1. Introduction
The food system is highly related to the Sustainable Development Goals (SDGs) involving hunger, nutrition and health, climate change, natural resources, biodiversity and socioeconomics [
1,
2]. Agriculture is the largest consumer of fresh water and the second largest contributor to greenhouse gas emissions worldwide [
3,
4]. Food consumption accounts for 20–30% of the environmental burden of total consumption in Europe [
5]. At the same time, non-communicable diseases (NCD) or chronic diseases, accounted for 71% of all deaths globally in 2015 [
6], while dietary risks (e.g., high consumption of red meat) and metabolic risks (e.g., high body-mass index) are predominantly associated with chronic disease burden. Diet plays a central role in the health of the population. There is an urgent need to limit environmental damage in order not to transgress the planetary boundaries [
7] while providing a nutritious diet to a growing world population.
Previous studies have found that demand side interventions such as dietary shift (e.g., towards vegan, vegetarian, Mediterranean diets) can have positive consequences for both human health and climate change [
8,
9,
10,
11,
12,
13]. However, others have found that shifting from current to vegetarian, vegan or even flexitarian diets carries the risk for deficiency of certain micronutrients that are currently supplied primarily through animal-based products [
14,
15]. In general, shifting to plant-based diets not only results in greenhouse gas emissions reduction but also other environmental benefits such as water savings [
16,
17,
18,
19].
Recent evidence shows that different dietary scenarios can have widely varying consequences on different dimensions of sustainability [
20]. It is therefore important to explore multiple scenarios of dietary change employing multiple indicators of sustainability in order to identify potential trade-offs and get a holistic and comprehensive picture of the consequences of dietary change.
Studies calculating more than one indicator of dietary sustainability are emerging [
14,
21,
22,
23,
24,
25,
26]. However, they are either restricted to a particular environmental aspect (e.g., greenhouse gas (GHG) emissions), a limited number of nutrient (e.g., calorie, protein) or health aspect while not taking into account intake of micronutrients, health effects of multiple food groups (e.g., pulses) or ignoring certain health risk-outcomes (e.g., obesity).
Recent years have seen development of new approaches to quantify the nutritional quality scores of individual diets taking into account the intake of essential and harmful nutrients and comparing it to their recommended levels [
14,
27,
28]. Assessment of human health consequences due to a shift in food consumption behaviours is also increasingly becoming common due to availability of food group-specific risk factors from long-term epidemiological studies [
29]. Regarding the environmental dimension, the emission factors and footprint values for different food items compiled through meta-analysis of life cycle assessment studies are now available enabling the calculation of damage across multiple environmental categories [
20,
30]. While dietary choices determine health and environmental outcomes, they in turn are driven by food cost especially in low-middle income countries [
31]. Healthy and environmental-friendly diet from sustainable food systems should be made economically affordable for communities. A reduced household expenditure on daily food purchase can help and drive the dietary shift.
In terms of various dietary change scenarios often employed, vegetarian and vegan diets are the most common ones. While the vegan diet excludes all animal-sourced foods, the vegetarian diet can include dairy products and an ovo-lacto vegetarian diet can include both egg and dairy products. Another variant, the pescatarian diet, excludes meat but can include fish. Much more relaxed than above, a flexitarian or semi-vegetarian diet, is a predominantly plant-based diet with the occasional inclusion of meat or fish [
32].
Few studies to date have integrated environment, nutrition, economic and human health indicators to assess the consequences of different dietary change scenarios. The aim of this study is to fill this research gap and demonstrate the calculation of multiple indicators of sustainability for a suite of dietary change scenarios through a case study of Switzerland and explore the potential synergies and trade-offs across different indicators.
4. Discussion
Our multi-indicator quantitative comparison of current and nine alternative diets for the Swiss population revealed that transitioning towards a healthy diet following the guidelines of Swiss society of nutrition will bring the highest sustainability benefits as it is projected to reduce 36% of the environmental footprint, save one third of expenditure on food and lower the adverse health outcome by 2.67% compared to the current diet (
Table 2). Achieving this sustainable diet would entail a high reduction in the intake of meat and vegetable oils, a moderate reduction in cereals, roots and fish products and at the same time increased intake of legumes, nuts, seeds, fruits and vegetables (
Table 1).
On the other extreme, our modelling results show that transitioning towards a meat or protein oriented diet is the least sustainable option out of all 10 diets considered here as it is projected to result in large increases in diet related adverse health outcomes, environmental footprint, daily food expenditure and low intakes in essential nutrient with risk of leading to deficiency (for vitamin C, fibre, potassium and calcium;
Table 2 and
Table 3).
Overall, our results are in line with other recent studies suggesting that shifting to a healthy diet is not only good for human health but also for the environment [
2,
8,
18,
24,
29] and that replacing meat with plant-based alternatives such as legumes, improves the nutrition quality and reduces the environmental footprints [
14]. This is also confirmed by our correlation analysis (
Table 5) where we found a significant negative correlation between health benefits and environmental footprints suggesting that healthier and nutrient dense diet can be in line with the goal of environmental sustainability. In addition, results show that daily food expenditure is positively correlated with environmental footprints and negatively correlated with health and nutrient density indicators. In other words, a healthy & nutritious diet that has low impact on the environment need not be expensive. It can be explained by the high meat price in Switzerland and shifting from current to nutritious diet (e.g., RSN scenario) entails cutting down on meat intake, ultimately leading to economic savings as the meat replacement products such as legumes and nuts are cheaper than meat. This is encouraging evidence showing a synergy between environmental, health and nutritional sustainability.
Our analysis presents several other insights useful from a sustainability point of view. For example, we found that while adopting a vegetarian or vegan diet has many benefits such as reduced adverse health outcomes, food expenditure and greenhouse gas footprint (
Table 2), it might lead decreased intake of essential micronutrients (e.g., Vitamin B
12, Choline and Calcium;
Table 3). Such insights were only possible because we employed a total of 10 indicators of dietary sustainability over four dimensions (human health, nutrition, environment and economics). A multi-indicator analysis enabled us to figure out possible trade-offs not only across the four dimensions of sustainability considered here but also between the different indicators of the same dimension (e.g., GHG vs. water footprint for vegetarian diet).
In our comprehensive nutrition quality analysis, we compared the intake of 23 essential and 4 harmful nutrients with their daily recommended levels and found that even a Swiss diet meeting the essential nutrients and health recommendations (RSN; HGD) exceeds in the intake of harmful (disqualifying) nutrients such as fats, cholesterol and sugar (
Table 3). Therefore, the disqualifying nutrient score (DNS), is currently low at <10 for all scenarios meaning that foods high in these nutrients of health concern needs to be cut down. We also found that the daily recommended levels of four essential nutrients (calcium, potassium, fibre and choline) are barely met in the current Swiss diet (ratio ~ 1) and therefore, food containing these nutrients should be encouraged in daily diet (
Table 3).
In our environmental analysis, we found that the current greenhouse gas emission and nitrogen footprint of daily average Swiss diet exceeds the respective food related planetary boundaries, meaning that food items high in GHG emissions (e.g., beef, lamb, pork;
Supplementary Table S3) and requiring high fertilizer application (e.g., oil seeds, rice or grain-fed poultry, pork) should be discouraged in order to meet environmental sustainability goals.
Our analysis comes with several limitations and uncertainties that should be considered while interpreting the results. Below, we list these limitations and sources of uncertainties as well as research gaps that needs to be filled through future efforts.
First, we relied on the current Swiss diet data from the UN FAO’s database [
33] which does not provide any uncertainty around its estimates, might suffer from errors and only provides the food intake aggregated across 82 broad food groups for Switzerland. In practice, the Swiss population eats thousands of individual food items (processed, cooked and modified from raw agricultural commodities to a varying degree) differing widely in their environmental footprint and nutrition content. For example, our study predicts remarkable low intake of vitamin B
12 in the vegan diet scenario compared with current levels. However, people can obtain vitamin B
12 from plant foods such as edible algae [
62] or dietary supplements, which are not included in 82 food categories considered here. A more detailed dietary database along with the mean and standard deviation estimates of food consumption is needed to overcome the potential errors in the calculated results.
Second, the Swiss food composition database (
http://www.naehrwertdaten.ch/ or USDA [
48]) and the environmental emission factor database [
20] that we use, do not provide the uncertainty intervals around their estimates and thus, we could not propagate this to the end result on nutrition quality and environmental indicators. For the nutrition, we did not take into account the bioavailability or the interaction of nutrients with one another that might affect the actual absorption of nutrients into the body [
63].
For the environmental emission factors, we used the global average values that were publicly available but future studies should use the country-specific emission factor values. Also, we could not consider the impacts of dietary change on other environmental categories such as biodiversity loss, ecosystem services [
64,
65,
66,
67,
68,
69] or human toxicity through air/water pollution due to a lack of emission factors for all food items considered here.
Third, our health impact estimates (DALY values,
Table 2) are likely an underestimate because we only included the health impacts due to intake of six food groups (red meat, fruits, vegetables, legumes, fish, nuts & seeds) and were not able to model the health consequences of changes in consumption of other food groups or components such as dairy, vegetable oils, whole grains, dietary fibre and so forth. [
21,
70,
71,
72,
73,
74,
75]. We simply used the relative risk factors (RR) for these food groups because the epidemiological evidence on impact of these foods on human health is robust and statistically significant and has been employed recently in the global scale EAT-Lancet commission calculating dietary impacts on health [
2]. More research is required to come up with reliable data on risk of other food groups to human health and address other limitations of epidemiological research [
76].
The underestimation on health impact can also be attributed to the limited number of diseases considered in our model (CHD, T2D, cancer, stroke, obesity, overweight) which simply follows the guidelines of Willet at al. [
2] although it might be that the six food groups considered here contribute to increased risk of some other diseases as well (e.g., Alzheimer, dementia etc.). In the health impact assessment, we were able to propagate the uncertainty in relative risk parameters to the calculated indicators (
Supplementary Table S2).
Fourth, we used the average price of 82 food groups from the largest food retailer website in our cost analysis (
Figure 4) but the cost of food items may differ depending upon its origin or whether it is processed and several other factors. Once more detailed food intake data are available, future studies should match them with their respective prices.
Finally, we only included nine popular alternative dietary scenarios but one can also obtain a sustainable diet using an optimization algorithm that starting with the current diet, generates optimum intake amounts of different food items meeting multiple nutritional, environmental or other constraints (see e.g., Jalava et al. [
19]).