The sweetening of the global diet, particularly beverages: patterns, trends and policy responses for diabetes prevention (original) (raw)

. Author manuscript; available in PMC: 2017 Feb 1.

Published in final edited form as: Lancet Diabetes Endocrinol. 2015 Dec 2;4(2):174–186. doi: 10.1016/S2213-8587(15)00419-2

I. Introduction

Over the past several decades, the world has become increasingly aware of the role of added sugar, particularly in beverages, as a major driver of increased weight gain and diabetes. This is particularly the case in higher income countries such as the United States, United Kingdom and Australia where sugar-sweetened beverage (SSB) levels and trends were upwards in the 20th century (1–4). This was echoed in a recent editorial in this journal titled “Sugar intake: lowering the bar”(5).

In this paper we report from a new dataset to update what is known about global trends in sales of beverages with caloric sweeteners (CS) and low calorie sweeteners (LCS) with a focus on the understudied countries outside of North America, Europe and Australasia. We include at caloric soft drinks, juice drinks and sports and energy drinks. We also provide an in depth analysis of trends for CS and LCS in both foods and beverages in the USA. We then review the types of policy responses that have been put into place around the world. We start by providing an overview of our understanding of the role of CS and LCS) on weight gain, diabetes, and other cardiometabolic problems.

II. Background: the effects of caloric and low calorie sweeteners on obesity and diabetes

The last 25 years have seen a revolution in our understanding of the role of CS (including sugar) on energy intake, obesity and diabetes. Though the relationship between sugar and insulin control has been understood since the 1920’s (6), earlier appeals to view sugar as a danger to our health were ignored by most of the health profession during the 1950’s and 60’s (7, 8). Public health and biomedical scholars focus on sugar-sweetened beverages (SSBs, .g. carbonated sodas, juice drinks with added sugar, energy drinks etc) increased significantly after path-breaking work by Mattes and others showing that intake of caloric beverages in any form added close to 95% or more of calories to daily energy intake (9–12). Though not fully understood, researchers have identified some of the potential biological mechanisms that might explain this lack of compensation for caloric beverage intake with reduced food intake (13, 14).

Extensive meta-analyses show that the risk from added sugars in beverages to weight gain and diabetes is very high. (4, 15–20). Studies claiming that the relationship between CS and weight gain and diabetes is weak have generally been funded by the sugar and beverage industries (21, 22). Risk from CS in food is lower but also important. In part based on this risk, in 2015 the World Health Organization generated the recommendation for added sugar intake of reducing the intake of added sugars to less than 10% of total energy intake2 (strong recommendation) and a further reduction of the intake of added sugars to below 5% of energy intake (conditional recommendation3).

The major source of sugar from beverages is 100% fruit juice. Fruit juice and SSB’s have a similar level of sugar and a similar effect in increasing total energy intake (12, 23, 24). A limited number of long-term studies suggest a strong adverse effect of consumption of each additional serving of 100% fruit juice on weight gain and diabetes risk similar to that of SSB’s (25–27). However no RCT’s have been conducted and this gap needs to be filled.

The role of LCS sweeteners as they affect weight gain and diabetes risk has been even more controversial. While careful reviews and random controlled trials (RCT’s) have not shown any adverse relationship of LCS on energy intake or increased consumption of sweet foods (14, 28), several longitudinal cohort studies implicate LCS as a cause of increased weight and diabetes and other adverse cardiometabolic outcomes (29–31). But these studies ignored a major issue, namely the potential for “affect modification” by an existing unhealthy diet. In fact, two analyses of earlier studies with the same cohort data showed that the health outcomes were affected by increased LCS intake. The revised studies considering the role of diet found that consumers of LCS with an unhealthy western diet had a higher risk, while those consumers of LCS with a prudent healthier diet had a lower risk (32, 33). However, there is not a consensus on the role of LCS on weight gain and diabetes risk; the many new and ongoing studies on LCS should help us come to consensus on their healthfulness.

Another issue that has received renewed attention is fructose, which forms a component of sugar (sucrose) and forms about 45–55% of the sugar in some high fructose corn syrups and 55–65% in some fruit juices (34, 35). Although it is clear that glucose has important effects on obesity and other adverse health responses (36), it appears that fructose, when consumed at high levels has additional adverse effects, such as on increased liver fat, visceral fat, muscle fat and triglycerides (37–51).

Almost all of this research has been undertaken in Western Europe and North America. The exceptions are the longitudinal studies in Singapore (27, 52) and one RCT in Mexico (53). Research in the United States shows that the relationships between BMI and SSB intake is stronger in Hispanics (54–57) and Asians, indicating the importance of conducting further research in other word regions.

In summary, current evidence shows that increased sugars intake affect the risk of weight gain, diabetes and many other health problems. All CS have comparable affects on these health outcomes. There is a need for further research on the role of LCS and 100% fruit juices on health.

III. Caloric and low calorie sweeteners in the US food supply

The US has one of the more complex food supplies in terms of number of products with unique ingredients. Using nationally representative samples of all bar coded foods purchased by Americans from 2000, 2006 and 2013 we analysed information on the consumer packaged foods and beverages with “unique formulations of ingredients” (1.2 million foods in the US bar-coded food supply from 2000–2013) which provide an indication of the extent of the use CS and LCS in the US food supply. Research already published showed that a large proportion of US foods and beverages included the new natural sweetener, fruit juice concentrate (58). We followed this same methodology (58) but made significant refinements in what we terms fruit juice concentrate to rule out any beverage with both fruit juice concentrate and water as have FJC as a sweetener, among other. There is a very large set of CS and LCS that could appears as ingredients or ingredients of the ingredients (59) in nutrition labels. Supplemental table 1 lists all the CS and LCS types of sweeteners searched. We exclude sugar alcohols and two new and rarely used lower calorie sweeteners--allulose and tagatose.

Figure 1 based on the products shows the large increase in the number of unique formulations in the US food supply between 2006 and 2013. There are several key points to note. First overall 68% of all US barcoded food products had added sweeteners, of which 63% contained a CS. Second, an increasing proportion of foods and especially beverages contain both CS and LCS (60). Third, LCS is mainly found in beverages.

Figure 1.

Figure 1

Pathways Linking Sugar-Sweetened Beverage Intake with Health

Source: copyright, D. Ludwig & W Willett, Harvard SPH, 2009

Figure 2 shows the proportion of purchases based on grams for each item with added CS and LCS or both for a nationally representative sample of purchases by US households. It shows there were significant increase in the number of products (especially beverages) with both CS and LCS, but there were also increases of foods and (especially) beverages with no added CS or LCS at all. (Figures 2 and 3) Figure 3 shows almost half of caloric beverages contained no added sweeteners at all and fruit juice concentrate use was minimal. This supports earlier work showing that total added sugar intake in the US has declined and much of this shift is due to a reduction in caloric beverages (61, 62).

Figure 2.

Figure 2

The % of Uniquely Formulated Foods and Beverages in the US Food Supply Containing Sweeteners (Mutually Exclusive Categories)

NS= Nutritive/Caloric Sweetener, NNS= Non-nutritive/non-caloric sweetener, FJC= Fruit Juice Concentrate (excluding lemon/lime and when reconstituted) (not for use or quotation)

Figure 3.

Figure 3

% of Uniquely Formulated CPG Purchases by Weight in Grams Containing Sweeteners (Weighted to be Nationally Representative)

NS= Nutritive/Caloric Sweetener, NNS= Non-nutritive/non-caloric sweetener, FJC= Fruit Juice Concentrate (excluding lemon/lime and when reconstituted) (not for use or quotation)

Comparable data on sales of SSBs is available for hundreds of countries from Euromonitor Passport International data (63). In a new database, Euromonitor worked in each country to obtain caloric information for most beverages on a country by country basis to allow us to estimate kcal/capita/day trends in sales for a limited number of years (2009–2014). We are utilizing these Euromonitor new data for which we were the beta-tester of the data and provided with permission to publish. Longer term trends from 2000 are available for volume in ml. In both cases we combined sales for both what they term off-trade volume (i.e. supermarkets, retailers) and on-trade volume (i.e., restaurants, cafeterias). All volume data are reported in ml/capita/day. We suspect these Euromonitor data omit many small local bottlers but no rigorous study exists to evaluate the completeness of Euromonitor data. We define as SSB’s caloric soft drinks (carbonated, noncarbonated), fruit drinks and several fast growing categories, namely energy drinks, sports drinks, and sugar-sweetened (and often flavored) waters (combined in our figures as sports and energy drinks).

Figure 4 Panel A shows that North America and Latin America are the largest consumers of SSBs, with Asia having very significantly lower levels. While caloric soft drinks dominate in most regions, juice drinks are equally important in Asia. Panel B shows that sales are declining for the 2009–14 period of kcal/cap/day in North America, Australasia and Western Europe, but increasing in all other regions (also see Supp Table 2). There are important differences in trends types of beverages between regions. Sales of caloric soft drinks has remained fairly stable in Latin America, whereas sales of juice drinks have increased. In North America, carbonated drinks have declined very significantly – but sports drinks have increased significantly, as they have in Australasia, though here, soft drinks have remained stable with the most significant overall declines being explained by juice drinks. When combined into global level trends, intakes of the soft drinks component of SSBs has remained fairly stable, whereas sales of juice drinks and energy drinks have increased.

Figure 4.

Figure 4

% of CPG Purchases by Calories Containing Sweeteners Weighted to be Nationally Representative (Excludes Low Calorie Sweeteners by Using Kcal)

NS= Nutritive/Caloric Sweetener, NNS= Non-nutritive/non-caloric sweetener, FJC= Fruit Juice Concentrate (excluding lemon/lime and when reconstituted) (not for use or quotation)

Figure 4 trends by region in kcal put the trends in panel A and the current 2014 levels in panel b

Current Country levels

There is enormous heterogeneity in levels of sales and trends within regions. In Figure 5 we present the data for highest and lowest consuming countries. Four of the six highest countries per capita daily kcal/cap SSB sales are in Latin America - Chile (number 1) Mexico, Argentina and Peru. The US and Saudi Arabia are also in the top six. This contrasts with 2000, when top spot was held by the US 2000 before the large decline in SSB consumption. National trends in different types of SSBs reflect regional trends. China and Chile, for example (supplemental figure X and X) shows growth in juice drink.

Figure 5.

Figure 5

Daily Calories Sold per Capita per Day from All Sugar-Sweetened Beverages

Source: Euromonitor Passport International with country-specific kcal data added (not for use or quotation). Only caloric beverages included)

Supplemental Table 4 great heterogeneity in growth patterns between different countries (e.g comparing Chile vs Mexico—two countries which over the last 4 years have been the focus of extensive publicity about the adverse health effects of SSB’s). These much more inclusive data show some countries, such as China, Thailand, Brazil and Chile are experiencing growth while others like the UK, Mexico, and the US are experiencing declines. In Mexico case this occurred before the SSB tax was instituted. Similarly these data show the decline in sales in the US – despite increases in sport and energy drinks – and low levels of intake in two Asian countries.

The caloric trends essentially mimic the 2000–2014 trends in sales in ml with a few exceptions (supplemental figures XYZ). The US is one where the two trends diverge somewhat as the US has shifted significantly to lower calorie SSB’s though sports and energy drinks have increased. Asia, including China, is characterized by rising consumption of juice drinks. China has undergone a remarkable shift since 2000 from a very low initial SSB sales to a much higher level of consumption, albeit with a less steep increase in Chile in Figure 5. Both the long and shorter-term data show that a major focus of global beverage companies is to push consumption is less saturated emerging markets beyond western countries (64).

Aside from North America, Western Europe and Australia and New Zealand, few countries in the world consume much LCS beverages. As shown in Figure 7 (also Supplemental Table 5), North America and Australia and New Zealand consume double the LCS beverage intake over other regions.

Figure 7.

Figure 7

Diet beverages with only low calorie sweeteners

V. Policy responses

The evidence presented in this paper has three important policy implications.

So what actions have countries been taking to reduce SSB consumption, how much of this action has been in low and middle income countries (LMICs) – and how have they dealt with the conundrum about SSB substitutes? Firstly, it is clear that countries have started to act to reduce SSB consumption. Actions implemented to date include (i) SSB taxes (ii) reducing availability in schools and other public institutions; (iii) restrictions on marketing of sugary foods to children; (iv), public awareness campaigns and; (v) front of pack labeling. Most of these actions are not specific to SSBs but include them.

Figure 9 shows the number of countries have implemented each of these policies. It is based from the best available information source but not a global survey, so some countries will be missing. It includes a total of 72 actions from 49 countries, indicating that countries are acting – but still a relatively small number countries have each taken a relatively small number of actions. Nevertheless, it shows that actions have been taken in all regions beyond North America, Western Europe and Australasia. However, further analysis of the income group of the 49 countries show that no actions have been reported from low income countries, only 2% from lower middle income countries (1 country), 33% from upper middle income countries, with the vast majority being taken in high income countries which in general are the highest consuming regions. This suggests more attention is needed beyond high income countries. Furthermore, half of the restrictions related to mandatory or voluntary school guidelines and another fifth relate to front-of-the-package labeling. So two areas felt to be most powerful in terms of impact—SSB taxation and restrictions on advertising—are used less than one-fifth of the times.

Nevertheless the most notable policy development over past two years has been the increasing level of interest by governments in SSB taxes. As of July 2015 national level taxes had been implemented in five countries in Latin America and Europe, four small island states, and one city (Berkeley, California: about 12–25% based on size of container) and one native American reservation (the Navajo Nation, a 2% tax) in the United States. While taxation levels of 20% or higher have been recommended by scholars (65), taxes to date have been lower. For example, 10% for SSB’s in Mexico (implemented January 2014) and France (January 2012), 8% in Chile (implemented January, 2014), higher in french Polynesia and other Western Pacific Islands (66). Several countries cover more than SSBs alone. In Hungary, for example, the tax (adopted 2012) applies to the sugar, caffeine and salt content of various categories of ready-to-eat foods and drinks, including energy drinks, which are widely consumed by youth. In Mexico, the SSB tax is combined with an approximately 8% tax on “nonessential foods” high in added sugar, sodium or unhealthy saturated fats.

What have been the effect of these taxes? A joint University of North Carolina-Mexican National Institute of Public Health team is evaluating the long-term effects of the taxes in Mexico on food purchase patterns. Preliminary results show a 6 percent average decline in purchases of taxed beverages over 2014 compared to pre-tax trends. This difference accelerated over 2014 and the reduction compared to pre-tax trends reached 12% by December 2014. All socioeconomic groups reduced purchases of taxed beverages. Reductions were higher among lower socioeconomic households, averaging 9% decline over 2014 compared to pre-tax trends and up to a 17% decline by Dec 2014. These are results only for the first year of the tax; in theory the longer-term impact could be greater if taxes influence people’s longer term preferences and habits away from SSBs (67–70). However, this does not account for changes industry behavior so the longer term impact is difficult to predict.

In Hungary, an initial evaluation reported that sales of products subject to the public health tax have fallen by 27%, with a 20–35% decrease in consumption observed. An additional benefit observed has been the response of manufacturers in removing entirely, or substantially reducing, the taxed ingredient in their products through reformulation (71).

Early data from the French tax suggest about a 5% price increase (71). Euromonitor data (Figure 8) suggests a small impact of the tax on French sales; however the WHO reported a 3.3% reduction in sales. No rigorous evaluation has been undertaken. Euromonitor data from Hungary mirrors the decline noted above while the Chilean data precedes the tax and show a continued increase up to the January 1, 2015 implementation of the tax.

Figure 8.

Figure 8

Countries with Taxes: Calories Sold per Capita per Day from Sugar-Sweetened Beverage by Country, 2009–2014

Source: Euromonitor Passport International with country-specific kcal data added (not for use or quotation). Only caloric beverages included

Restricting the availability of SSBs has been a major focus of government actions to improve the quality of foods available in schools around the world (72). While the policies that governments have put into place vary significantly from place to place – for example, some set standards for meals, some for vending, some for all food in schools, some specific to SSBs – all are united by including a restriction on SSBs. Policies have been implemented at the national, state/province, municipal and school district levels; most are mandatory, through there are also many examples of official guidelines for voluntary application.

Evidence on the effects of SSB restrictions on consumption come largely from the United States. There the evidence is consistent that state-level bans lead to lower availability in schools, but the evidence on daily consumption is mixed (73). Some evidence suggests that specific populations (e.g. African Americans) are more responsive to restrictions (74). Others show that restrictions are ineffective if not comprehensive across the school environment, leading researchers to conclude that isolated changes to reduce soda alone need to be accompanied by complementary measures in order to be effective (75). Where bans are comprehensive, evidence shows that they influence in-school purchasing of SSBs – but do not reduce overall consumption since children can bring SSBs into school and consume before and after school. These findings suggest a range of synergistic measures are needed inside school and out where preferences and habits for SSB consumption are already deeply entrenched (76, 77). More evidence is also needed from LMICs where effects may be different due to less entrenched preferences and habits.

Efforts to reduce the exposure and power of marketing of soft drinks have been less widely implemented. The UK, Ireland and South Korea restrict SSB advertising as part of restriction on high fat, sugar, salty foods, but these are limited to specific communications channels. Iran is reported to have had a ban on all soft drink advertising on television since 1994. France requires health messages on all food and drink advertising. Chile will take a new approach with a law scheduled to be implemented in July 2016 by requiring warnings on advertising of foods with levels of sugar (including SSBs), fat and salt. (78)). The Peruvian government created a more limited version of regulating advertising for junk food as a law in 2013(79). Evidence on the effect of marketing restrictions on SSB consumption is not yet available. There is currently no way to evaluate the impact of comprehensive marketing bans recommended as most effective by the WHO since no country has comprehensively restricted all forms of SSB marketing to children.

Governments have also taken actions deigned to inform people (80). The New York City Department of Health set a precedent with its 2009 “Pouring on the Pounds” campaign to reduce SSB consumption through the use of a graphic public awareness campaign with slogans such as “Don’t drink yourself fat. The campaign has been adapted for use in other US states, including Los Angeles Choose Health LA “Sugar Pack” campaign (October 2011 – December 2012). Sugary drinks have also been included in public awareness campaigns in Australian states and territories (as part of the “LiveLighter” campaign) (81), in Englands Change4Life social marketing campaign (which ran a “Smart Swaps” campaign in January 2014 to encourage alternatives to sugary drinks), in Tonga (where the 2012 “A Mouthful of Sugar” campaign featured a bottle of soda with the label ‘diabetes’), and Hungary, Thailand and Venezuela.

Thw body of evidence on public awareness campaigns suggest they are effective at reducing consumption of less healthy foods and drinks if sustained and use multiple modes of communication (82). There are few evaluations specific to SSBs. According to a policy brief on sugar published by the NGO WCRF International (83), one exception is the Choose Health LA “Sugar Pack” campaign which increased the public’s knowledge of sugar in drinks and over 60% of respondents reported they were likely or very likely to reduce their daily intake of sugary drinks as a result.

Another information-oriented approach is placing labels on the front of food packages which clearly indicate the level of sugars in a product. In 2014 the Ecuadorian government set a precedent by requiring “traffic light” labels on packaged foods and drinks which indicates the levels of sugar, fats and salt are by red (high), orange (medium) or green (low). The approach was originally developed in the UK, where the government has voluntary guidelines for traffic light labelling. Chile has developed a “warnings” approach of requiring foods high in added sugar, sodium and saturated fats to carry a warning label about the ill-health effects, also to be applied to advertising in July 2016(78). Another approach is to label foods and drinks with a visible and readable calorie label. To date it appears no countries require this for SSBs, but has been adopted voluntarily by industry. Other labels take a “positive” approach which indicates healthier products, including products lower in sugar – such as the “Healthy Stars Rating” in Australia, the “Green Keyhole” in Denmark, Norway, Sweden and Iceland and the Choices symbol in the Netherlands and other countries. When applied on beverages, the aim of these positive symbols is to encourage the purchase of alternatives to SSBs (84).

The impacts of these labels have yet to be fully evaluated for their impact on purchasing and intake. The currently available suggests that these types of labels are easier to understand and interpret correctly relative to traditional “nutrient lists”, but their effect depends on the nature of the population, with some people far more likely to be responsive than others (85). Different types of labels appear to give rise to different responses (86, 87). One small US study specific to SSBs suggests that providing prominent caloric information was associated with purchasing fewer SSB (88). There is evidence that front-of-pack symbols can have a positive impact on reformulation of foods (i.e. companies reduce levels of nutrients to avoid negative labelling or to gain postive labelling), but not from SSBs.

With regard to reformulation of processed foods, probably the most widespread of all actions to promote healthy eating around the world are mandatory or voluntary targets to reduce population level salt intake through reformulation (89). Yet relatively few countries have made a concerted effort to set sugar-reduction targets in foods and drinks. Exceptions include France, where there is a strategy aimed to reduce sugar consumption by 25%, but this appears to apply to foods only.. In New York City, efforts to introduce legislation on portion sizes of SSBs in restaurants failed to pass.

Given the lack of consensus about the health impact of drinks with LCS and fruit juices (see section 1), as well as the trends in sales of sugar-sweetened juice, sports and energy drinks, it is instructive to examine the decision policy makers have made about how to treat drinks which people potentially turn to to replace SSBs. This is critical since it has the potential to influence what beverages people consume instead and, therefore, the impact of the policy on overall caloric intake. Though from a public health standpoint water is the best alternative, policy makers nevertheless need to make very practical decisions about what drinks are covered by their policies are what are not. For example, should they impose restrictions on LCS beverages as well as SSBs? If they do – but not 100% fruit juices – will it drive substitution for fruit juices and so fail to address excess caloric intake of the sugars in the 100% fruit juices? Further, in many LMICs, public water supplies are inadequate, unsafe or mistrusted and bottled water is more expensive, potentially making water less attractive as a substitute. So what should policy do on water?

For taxes, most countries tax according to the level of sugar in drinks and/or apply them to SSBss only. France is a rare exception in covering drinks with LCS. 100% fruit juices are not taxed in any country. The preliminary evaluation of the Mexican soda taxes suggests people are switching to water: results show roughly a 4 percent increase in purchases of untaxed beverages over 2014, mainly driven by an increase in purchased bottled plain water (tap water intake is not collected). There has been no concomittant increase in sales of 100% fruit juices (they are too expensive), nor drinks with L CS, which have never been promoted and are not popular in Mexico. A widely supported strategy in Mexico was to use a proportion of the SSB taxation for providing potable water in schools, but this does not appear to have been implemented.

School policies on SSBs tend to take a different perspective to categorisation of drinks. Unlike most taxes, policies in, for example, Queensland, Australia, France and Brazil, exclude drinks with LCS from schools on the basis that they are of mimimal nutritional value, and the emphasis of school meals should be nutritious foods and drinks (90). 100% fruit juices are, however, typically allowed although generally with US portion sizes are limited (e.g 4 ounces in the US). The case of Mexico shows how important it is to consider substitution for drinks of equal calorie content to carbonated SSBs. Preliminary analysis of the school food standards implemented in Mexico in 2013, which permits 100% juice, showed that soda was replaced by drinks with equal caloric content (Barquera, 2014). These dynamics may change if Mexico’s Department of Education implements plans to provide potable water in its schools.

Public awareness campaigns tend to promote drinks which are low in calories as alternatives. The slogan used in the New York City 2009 campaign was “Go with water, seltzer or low-fat milk instead”. In the UK, the “Smart Swaps” campaign recommends LCS drinks as potential swaps for SSBs along with water and any drink with no-added sugar (91, 92). The campaign in Tonga recommended subsitution with water or coconut water. Overall, a better understanding is needed of the effects of policy actions on what people replace SSBs with – as well as the health impacts of alternatives to SSBs, to ensure that SSB poilcies can be more effectively designed to have their intended effect of reducing excessive calorie intake from beverages.

Supplementary Material

Supplemental fig

Figure 6.

Figure 6

Calories Sold per Capita per Day from all Sugar-Sweetened Beverage by Country, 2009–2014

Source: Euromonitor Passport International with country-specific kcal data added (not for use or quotation). Only caloric beverages included

Acknowledgments

We thank the Robert Wood Johnson Foundation (Grants 67506, 68793,70017, 71837) and the National Institutes of Health (R01DK098072 and CPC 5 R24 HD050924) for financial support. Shu Wen Ng, Greg Bricker and Donna Miles, UNC Food Research Program, are thanked for exceptional help in creating the US measurement of CS and LCS in our food supply. Chris Ford is thanked for all his work with the Euromonitor data for organizing it. We also wish to thank Ms. Frances L. Dancy for administrative assistance and Ms. Denise Ammons for graphics support.

Footnotes

None of the authors have conflict of interests of any type with respect to this manuscript.

Contributor Information

Barry M. Popkin, Email: popkin@unc.edu, W. R. Kenan, Jr. Distinguished Professor, School of Public Health, Carolina Population Center, University of North Carolina, 137 E. Franklin St., Chapel Hill, NC 27516, Phone: 919-962-6139, Fax: 919-966-9159.

Corinna Hawkes, Centre for Food Policy, School of Arts & Social Sciences, City University London, Northampton Square, London EC1V 0HB, UK.

References

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental fig