Wider adoption of the dietary practices identified by the “Mistra Sustainable Consumption” programme as niche, sustainable and with potential to become mainstream would be expected to result in considerable public health benefits, especially for men, in addition to a likely reduction in diet-related GHGEs. Taking 2010–11 as the starting point for implementation, we modelled what the impact in Sweden might have been in terms of deaths prevented or postponed over 20 and 30 years. The results suggest that the gains could have been in the hundreds of thousands of YLL, possibly in the region of a million for the combination of scenarios labelled “moderate” (changes at the 50% level) or “extensive” (changes at the 100% level). To put this in perspective, in Sweden in 2017, the number of YLL from all causes was approximately 1,248,000 according to GBD 2017 . Approximately 16% of these were due to IHD (ca 203,000), and poor dietary habits (the cumulative effect of all 15 dietary risks included in GBD) was the biggest risk factor. As the biggest change in our dietary scenarios was for YLL due to IHD, it is easy to see how the cumulative figure over a longer time period reaches a substantial number. The practice that had the most impact was reducing the intake of red and processed meat and replacing it with a mixture of vegetables and legumes. The results suggest that this practice alone could prevent about a fifth of YLL due to IHD.
As the most recent national dietary data in Sweden is from 2010 to 11, it is worth considering how relevant the proposed scenarios are today. Changes in per capita supply data between 2010 and 2019  suggest a 18% decrease in pork, a 7% decrease in beef and a 20% increase in poultry volumes. No clear trend is seen for vegetables using the same source, but the market for “meat alternatives” has increased over 15% year-on-year between 2017 and 2019 . A internet panel survey of vegetarian practices suggests a similar trend between 2016 and 2019: while the proportion of adults who rarely or never ate vegetarian meals was 48% in 2019, the proportion who never ate vegetarian meals had fallen from 22 to 15% . It should be noted that neither per capita supply data nor market analysis is a substitute for individual consumption data but can be useful to observe trends . These would suggest that the scenario of replacing red meat (with poultry or something else) is already being practiced by some consumer segments – from “innovators” to the “early majority”  – and have the potential to be scaled up. In contrast, the per capita sales volume of SSBs has increased slightly since 2010 . Rogers lists compatibility with existing values, norms and practices as one of five factors determining the success of an innovation. Others are 2) relative advantage (the greater the perceived relative advantage of an innovation by the user group, the more rapid its rate of adoption is likely to be); 3) simplicity and ease of use; 4) trialability (the degree to which an innovation can be experimented with on a limited basis); and 5) observable results (the easier it is for individuals to see the results of an innovation, the more likely they are to adopt it) . In order to accelerate the dietary shift required, all these factors could be taken into consideration by stakeholders and decision-makers who want to bring about change.
It is always important to take account of several aspects of sustainability simultaneously so that e.g. health is not prioritised at the expense of the environment, or social sustainability. In general, the overlap between foods that have lower environmental impact and also improve health is usually high, with the notable exception of fish and sugar, where health and environmental impacts may act in the opposite directions . We were not able to consider other impacts that substitutions within food groups could have in more detail. However in another part of the “Sustainable Consumption” project, the impacts of a number of potential “replacement” products relevant for these scenarios (e.g. legume-based products, plant-based drinks) have been quantified from the Swedish perspective , confirming their lower GHGE impact, even when production, transport from abroad, packaging, etc. is taken into account.
One of the major limitations of simulations is that replacements may have consequences for energy balance and nutritional adequacy, as what and how much replacement occurs in reality is difficult to know. We assumed here replacement by equal weight, not by energy. For example in one scenario we assumed a total reduction in processed and red meat, and compensating for this by increasing vegetables and legumes (in equal amounts) by the same weight. This increase was, on paper, not excessive and would bring average intakes of total fruits, vegetables and legumes up to 384 g for men and 396 g for women, still far below the recommended intakes of 500 g . If these were consumed in their unprocessed form it would likely mean a shortfall in energy intake, due to the lower energy density of this food group. It is however likely that replacement would also involve more processed vegetable- and legume-based products, which are more energy-dense than unprocessed vegetables and legumes, as the availability of these has increased dramatically in recent years in Sweden . Given half of all Swedish adults are living with the effects of a prior or ongoing positive energy balance (i.e. are overweight or obese ), for many energy deficits may lead to further health gains. Similarly, the impact associated with reduced SSB consumption are also possibly underestimated as further gains could be mediated through a reduction in obesity, for which high SSB consumption is a risk factor.
A more critical issue is if the foods that are reduced are important sources of nutrients that are not compensated for. Some of our modelled scenarios would be almost neutral in terms of impact on micronutrient intake, for example replacing SSB with water. Others are more complex. For example, meat is a rich source of nutrients such as iron, selenium, zinc and some B-vitamins. However, a study from the Nordic region examined this (using, for the Swedish part, the same dietary survey data as in our study) and concluded that the effects on overall dietary quality would be minimal if processed meat was reduced to zero, and if average red meat was reduced to the WCRFs 2007 population-level recommendation of 43 g per day . They modelled scenarios involving both replacement by other meat, non-meat, and with or without energy compensation. Another way of looking at the replacements at food level and ensuring that the overall diet is nutritionally adequate would be to perform an optimisation analysis using linear programming  and this is planned in a future study.
Another limitation of simulations or models is that they remain theoretical. Indeed, the relationship between the health and environmental impacts of self-selected diets is more complex than that between single foods/food groups . The modelled changes may also be less acceptable to consumers than what is assumed. Vieux et al. examined actual dietary patterns in six European countries, including Sweden, and concluded that exclusion of entire categories of food is not necessary to achieve health and climate benefits, and a “more sustainable” diet with “moderate” amounts of animal-based products is probably realistic, as it is already adopted by nearly one in five adults  corresponding to the population segments innovators, early adopters and some of the early majority according to the Diffusion of Innovations theory . We therefore made sure to include scenarios where animal-based products were still included to a large degree, as well as being more extensively reduced.
Our results complement those of Saha et al.  who estimated 1-year health gains if food and nutrient intakes were in line with Nordic Nutrition Recommendations. They used the PRIME model, which is different from our method and based on other premises. They estimated that 6405 deaths in a year in Sweden would be prevented/delayed from cardiovascular diseases and diet-related cancers, or 14.4% of the total. In line with our results, the majority of YLL reductions were also from IHD. However other differences make the results difficult to compare: they used the same dietary data but based their model on population data from 2016, not 2011 which was 5.4% smaller; they modelled nutrients (fat, salt, dietary fiber, energy) and only one food group (fruits and vegetables (and not meat)). They also assumed that their health gains occurred in the same year, not allowing for a lag time as we have done, and not taking into account the impact on the population structure over time. Assuming a constant effect over 20 years or 30 years, this would correspond to 128,000 or 192,000 prevented or postponed deaths in their study. Using the same IOMLIFET approach, a scenario for the UK (population 67 million) suggested almost 7 million YLL would be saved over 30 years if diets were in line with WHO dietary recommendations . A study from Italy using the IOMLIFET model that we used predicted that reducing beef by 63% (to 150 g per week) and processed meat by 80% (to 50 g per week) would reduce YLL by 9 and 20 million respectively over 30 years . This is a similar range to our numbers, when Italy’s population size (about 6 times larger) is taken into account.
Other strengths and limitations are that we included all food-based dietary factors in GBD 2017 that could be connected to the proposed scenarios but did not consider nutrient-based dietary factors. Where several dietary exposures affected the same disease, the risks were multiplied together. It is possible that this leads to an over-estimation, but this is commonly done in other models too, e.g. PRIME, and due to lack of information on mediation/overlap. We only examined YLL, not years of life lived with disability (YLD), another widely used measure of health impact, which means that we have most likely underestimated total health benefits by not accounting for impacts on morbidity. We also did not consider here the effects that any reduction in body mass index (BMI) may have had, either due to negative energy balance as a result of a scenario, or as a scenario in its own right. Reducing excess consumption (and waste) is one obvious way to reduce the environmental impact of a diet, as this is determined by both the quality and quantity of food consumed .
The population has increased since 2011, from 9.48 million in 2011 to 10.33 million in 2019 , an increase of 8.9% so the reductions in YLLs may be underestimated. The population structure has remained similar: the proportion of men increased by less than 1%, life expectancy at birth increased by 1.9% for men and 1.3% for women during the last decade. Data on disease-specific death rates were taken from GBD 2017 rather than directly from the national source to ensure deaths from and RRs for diseases were for exactly the same disease codes, but the differences in deaths from both sources were minimal (< 1%). The expected benefit to the environment was limited to the effect on GHGEs, but other aspects such as water and land use are also important. Although a clear socioeconomic gradient is seen with dietary quality, the impact of dietary changes on economic sustainability was not possible to include in this analysis.