How to Obtain Forty Percent Less Environmental Impact by Healthy, Protein-Optimized Snacks for Older Adults

It is well known that meals containing less meat are more sustainable, but little is known about snack-meals, which typically do not contain meat. This study investigates the diversity in environmental impacts associated with snack production based on 20 common recipes optimized for protein content, energy content and sensory aspects for older adults. The purpose is to improve sustainability of public procurement by serving more sustainable snack-meals. Public procurement serves Danish older adults over millions of snack-meals every year, and millions more are served in countries with a similar social service. The environmental impact of snack production was estimated by consequential life cycle assessment. The average impact of producing the 10 least environmentally harmful snacks was 40% less than the average impact of producing the 10 most harmful snacks. This is true whether the functional unit was mass, energy, or protein content, and whether the environmental impact was measured as global warming potential or the monetized value of 16 impact categories. We conclude that large-scale public procurement of snack-meals by private and municipal kitchens can be reduced by up to 40% if the kitchens evaluate the environmental impact of all their snacks and serve the better half more frequently.


Introduction
The environmental impact of food production and consumption is important to the local, regional and global environment [1]. Production of food and beverages is responsible for about half of the environmental impact caused by the personal activities of people living in industrialized countries [2], which in turn is responsible for about half of the overall environmental impact caused by these societies; the remaining half is caused by industry. Selection of healthy and environmentally friendly meals can reduce the environmental impact of food and beverage in the order of 25-65 percent [3,4]. This makes every meal, but particularly the numerous meals prepared in large numbers by central municipal or private kitchens, important, not only to the health of its customers but also to the overall environmental impact, both of which have significant socioeconomic implications [5].
While it is well known that the environmental impact of main meals largely depends on selecting meals with less meat, particularly beef [4,6,7], it is less clear whether selecting certain in-between meals above others can also contribute to reducing the environmental impact of food consumption, since snacks are typically meat-free.
In Denmark, and similar countries with a high level of public service, older adults (older adults: EUROSTAT definition of people aged 65-84. http://ec.europa.eu/eurostat/statistics-explained/ index.php/People_in_the_EU_%E2%80%93_statistics_on_an_ageing_society) who are unable to cook for themselves receive main meals and in-between meals produced by large municipal and private kitchens. For various reasons, older adults tend to eat inadequate amounts of food [8], and in-between meals (from here on called snacks) are therefore of particular importance to tempt the older adults to increase their energy and protein intake. The Danish recommendations for institution-diet advise people with poor appetite to eat at least three snack-meals daily [9]. In a study by Okkels et al. [10], 20 of the most popular/most ordered snacks were assessed by 30 older adults for flavor and appearance, and these snacks were further optimized in terms of energy, protein content and sensory aspects. The primary objective of this study is to determine if the variation in environmental impact of these 20 common snacks is significant and so large, that more often choosing snacks from the better half improves the overall sustainability. The secondary objective is to determine if there are certain ingredients in the snacks that are mainly responsible for their environmental impact. This is important since municipal and private commercial kitchens serve millions of snack-meals every year to older adults in Danish municipalities.

Recipes
Besides the importance of flavor and appearance, the nutritional criteria for all the optimized snacks were a content of at least 6 g protein and 400 KJ per 100 g, the typical weight of a single snack-meal. Figure 1 gives the approximated amounts of ingredients in the 20 recipes, while the precise recipes and energy consumption associated with production are given in Appendix A.

Assessment of Environmental Responses and Monetized Impact
The environmental impact of the 20 recipes was calculated by consequential life cycle assessment (cLCA) of each snack recipe including all ingredients and cooking, baking, cooling and freezing. Life cycle assessment calculates the environmental impact of a product from soil to table using a range of impact categories and a selection of functional units. Consequential LCAs seek to identify the environmental consequences of a decision or a proposed change in a system under study (oriented towards the future), which means that market and economic implications of a decision are taken into account [11].

Figure 1.
Ingredients in the 20 recipes of texture modified snacks where a specific amount of each ingredient is approximated due to proprietary rights. The relative content of all ingredients are given by the color codes. The estimated price of each snack is given in euros based on ingredients, but excluding the capacity cost.

Content of each ingredient:
€ per serving (ingredients) Adosan   However, for clarity, only data for the three most important impact categories (respiratory inorganics, nature occupation, and global warming) were presented separately in this study, along with the sum of the 13 other impacts. The environmental assessment was taken from the Ecoinvent database version 3.3 [12] using the Simapro 8.3 software (Simapro, Amersfoort, The Netherlands) [13]. The Stepwise 2006 version 1.05 method was applied to facilitate monetizing [14,15]. The Stepwise method combines methods from Impact 2002+ version 2.1 and EDIP 2003 with small modifications. Stepwise normalizes data by monetization expressed in Euro, thus calculating the potential socioeconomic cost of environmental externalities. All environmental impacts were calculated according to the ISO standard 14040 [16]. The functional units (i.e., references) were weight, energy content (MJ) or protein content in the manufactured snacks. The energy and protein contents of the ingredients were taken from the Danish FRIDA food database [17].

Statistics
A t-test was applied to determine the significance level of the difference between the 10 best and the 10 worst of the recipes in terms of global warming and monetized environmental. T-tests were also applied to determine if the 10 best recipes had a different content of the most impacting ingredients (cream and protein powder) than the 10 worst, and to determine if the 10 best recipes had a different cost of ingredients than the 10 worst. To investigate possible correlations between impact results using either of the two impact categories and any of the three functional units, a Spearman Rank-order correlation test was applied [18].

Cost of Ingredients
Based on recipes for the 20 optimized snack-meals, combined with food service prices for the ingredients, the ingredient costs per serving for the 20 snacks were calculated. Price data were for the period 2013-2014, supplied from one of the major Danish suppliers of groceries for food service operators.
For ingredients, where more than one variety was available from food service suppliers, the variety with the lowest price per kg was selected for the price calculation. This implies that the price estimates represent the lowest possible ingredient cost for the respective snacks. If further requirements to the ingredients are stated (e.g., that they should be organic as far as possible, should be domestically produced, should be semi-processed, etc.) the unit prices will tend to be higher.
The recipes for the 20 snacks include data on the amounts of ingredients and cooking, baking, cooling and freezing, but not full information about other energy use, personnel hours, depreciation and maintenance of kitchen facilities, waste management etc. Data from Danish municipal kitchens for provision of meals to the older adults suggest that ingredients on average constitute 30-40 percent of the total cost of meal service. This might suggest that the capacity costs would constitute around 1 € per snack, which should be added to the cost figures in Figure 1, although there is of course some variation in the time and energy requirements across snacks.

Packaging and Delivery
The snack-meals all weighed 100 g and were delivered on plastic trays for food with an average weight of 8 g covered with 0.5 g of plastic film. Small diesel vans, e.g., Fiat Ducato or Iveco were used to deliver meals from the kitchens to its customers in the shortest possible overall route. Figure 1 shows four frequent commonalities of the 20 recipes: Fortification with protein powder for protein delivery, cream added for energy delivery and texture, sugar added for energy delivery and taste, and vanilla added for taste. Figure 1 also gives the estimated prices of the snacks. There are no significant differences between the average prices of snacks belonging to the best and worst half of global warming or monetized environmental impacts with any of the 3 functional units (kg, MJ or protein content).

Recipes
In spite of these commonalities, the global warming impact ( Figure 2) and monetized overall environmental impact (Figure 3) of the 20 snacks differ significantly between the 10 best and 10 worst recipes (impacts, differences, and p-values are given in the figures). In spite of these commonalities, the global warming impact ( Figure 2) and monetized overall environmental impact (Figure 3) of the 20 snacks differ significantly between the 10 best and 10 worst recipes (impacts, differences, and p-values are given in the figures). c. 14 19 6 2 1 10 17 4 9 15 3 12 11 16 8 18 7 20 5    c.
half in terms of climate impact is given in the figure together with the relative difference and associated significance level (t-test).   Figure 2 shows the global warming impact (GWP, kg CO 2 -eq) associated with production of the 20 snacks where the numbers below each sub-figure correspond to recipe numbers in Figure 1. Each set of three columns for each snack represents impact per 100 g (red), per MJ (green), and per kg protein (blue) respectively, arranged by increasing impact relative to snack weight (Figure 2a), relative to energy content ( Figure 2b) and relative to kg protein (Figure 2c). On average, the best half of the snacks impacts global warming significantly about 40 percent less than the average of the worst half, whether the functional unit is weight, energy content (MJ) or protein content. However, the snacks belonging to the better half changes with the choice of functional unit.

Global Warming Impact
Only six snacks in Figure 2 are on the environmentally-friendly top-10 list for both energy and protein delivery: Rye bread soup with whipped cream, milkshake, lemon mousse, tuna mousse, rum mousse, and panna cotta. Only the first three of these are on the kg-based top-10 list of low GWP, demonstrating that environmental impact per kg (most commonly reported in the literature) is useless as a functional unit in relation to meaningful calculations of the GWP impact of snack-meals, i.e., in order to improve the energy and protein intake of older adults. The order of the global warming impact of the snacks differ between the three functional units, and the closest match is seen between weight and protein content ( Figure 2). On average, the best half of the snacks have a monetized environmental impact which is significantly about 40 percent less than the average of the worst half, whether the functional unit is weight, energy content (MJ) or protein content ( Figure 3). However, the snacks included in the better half vary with the choice of functional unit.

Monetized Overall Environmental Impact
Only five snacks are on the top-10 list for both energy and protein delivery in Figure 3: Rye bread soup with whipped cream, milkshake, lemon mousse, tuna mousse, and rum mousse. Since only four of these are also on the kg-based top-10 list of low overall environmental impact and the ranking differ among all three functional units, the environmental impact per kg is useless as a functional unit in relation to meaningful calculations of the GWP impact of food. The order of the global warming impact of the snacks differ between the three functional units.

Correlation between Indicators
Spearman-rank-order correlations between the impacts measured by the three functional units (kg, MJ, or protein content) combined with the two investigated impact categories (GWP or €) are significant (p < 0.04 to p < 0.00003) in 11 of the 15 combinations. In spite of this, Figures 2 and 3 clearly demonstrate that the choice of functional unit and impact category are important to select snacks according to the specific purpose, i.e., to deliver energy and protein to the older adults. Table 1 illustrates the 4 to 6 most environmentally harmful ingredients in each of the 20 snacks totalling nearly 100% of the total impact. When comparing these impacts in the best and worst recipes, Table 1 shows that cream is the first or second most impacting ingredient measured both as GWP and as monetized overall environmental impact in 18 of the 20 recipes. Cream is not an ingredient in Nutricia ® protein drink, and in prune trifle cream takes second place only when the impact is measured as GWP and third place when the impact is measured by the monetized overall environmental impact.

What Makes a Snack Less Sustainable?
In the 10 recipes with the lowest GWP impact relative to energy-content (Figure 2b), the average cream content is 44%, and in those with the highest GWP impact it is 27%. A t-test reveals that the 38% lower cream content in the 10 snacks with the highest GWP impact compared with the 10 snacks with the lowest impact is significant (p < 0.02). Similar results are found for GWP impact relative to protein-content, monetized overall environmental impact relative to MJ and to protein content, but the results are less or not significant, respectively 34% lower (p < 0.03), 31% lower (p < 0.05), and 16% lower cream content in the 10 snacks with the highest impact compared with the 10 snacks with the lowest impact (p < 0.23). Table 1 shows that the protein powders Adosan ® (a registered trademark including four types, all sourced from veal and milk: "cold/warm", "high-energy", "high protein" and "transparent"), whey and casein are on average the second most impacting ingredients measured both as GWP (CO 2 -eq) and as monetized overall environmental impact (€) in 18 of the 20 recipes. The exact composition of Adosan ® , which cannot be publicly disclosed, is known to the authors and used in the LCA calculations. In rum mousse, eggs supply the protein, and in tuna mousse canned tuna supplies the protein. In the 10 recipes with the lowest GWP impact relative to protein-content (Figure 2c), the average protein powder content is 12%, and in those with the highest GWP impact it is 31%. A t-test reveals the 163% higher protein powder content in the 10 snacks with the highest GWP impact compared with the 10 snacks with the lowest impact is significant (p < 0.02). Similar results are found for GWP impact relative to energy-content (Figure 2b), monetized overall environmental impact relative to MJ ( Figure 3b) and to protein content (Figure 3c), but the results are less or not significant, respectively 59% higher (p < 0.14), 128% higher (p < 0.04), and 109% higher protein powder content in the 10 snacks with the highest impact compared with the 10 snacks with the lowest impact (p < 0.06).

Packaging and Delivery
Using 8 g plastic trays and 0.5 g plastic film for cover, the environmental impact associated with packaging added a global warming impact of 0.02 kg CO 2 -eq and a monetized overall environmental impact of 0.002 € per meal. This makes up only 2-10 percent of the production impact, which does not affect the conclusion that it is the production rather than packaging that impacts the environment.
The delivery vans typically travelled about 1 km for delivery to each customer. Deliveries consist of a single main meal, and up to seven main meals in addition to the snack-meals. The environmental impact of meal delivery by medium size vans ranged from 0.026 kg CO 2 -eq and 0.005 € per meal, and up to 0.435 kg CO 2 -eq and 0.09 € per meal depending on the average distance between customers (0.92-1.56 km) and the frequency of meal delivery (1-7 times per week). If the meals are delivered on a weekly basis, the environmental impact of delivery is insignificant compared with the meal production. A minor importance of packaging and delivery was also found by [7]. Table 1. Ingredients of the 20 snack meal recipes and their relative share of total impact in terms of GWP (climate) and monetized overall environmental impact (€). 97-100% of the total impact is accounted for by 3-6 ingredients and processes. Adosan stands for one of several types of Adosan ® protein powders.

Recipe
Type of Impact Ingredients and Their Share of Total Impact in Terms of GWP (Climate) and Monetized Overall environmental Impact (€) Total Impact

A Path towards Greater Sustainability
Snack meals often account for half the energy intake of older adults in Denmark. If further tests of the snack recipes discussed in this paper are successful in terms of liking, life quality, physical performance, cognitive status and body weight conservation of the older adults, the recipes could be produced and offered to this group on a larger scale. Approximately 114,000 people received snack meals in Denmark in 2016 from private and municipal kitchens. With two snack meals per day, this totals 83 million snack meals per year.
With a potential of 40 percent savings on global warming by choosing the best half of the snack-meals for the older adults in Denmark, it would equal an annual saving of 1165 t CO 2 , or the annual emission from driving a Euro 5 passenger car approximately 10 million km. Even with the environmentally best half of the snack-meals being chosen more often than the worst half, there can be substantial environmental savings, with a reasonably broad choice of recipes. If you consider the potential savings on a larger European scale, the savings on environmental impact grows far bigger.

Choice of Functional Unit and Impact Category
Which functional unit and impact category is the best to point the kitchens towards snacks with the lowest environmental impact? The answer to the first question depends on the purpose of the snack-meals, whether they are meant to improve the energy or protein content of the daily food intake of the older adults, or in this study both. There is increasing focus on the choice of functional unit when LCAs are used to compare foods [19,20]. Recently, Ref. [21] pointed out that protein quality may be an even better functional unit that protein quantity. The most common functional unit in LCA studies, weight (kg), was included in this study to be able to calculate the others, but it proved meaningless in selecting the most sustainable snack-meals, with the purpose of delivering both energy and proteins to the older adults.
The GWP is the dominant environmental impact of snack production, but it makes up a little less than half of the average monetized overall environmental impact of all snacks. So the answer to the second question is that the sum of the monetized impacts of all 16 impact categories is the best environmental indicator, since it covers the overall environmental impact better than any single category, even the GWP. Therefore, even though the use of different functional units and impact categories was shown to be significantly correlated in 11 out of 15 combinations, the choice of functional unit and impact category matters as it affects the ranking of recipes in terms of environmental sustainability.

Which Ingredients Mainly Cause the Environmental Impact?
All ingredients cause environmental impact, but as Table 1 shows, it is particularly cream and protein powders that are responsible for the environmental impact of the 20 different snacks. The 10 snacks with the highest impact on average (surprisingly) contain the least cream, and (not surprisingly) the most protein powder. Does this indicate that a generic increase in cream and decrease in protein powder content would increase sustainability of the snacks? No, an increase in cream would decrease sustainability in any given snack and add unnecessary energy, while a decrease in protein powder content would indeed increase sustainability, but miss the purpose of the snacks to supplement older adults with more protein in their daily food consumption. Reducing protein powder and increasing cream content in the recipes may not even be compatible with healthy and tasty snacks. There is therefore no general advice on how to modify the recipes for all snacks in order to reduce their environmental impact.
The logical way forward is to more frequently choose snacks among the 10 least impacting recipes and less frequently choosing snacks among the 10 with the highest environmental impact in order to reduce the environmental impact of the snacks offered to the older adults by up to 40 percent.
Each private and municipal kitchen should evaluate the environmental impact of all their snacks, and serve the better half more frequently.

Prices Are Not Correlated with Environmental Impact
Although there is a wide range of snack prices (0.27 € to 1.61 € per serving, or a cheapest-to-most-expensive ratio of 6.2) based on the cost of ingredients and cooking, baking, cooling and freezing, there is no economic incentive not to choose the best half of the snacks in terms of either climate change impact or the monetized overall environmental impact. The average cost of all snacks is about 0.5 €, and the average price of the five snacks that are on the top-10 list for both energy and protein delivery in Figure 3 is similarly about 0.5 €.
The total cost of any snack also include the capacity cost, and this was estimated as twice (around 1 €) the average cost of ingredients. With more information on energy use, personnel hours, depreciation and maintenance of kitchen facilities and waste management that was available for this study, this value could be more precisely calculated. Adding the fabricating cost to that of the ingredients reduces the relative cost difference between individual snack-meals (1.27 € to 2.61 €) reducing the cheapest-to-most-expensive ratio to a third: 2.1.
We recommend that the choice of snacks should be selected based on taste/acceptability (it is important that the older adults eat the additional snack-meals), health (older adults are vulnerable), and environmental impact (food is a major contributor to our overall environmental impact)-in that order-but without neglecting any of these three aspects of snack-meals. Fortunately, low environmental impact often follows a positive health impact of foods [22], and monetized health impact has been shown to be more important than the monetized environmental impact [4,5]. The price of the snack should not influence the choice of snacks to be served to the older adults.

Conclusions
There is a 40 percent improvement in environmental impact of snack-meals for the older adults if choosing the environmentally best half rather than the worst half of 20 snack-meals, independent of the functional unit (weight, energy or protein content) and impact category (GWP or the monetized value of 16 impact categories). Choosing more snacks from the best half improves the overall sustainability. However, taste, health and environmental impact should, in that order, decide the choice of snack-meals for the older adults. The data provided by this study give no support for general modification of the recipes. However, future studies should investigate whether protein powders sourced from plants rather than from animals could result in a lower environmental impact of snack meal recipes with added protein powder.
In view of the large differences between the least and the most environmentally harmful snacks, it is advised that public and private providers of meals for the older adults (and other population groups) make an environmental inventory of all their meals, in order to offer not only the tastiest, healthiest and most attractive, but also the environmentally best snack meals. A conscious diet choice is now a proven path towards greater sustainability. 90 g sweat potato 40 g organic celery (root) 220 g organic cream (38%) 470 g baking potatoes 50 g parsnips Energy for boiling Procedure: Peel and chop the roots. Melt the butter together with the chopped roots. Add Adosan ® and cream. Boil the roots slowly and blend it all.

Recipe 8 Panna cotta
Panna cotta 1500 g cream (38%) 135 g powdered sugar 12 g isinglass 15 g vanilla sugar 15 g organic lemon peel 225 g Adosan ® 'high protein' Energy for cooling Fruit puree 1200 g frozen blueberries 150 g cane sugar 1.5 pod vanilla = 7.5 g vanilla pod 3 cinnamon pods = 3 × 2.5 g = 7.5 g cinnamon pod 30 g Adosan ® 'transparent' Energy for freezing and boiling Procedure: Panna cotta: Hydrate the isinglass. Heat the cream to boiling with the other ingredients, while stirring carefully. Melt the isinglass in the lukewarm creamy substance. Place portions under cooling. Fruit puree: Boil berries, sugar, Adosan ® , cinnamon and vanilla pods. Cool, and remove the pods before mixing with a hand blender. Run the produce through a sieve and put it on top of the Panna Cotta.

Recipe 9
Prune trifle 33 portions = 3300 g Prune trifle 600 g prunes 1500 g water 170 g sugar 10 g vanilla sugar 150 g Adosan ® 'transparent' Energy for boiling Cream 400 g cream (38%) 300 g plain yoghurt, 3.6% fat 60 g sugar 100 g Adosan ® 'high protein' 10 g vanilla sugar Procedure: Prune porridge: Hydrate the prunes in water for 24 h. Then boil the prunes with sugar, vanilla and Adosan ® . Let the porridge cool down. Cream: Whip the cream and mix the cream ingredients into the cream, and place the cream on top of each portion prune trifle.

Recipe 10 Raspberry jelly with vanilla cream
Ca. 31 portions = 3158 g (2268 g jelly + 890 g cream) Raspberry jelly 1200 g raspberry puree 300 g sugar 120 g Adosan ® 'transparent' 48 g isinglass 600 g water Energy for cooling Cream 400 mL = 420 g cream (38%) 300 g plain yoghurt (3.6% fat) 60 g sugar 100 g Adosan ® 'cold/warm' 10 g vanilla sugar Energy for cooling Procedure: Raspberry jelly: Boil water and sugar with Adosan ® . Melt hydrated isinglass into the substance. Add the raspberry puree and stir. Pour into small bowls or cups and preserve cold. Cream: Whip the cream; add sugar, vanilla sugar and Adosan ® . Stir with yoghurt and put it all over the jelly, when this is solid.