basis of GHGE, other relevant characteristics of environmental sustainability in the context of


diets such as eco-toxicity, land use change, water use, eutrophication, acidification, animal welfare and biodiversity loss were not included due to lack of detailed data for Sweden. Not


including these aspects is a limitation since different foods vary in their environmental impacts [49]; animal products tend to be the most GHGE-intense while staple crops (for human


consumption), fruits and vegetables, generally are the main contributors to freshwater use per kg of food. However, a drop in GHGE of diets has been observed to be accompanied by substantial


reductions in land use and water footprint [50]. Although this study used only the GHGE as an active environmental constraint, it can be assumed that the associated land use and water


footprint of the optimised diets are considerably smaller compared to the observed diet. Our LP modelling did not include foods that were not already present in the baseline diets. There are


various new, climate-friendly meat/dairy replacements emerging on the market; many of them fortified with nutrients such as vitamins B12, D and calcium [51, 52]. These are nutrients that


tend to be insufficient in plant-based diets. Allowing for these foods to be chosen by the LP-algorithm could be an alternative path to providing climate- and nutrient efficient foods with


sensory traits similar to those of animal products. Future optimisations could therefore explore the effects of also including such foods in the modelling as a way to deliver nutritious,


climate-friendly and acceptable diet solutions. This study shows that this novel modelling approach is useful for integrating goals of nutrition, health promotion, climate friendliness and


cultural acceptability for different self-selected dietary patterns. Switching to a diet following current nutritional recommendations and Swedish FBDGs is not sufficient to stay below the


IPCC CO2eq threshold. The fully optimised diets remain within planetary boundaries for climate change while leaving a considerable part of diet unchanged and being lower in cost, suggesting


that acceptance among consumers could be high. This is based on the assumption that similarity to existing diets is a predictor of cultural acceptability. The changes seen for all diets were


predominantly characterised by shifts from animal products to plant-based foods. However, the shifts required to meet nutrient, FBDG and CO2eq constraints varied between the dietary


clusters as well as in comparison to the diet optimised for the total population. This suggests that explorative cluster analysis combined with LP is likely to propose dietary shifts that


are easier to achieve across a broader range of consumers. The nutritionally adequate, health-promoting and climate-friendly diets in this study did, in various aspects, not match the


EAT-Lancet diet. This indicates that there are several approaches through which sustainable diets can be defined, but also that the cultural dietary context plays a bearing role in the


optimisation of such diets for specific populations. This study may offer policymakers with insights into how both health promotion and environmental protection may become better connected


and thus plausibly also more effective. DATA AVAILABILITY Data can be found within the published article and its supplementary files. Requests for additional materials should be addressed to


PEC. REFERENCES * Willett W, Rockström J, Loken B, et al. Food in the Anthropocene: the EAT–Lancet Commission on healthy diets from sustainable food systems. Lancet. 2019;393:447–92.


Article  PubMed  Google Scholar  * Government Offices of Sweden. Sweden ratifies the climate agreement from Paris. 2016.


https://www.regeringen.se/pressmeddelanden/2016/10/sverige-ratificerar-klimatavtalet-fran-paris/ * Sarkar A. Addressing consumerism and the planetary health crisis: behavioral economics


approach in public policy. Front Energy Res. 2022;10:800588. Article  Google Scholar  * Béné C, Fanzo J, Haddad L, Hawkes C, Caron P, Vermeulen S, et al. Five priorities to operationalize


the EAT–Lancet Commission report. Nat Food. 2020;1:457–9. Article  PubMed  Google Scholar  * Springmann M, Wiebe K, Mason-D’Croz D, Sulser TB, Rayner M, Scarborough P. Health and nutritional


aspects of sustainable diet strategies and their association with environmental impacts: a global modelling analysis with country-level detail. Lancet Planet Health. 2018;2:e451–61. Article


  PubMed  PubMed Central  Google Scholar  * Gazan R, Brouzes CMC, Vieux F, Maillot M, Lluch A, Darmon N. Mathematical optimization to explore tomorrow’s sustainable diets: a narrative


review. Adv Nutr. 2018;9:602–16. Article  PubMed  PubMed Central  Google Scholar  * Mertens E, Van’t Veer P, Hiddink GJ, Steijns JM, Kuijsten A. Operationalising the health aspects of


sustainable diets: a review. Public Health Nutr. 2017;20:739–57. Article  PubMed  Google Scholar  * Perignon M, Masset G, Ferrari G, Barré T, Vieux F, Maillot M, et al. How low can dietary


greenhouse gas emissions be reduced without impairing nutritional adequacy, affordability and acceptability of the diet? A modelling study to guide sustainable food choices. Public Health


Nutr. 2016;19:2662–74. Article  PubMed  PubMed Central  Google Scholar  * Milner J, Green R, Dangour AD, Haines A, Chalabi Z, Spadaro J, et al. Health effects of adopting low greenhouse gas


emission diets in the UK. BMJ Open. 2015;5:e007364. Article  PubMed  PubMed Central  Google Scholar  * Eustachio Colombo P, Patterson E, Elinder LS, Lindroos AK, Sonesson U, Darmon N, et al.


Optimizing school food supply: integrating environmental, health, economic, and cultural dimensions of diet sustainability with linear programming. Int J Environ Res Public Health.


2019;16:3019. Article  PubMed  PubMed Central  Google Scholar  * Darmon N, Ferguson EL, Briend A. A cost constraint alone has adverse effects on food selection and nutrient density: an


analysis of human diets by linear programming. J Nutr. 2002;132:3764–71. Article  CAS  PubMed  Google Scholar  * Horgan GW, Perrin A, Whybrow S, Macdiarmid JI. Achieving dietary


recommendations and reducing greenhouse gas emissions: modelling diets to minimise the change from current intakes. Int J Behav Nutr Phys Act. 2016;13:46. Article  PubMed  PubMed Central 


Google Scholar  * Maillot M, Vieux F, Amiot MJ, Darmon N. Individual diet modeling translates nutrient recommendations into realistic and individual-specific food choices. Am J Clin Nutr.


2010;91:421–30. Article  CAS  PubMed  Google Scholar  * Lluch A, Maillot M, Gazan R, Vieux F, Delaere F, Vaudaine S, et al. Individual diet modeling shows how to balance the diet of french


adults with or without excessive free sugar intakes. Nutrients. 2017;9:162. Article  PubMed  PubMed Central  Google Scholar  * Cocking C, Walton J, Kehoe L, Cashman KD, Flynn A. The role of


meat in the European diet: current state of knowledge on dietary recommendations, intakes and contribution to energy and nutrient intakes and status. Nutr Res Rev. 2020;33:181–9. Article 


PubMed  Google Scholar  * Gibbons H, Carr E, McNulty BA, Nugent AP, Walton J, Flynn A, et al. Metabolomic-based identification of clusters that reflect dietary patterns. Mol Nutr Food Res.


2017;61:1601050. Article  Google Scholar  * World Wildlife Fund. One Planet Plate 2019 – kriterier och bakgrund (One Planet Plate 2019 – criteria and background). 2019.


https://wwwwwfse.cdn.triggerfish.cloud/uploads/2019/04/kriterier-fr-one-planet-plate-rev-2019.pdf. Accessed 30 Jan 2020. * Amcoff E, Sverige, L. Riksmaten - vuxna 2010-11 Livsmedels- och


näringsintag bland vuxna i Sverige. Uppsala: Livsmedelsverket; 2012. * Livsmedelsverket. https://www.livsmedelsverket.se/om-oss/psidata/apimatvanor. Accessed 10 Apr 2020. * Florén B, Amani


P, Davis J. Climate database facilitating climate smart meal planning for the public sector in Sweden. Int J Food Syst Dyn. 2017;8:72–80. Google Scholar  * International Organization for


Standardization. ISO 14040:2006 - Environmental management -- Life cycle assessment -- Principles and framework. https://www.iso.org/standard/37456.html. Accessed 9 Oct 2017. * International


Organization for Standardization. ISO 14044:2006 - Environmental management -- Life cycle assessment -- Requirements and guidelines. https://www.iso.org/standard/38498.html. Accessed 9 Oct


2017. * Parry ML, Intergovernmental Panel on Climate Change (eds). Climate change 2007: impacts, adaptation and vulnerability: contribution of Working Group II to the fourth assessment


report of the Intergovernmental Panel on Climate Change. Cambridge: Cambridge University Press; 2007. * Matpriskollen (Food price check). https://matpriskollen.se/. Accessed 27 Jan 2020. *


Brock G, Pihur V, Datta S, Datta S. clValid: an R package for cluster validation. J Stat Softw. 2008;24:1–22. Google Scholar  * Charrad M, Ghazzali N, Boiteau V. NbClust: an R package for


determining the relevant number of clusters in a data set. J Stat Softw. 2014;61:1–36. Article  Google Scholar  * R Core Team. R: a language and environment for statistical computing; R


Foundation for Statistical Computing, Vienna, Austria; 2021. * Moraeus L, Lindroos AK, Warensjö Lemming E, Mattisson I. Diet diversity score and healthy eating index in relation to diet


quality and socio-demographic factors: results from a cross-sectional national dietary survey of Swedish adolescents. Public Health Nutr. 2020;23:1754–65. Article  PubMed  PubMed Central 


Google Scholar  * Parlesak A, Tetens I, Dejgard Jensen J, Smed S, Gabrijelcic Blenkus M, Rayner M, et al. Use of linear programming to develop cost-minimized nutritionally adequate health


promoting food baskets. PLoS ONE. 2016;11:e0163411. Article  PubMed  PubMed Central  Google Scholar  * Dantzig GB (1951). Maximization of a linear function of variables subject to linear


inequality. In: Koopmans TC editors. Activity Analysis of Production and Allocation. New York: Wiley; 1947. p. 339–47. * Nocedal J, Wright SJ. Numerical optimization. New York: Springer;


2006. * Mason AJ. OpenSolver - an open source add-in to solve linear and integer progammes in Excel. In: Klatte D, Lüthi H-J, Schmedders K, editors. Operations Research Proceedings 2011.


Berlin: Springer; 2012. p. 401–6. * Nordisk Ministerråd. Nordic nutrition recommendations 2012. 5th ed. Copenhagen: Nordic Council of Ministers; 2014. * The Swedish dietary guidelines: find


your way to eat greener, not too much and be active. The Swedish Food Agency. Uppsala; 2017. * Territorial emissions and uptake of greenhouse gases.


https://www.naturvardsverket.se/data-och-statistik/klimat/vaxthusgaser-territoriella-utslapp-och-upptag. Accessed 15 Mar 2022. * Macdiarmid JI, Kyle J, Horgan GW, Loe J, Fyfe C, Johnstone A,


et al. Sustainable diets for the future: can we contribute to reducing greenhouse gas emissions by eating a healthy diet? Am J Clin Nutr. 2012;96:632–9. Article  CAS  PubMed  Google Scholar


  * Reynolds CJ, Horgan GW, Whybrow S, Macdiarmid JI. Healthy and sustainable diets that meet greenhouse gas emission reduction targets and are affordable for different income groups in the


UK. Public Health Nutr. 2019;22:1503–17. Article  PubMed  PubMed Central  Google Scholar  * Broekema R, Tyszler M, van ‘t Veer P, Kok FJ, Martin A, Lluch A, Blonk HTJ, et al. Future-proof


and sustainable healthy diets based on current eating patterns in the Netherlands. Am J Clin Nutr. 2020;12:1338–47. Article  Google Scholar  * Vieux F, Perignon M, Gazan R, Darmon N. Dietary


changes needed to improve diet sustainability: are they similar across Europe? Eur J Clin Nutr. 2018;72:951–60. Article  PubMed  PubMed Central  Google Scholar  * Brink E, van Rossum C,


Postma-Smeets A, Stafleu A, Wolvers D, van Dooren C, et al. Development of healthy and sustainable food-based dietary guidelines for the Netherlands. Public Health Nutr. 2019;22:2419–35.


Article  PubMed  PubMed Central  Google Scholar  * Green R, Milner J, Dangour AD, Haines A, Chalabi Z, Markandya A, et al. The potential to reduce greenhouse gas emissions in the UK through


healthy and realistic dietary change. Clim Change. 2015;129:253–65. Article  ADS  Google Scholar  * Eustachio Colombo P, Elinder LS, Lindroos AK, Parlesak A. Designing nutritionally adequate


and climate-friendly diets for omnivorous, pescatarian, vegetarian and vegan adolescents in sweden using linear optimization. Nutrients. 2021;13:2507. Article  CAS  PubMed  PubMed Central 


Google Scholar  * Verly-Jr E, de Carvalho AM, Marchioni DML, Darmon N. The cost of eating more sustainable diets: A nutritional and environmental diet optimisation study. Glob Public Health.


2021;17:1–14. Google Scholar  * Maillot M, Drewnowski A. Energy allowances for solid fats and added sugars in nutritionally adequate U.S. diets estimated at 17–33% by a linear programming


model. J Nutr. 2011;141:333–40. Article  CAS  PubMed  Google Scholar  * Nykanen E-PA, Dunning HE, Aryeetey RNO, Robertson A, Parlesak A. Nutritionally optimized, culturally acceptable,


cost-minimized diets for low income ghanaian families using linear programming. Nutrients. 2018;10:461. Article  PubMed  PubMed Central  Google Scholar  * Barosh L, Friel S, Engelhardt K,


Chan L. The cost of a healthy and sustainable diet – who can afford it? Aust N Z J Public Health. 2014;38:7–12. Article  PubMed  Google Scholar  * Springmann M, Clark MA, Rayner M,


Scarborough P, Webb P. The global and regional costs of healthy and sustainable dietary patterns: a modelling study. Lancet Planet Health. 2021;5:e797–807. Article  PubMed  PubMed Central 


Google Scholar  * Beal T, Ortenzi F, Fanzo J. Estimated micronutrient shortfalls of the EAT–Lancet planetary health diet. Lancet Planet Health. 2023;7:e233–37. Article  PubMed  Google


Scholar  * Springmann M, Clark M, Mason-D’Croz D, et al. Options for keeping the food system within environmental limits. Nature. 2018;562:519. Article  ADS  CAS  PubMed  Google Scholar  *


Aleksandrowicz L, Green R, Joy EJM, Smith P, Haines A. The impacts of dietary change on greenhouse gas emissions, land use, water use, and health: a systematic review. PLoS ONE.


2016;11:e0165797. Article  PubMed  PubMed Central  Google Scholar  * Franco D, Martins AJ, López-Pedrouso M, Purriños L, Cerqueira MA, Vicente AA, et al. Strategy towards replacing pork


backfat with a linseed oleogel in Frankfurter sausages and its evaluation on physicochemical, nutritional, and sensory characteristics. Foods. 2019;8:366. Article  CAS  PubMed  PubMed


Central  Google Scholar  * Heck RT, Fagundes MB, Cichoski AJ, de Menezes CR, Barin JS, Lorenzo JM, et al. Volatile compounds and sensory profile of burgers with 50% fat replacement by


microparticles of chia oil enriched with rosemary. Meat Sci. 2019;148:164–70. Article  CAS  PubMed  Google Scholar  * Lag om etikprövning av forskning som avser människor (SFS 2003:460) [Law


on ethical review of research concerning humans (SFS 2003:460)]. Stockholm: Utbildningsdepartementet; 2003. Download references FUNDING The contribution by all authors was funded by the


Swedish Research Council FORMAS (grant number 2016-00353). The funder had no role in the study design, data analysis or writing, or the decision to submit for publication. Open access


funding provided by Karolinska Institute. AUTHOR INFORMATION AUTHORS AND AFFILIATIONS * Department of Global Public Health, Karolinska Institutet, Stockholm, Sweden Patricia Eustachio


Colombo, Liselotte Schäfer Elinder, Esa-Pekka A. Nykänen & Emma Patterson * Centre on Climate Change and Planetary Health, London School of Hygiene and Tropical Medicine, WC1E 7HT,


London, UK Patricia Eustachio Colombo * Centre for Epidemiology and Community Medicine, Region Stockholm, Stockholm, Sweden Liselotte Schäfer Elinder * Functional Foods Forum, University of


Turku, Turku, Finland Esa-Pekka A. Nykänen * The Swedish Food Agency, Uppsala, Sweden Emma Patterson & Anna Karin Lindroos * Department of Internal Medicine and Clinical Nutrition, the


Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden Anna Karin Lindroos * Department of Nutrition, Exercise and Sports, Copenhagen University, Copenhagen, Denmark Alexandr


Parlesak * Personalized Nutrition, Duale Hochschule Baden-Württemberg, Heilbronn, Germany Alexandr Parlesak Authors * Patricia Eustachio Colombo View author publications You can also search


for this author inPubMed Google Scholar * Liselotte Schäfer Elinder View author publications You can also search for this author inPubMed Google Scholar * Esa-Pekka A. Nykänen View author


publications You can also search for this author inPubMed Google Scholar * Emma Patterson View author publications You can also search for this author inPubMed Google Scholar * Anna Karin


Lindroos View author publications You can also search for this author inPubMed Google Scholar * Alexandr Parlesak View author publications You can also search for this author inPubMed Google


Scholar CONTRIBUTIONS PEC contributed to the conceptualisation and design of the study, the data analysis, presentation, interpretation of the results, as well as drafted and edited the


manuscript. LSE contributed to the conceptualisation and design of the study, and to the critical revising of the manuscript. EPN contributed to the conceptualisation and design of the


study, data curation, and to the critical revising of the manuscript. EP contributed to the conceptualisation and design of the study, and to the critical revising of the manuscript. AKL


provided data, contributed to the conceptualisation and design of the study, and to the critical revising of the manuscript. AP maintained study oversight, contributed to the


conceptualisation and design of the study, and to the critical revising of the manuscript. The corresponding author attests that all listed authors meet authorship criteria and that no


others meeting the criteria have been omitted. All authors approved the final article version to be submitted. CORRESPONDING AUTHOR Correspondence to Patricia Eustachio Colombo. ETHICS


DECLARATIONS COMPETING INTERESTS The authors declare no competing interests. ETHICAL APPROVAL Ethical approval for the original Riksmaten vuxna 2010–11 dietary survey was granted by the


Regional Ethical Review Board in Uppsala. This data is now fully anonymized and publicly available and so the current study involved no personal data. Ethical approval was therefore not


required for this study in accordance with Swedish law [53]. ADDITIONAL INFORMATION PUBLISHER’S NOTE Springer Nature remains neutral with regard to jurisdictional claims in published maps


and institutional affiliations. SUPPLEMENTARY INFORMATION SUPPLEMENTARY INFORMATION RIGHTS AND PERMISSIONS OPEN ACCESS This article is licensed under a Creative Commons Attribution 4.0


International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the


source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative


Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by


statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit


http://creativecommons.org/licenses/by/4.0/. Reprints and permissions ABOUT THIS ARTICLE CITE THIS ARTICLE Eustachio Colombo, P., Elinder, L.S., Nykänen, EP.A. _et al._ Developing a novel


optimisation approach for keeping heterogeneous diets healthy and within planetary boundaries for climate change. _Eur J Clin Nutr_ 78, 193–201 (2024).


https://doi.org/10.1038/s41430-023-01368-7 Download citation * Received: 10 January 2023 * Revised: 02 November 2023 * Accepted: 08 November 2023 * Published: 21 November 2023 * Issue Date:


March 2024 * DOI: https://doi.org/10.1038/s41430-023-01368-7 SHARE THIS ARTICLE Anyone you share the following link with will be able to read this content: Get shareable link Sorry, a


shareable link is not currently available for this article. Copy to clipboard Provided by the Springer Nature SharedIt content-sharing initiative