Global benchmarking of children's exposure to television advertising of unhealthy foods and beverages across 22 countries

Summary Restricting children's exposures to marketing of unhealthy foods and beverages is a global obesity prevention priority. Monitoring marketing exposures supports informed policymaking. This study presents a global overview of children's television advertising exposure to healthy and unhealthy products. Twenty‐two countries contributed data, captured between 2008 and 2017. Advertisements were coded for the nature of foods and beverages, using the 2015 World Health Organization (WHO) Europe Nutrient Profile Model (should be permitted/not‐permitted to be advertised). Peak viewing times were defined as the top five hour timeslots for children. On average, there were four times more advertisements for foods/beverages that should not be permitted than for permitted foods/beverages. The frequency of food/beverages advertisements that should not be permitted per hour was higher during peak viewing times compared with other times (P < 0.001). During peak viewing times, food and beverage advertisements that should not be permitted were higher in countries with industry self‐regulatory programmes for responsible advertising compared with countries with no policies. Globally, children are exposed to a large volume of television advertisements for unhealthy foods and beverages, despite the implementation of food industry programmes. Governments should enact regulation to protect children from television advertising of unhealthy products that undermine their health.


Summary
Restricting children's exposures to marketing of unhealthy foods and beverages is a global obesity prevention priority. Monitoring marketing exposures supports informed policymaking. This study presents a global overview of children's television advertising exposure to healthy and unhealthy products. Twenty-two countries contributed data, captured between 2008 and 2017. Advertisements were coded for the nature of foods and beverages, using the 2015 World Health Organization (WHO) Europe Nutrient Profile Model (should be permitted/not-permitted to be advertised). Peak viewing times were defined as the top five hour timeslots for children. On average, there were four times more advertisements for foods/beverages that should not be permitted than for permitted foods/beverages. The frequency of food/beverages advertisements that should not be permitted per hour was higher during peak viewing times compared with other times (P < 0.001). During peak viewing times, food and beverage advertisements that should not be permitted were higher in countries with industry self-regulatory programmes for responsible advertising compared with countries with no policies. Globally, children are exposed to a large volume of television advertisements for unhealthy foods and beverages, despite the implementation of food industry programmes. Governments should enact regulation to protect children from television advertising of unhealthy products that undermine their health. The marketing of unhealthy foods and beverages to children has been repeatedly identified in comprehensive evidence reviews as a negative influence on children's food knowledge, preferences, consumption, diet quality, and health. [1][2][3][4][5][6][7] Therefore, restricting this marketing has been a global priority for obesity and diet-related noncommunicable disease (NCD) prevention. [8][9][10] The current World Health Organization (WHO) Global Action Plan for the prevention and control of NCDs 2013-2020 identifies policies to reduce the impact of the marketing of unhealthy foods and beverages on children as one of 25 indicators for change in order to reduce premature mortality from NCDs by 25% by 2025. 11 In 2010, 193 Member States endorsed Resolution WHA63.14 to restrict the marketing of food and nonalcoholic beverage products high in saturated fats, trans fats, free sugars, and/or sodium to children and adolescents. 8 Global progress has been limited on restricting the nature and extent of children's exposure to the marketing of unhealthy foods and beverages and has been addressed primarily by industry self-regulatory programmes. [12][13][14] Globally, almost one in five people aged 5 to 19 years were overweight or obese in 2016, and the prevalence of obesity in young people has increased 10-fold in the last 40 years. 15 While the overall rate of increase in children's body mass index seems to have plateaued (at a high level) in high-income countries since 2000, rates continue to increase in low-and middle-income countries. 15 The globalization of food systems has been implicated in major shifts towards poor dietary patterns and increasing NCD risk in low-and middle-income countries. 16 Globalization includes international food trade and foreign direct investment by transnational food and beverage companies that This article is part of the upcoming supplement 'Future Directions in Obesity Prevention' by the Lancet Commission on Obesity.
predominantly manufacture and sell unhealthy foods and beverage products. This marketing increases the desirability and normalcy of consuming these products. 16 The International Network for Food and Obesity/noncommunicable diseases Research, Monitoring and Action Support (INFORMAS) is a global network of researchers that aims to monitor, benchmark, and support public and private sector actions to create healthy food environments and reduce obesity and NCDs. 17 This includes monitoring and benchmarking food marketing environments over time and across place, 18 as well as government and industry policies relating to food marketing. 19,20 Monitoring data is required throughout policy development and implementation. Initially, these data can highlight the need for policy interventions by identifying the nature and extent of children's food marketing exposures. During policy development, monitoring data can inform policy specifications, including the media, broadcast times, food products, and marketing techniques that should be restricted. Once implemented, monitoring is essential for evaluating the impact of policies on reducing children's exposures to unhealthy food and beverage marketing.
The WHO has recommended that Member States (or national governments) monitor children's exposure to, and the persuasive power of, food and beverage marketing messages. 21 Exposure refers to the reach and frequency of contact with marketing messages, while power relates to the marketing content and design. This study had three research objectives. Firstly, given the substantial number of monitoring studies that have measured exposure and power of television food advertising to children globally, 22 we aimed to combine these data from 22 countries to form a global overview of children's estimated exposure to advertising of unhealthy foods and beverages on television. We also sought to determine the potential impact of policy interventions, by comparing rates of television advertising for unhealthy foods and beverages to children across countries with different policy arrangements (statutory, coregulatory, or industry self-regulation).
Lastly, to identify the influence of international food trade and investment on food advertising exposures, we identified parent companies of branded food and beverage products and the penetration and frequency of advertising by parent companies (ie, the owners of subbrands or individual product brands) across markets.
Several countries (including Australia, Canada, Chile, Costa Rica, Guatemala, Malta, Mexico, New Zealand, and Slovenia) contributed data on television food advertising that had been collected using the INFORMAS standardized protocol. 23 These data were supplemented with data from other countries, identified by a literature of studies measuring the nature and extent of food advertising on television. There was no limit on year of data collection. Included studies were those that captured television broadcasting during consecutive hours across recorded days (not limited to children's programmes) and those that included information on all food and beverage advertisements (not limited to specific food categories). Data were mostly recorded by the research team, although one dataset was purchased from a market research company (2009 UK dataset) and one was freely obtained from the national television regulator (Chile dataset). Corresponding authors were contacted and invited to participate by contributing their dataset. Datasets were eligible for inclusion if they contained sufficient information on the time each food advertisement was shown to enable calculation of the frequency of advertisements per hour, as well as sufficient detail on the advertised food and beverage products to allow reclassification using a standardized food classification system. In addition, studies needed to include at least one weekday and one weekend day, preferably randomly selected. Advertisements broadcast during school holiday periods were excluded. Datasets were cleaned by the country research teams and then sent to the lead author, who processed these for compilation.
Television food advertising policy arrangements of included countries were also captured from repositories of government and industry policies on food marketing to children 24,25 and cross-checked with corresponding authors. Extracted policy information included the type of regulatory control (self-regulatory codes of practice from the food industry, statutory government regulations, coregulation, or no regulation) and the year of implementation.

| Classification of food and beverage advertisements
Advertisements included paid commercial messages that were broadcast before, during, or after television programmes. Product placements embedded within programmes were excluded, as were messages about commercial sponsorship of programmes (eg, "this program was brought to you by …"). The term "food advertisement" was used to refer to advertisements for retail food and nonalcoholic drink products, as well as advertisements for retailers themselves (supermarkets) and food service outlets (restaurants). This also included advertisements for food companies, retailers, and outlets where no specific foods or beverages were visually depicted. Food advertisements were recoded, where necessary, according to the INFORMAS protocol for television food advertising monitoring. 23 For each advertisement, coded variables included orienting information on the placement of the advertisement (country, day, date, channel, and time of broadcast); descriptive information on the nature of the food or beverage product, company, retailer, or outlet promoted (brand/company name and description); and the use of selected persuasive marketing techniques including promotional characters (company-owned media characters or brand-equity mascots, third-party licensed characters, entertainment, or sports celebrities) and premium offers (eg, competitions, rebates, and games). For advertisements where multiple products were promoted, the first product listed the description in country datasets was coded.
Advertised foods and beverages were then coded according to the WHO Regional Office for Europe Nutrient Profile Model, 26 which was designed for the purpose of restricting the marketing of foods and beverages to children (no age range given). The model differentiates products into 16 food and four beverage categories and designates these as recommended to be "not-permitted" or "permitted" to be advertised to children. Certain categories are recommended to not be permitted to be marketed to children regardless of their nutritional composition. These include chocolate and confectionery, cakes and sweet biscuits, juices, and energy drinks. Conversely, unprocessed meat and fish and fresh/frozen fruit and vegetables are recommended to be marketed without restriction. For other categories, threshold criteria per 100 g/mL for total fat, saturated fat, trans fat, total sugar, added sugar, nonsugar sweeteners, salt, and/or energy apply. Notably, advertisements for coffee, tea, nutritional supplements, baby food, and toddler formula are not covered by the nutrient profiling model, and these were identified separately. Advertisements for food companies, retailers, and outlets that do not promote specific food or beverage products are also not covered by the model and were similarly identified separately.
Advertisements were identified as being broadcast during children's peak television viewing times or during other viewing times.
For the purposes of this study, peak viewing times were defined as the top five hour timeslots based on the maximum child audience of the day for each country, separately for weekdays and weekend days.
The age of "children" varied between countries, depending on the definitions used in available audience data (see Table 1). Contributing authors from each country identified children's television audience data from available sources, reflecting average daily viewing patterns as close to the year of data collection as possible. Children's peak viewing times, children's age groups, and the source of audience data for each country are provided in Table S1.
Food and beverage advertisements were also classified according to the parent company of promoted products, supermarkets, and restaurants. Parent companies of product brands were identified through local brand webpages and verified on parent company websites. This parent company information was cross-checked against a wider web search to identify any changes in acquisition since the time of television data capture and any company subsidiaries, where particular product brands were owned by different parent companies in different countries. This web search included Bloomberg.com (equity trading platform) and/or Euromonitor (market research database).

| Analyses
Data were compiled into SPSS for Windows version 21 (IBM Corp, Armonk, New York). Given the variations in advertising between weekdays and weekend days, to derive estimates from combined weekday and weekend day data, data were weighted to take account of the unequal probabilities of selection and the number of weekdays and weekend days in the sample frame for each country. The primary indicators of interest were the frequency of total food and beverage advertisements and food and beverage advertisements that should not be permitted broadcast overall and during children's peak viewing times.
Advertising frequency was calculated as the mean number of advertisements per hour, per channel for each country. Differences in the frequency of advertising for foods and beverages that should not be permitted between children's peak viewing times and other viewing times were examined using paired sample t tests. Statistical significance was accepted at the level of α = 0.05. The frequency of advertisements for foods and beverages that should not be permitted during children's peak viewing times was compared against the regulatory arrangements in countries (government statutory regulation and/or industry code or no regulation) using one-way analysis of variance (ANOVA) with Scheffe post hoc tests. The overall frequency of advertising across all countries was identified by parent company and the number of markets in which the top advertising companies advertised.

| Sample description
Twenty-three research groups from 22 countries contributed data for this study. The final compiled dataset spanned countries from the Asia Pacific region, Africa, Central and South America, Europe, and North America ( Table 1). The total included broadcast time was 11 191 hours (mean = 486.6, SD 499.8). Most countries captured television for the majority of the day across the recording period (14-18 h), with the exception of Tonga where data were only captured from 4:00-9:00 PM daily and Thailand where data were captured from 3:00-8:00 PM on weekdays and 6:00-10:00 AM and 3:00-8:00 PM on weekend days.

| Overall food and beverage advertising by country and region
Across countries that also captured non-food advertisements (N = 15), 23% of all television advertisements were for food or beverage products (Table 2). Overall, there were four times more advertisements for foods and beverages that would not be permitted to be advertised to children using the WHO European Region nutrient profiling model than for foods and beverage that would be permitted to be advertised (2.4 per hour vs 0.6 per hour).
There was an average of 0.3 ads per hour (SD = 0.9) for foods and beverage that are not considered as part of WHO nutrient profiling model (2119 ads across the sample). This included coffee, tea, supplements, and baby food and toddler or follow-up formula. Overall, 3% of food advertisements were for toddler formulas, although this was markedly higher in some countries, including China where these products comprised 21% of food ads, Guatemala (11%), Malaysia (8%), and Costa Rica (7%).
Further, there was an average of 0.3 ads per hour (SD = 0.9) for food companies or brands (no specific foods depicted), which could not be classified by the nutrient profiling model given the range of food and beverage products that the companies manufactured or sold (2206 ads across the sample). This frequency was fairly consistent across countries, ranging from 0 food company brand ads per hour in Malta to 1.2 (SD = 2.0) in Slovenia. Almost all of these types of advertisements (80%) were for supermarkets, where only the store was advertised and not food products, and a further 14% were for fast food restaurants without products advertised.

| Food and beverage advertising during children's peak viewing times
Overall, for those countries where children's television audience data were available (N = 12), the mean frequency of food and beverage advertisements that should not be permitted was significantly higher during children's peak viewing times compared with other viewing times (Table 4). These timeslots when the highest number of children was watching television were typically during the late evening, up to 10:00 PM or later in most countries (Table S1). The frequency of food and beverage advertisements that should not be permitted was around 35% higher during peak viewing times overall compared with other viewing times. In countries where the frequency was significantly higher in peak viewing times, the frequency of food and beverage advertising that should not be permitted was between 48% higher (Australia) to 85% higher (Chile).

| Persuasive marketing techniques
Most countries, with the exception of the five Pacific Island countries, Chile and the United Kingdom (2009) had comparable information on the use of promotional characters and premium offers in food advertisements. Of these 16 countries, 30% of food and beverage advertisements contained promotional characters, and of these, three-quarters (73%) were for products that would not be permitted   All food" includes advertisements for coffee, tea, nutritional supplements, baby food, and toddler formula. In addition, it also covers advertisements for food companies, retailers, and outlets that do not promote specific food products. c Standard deviation cannot be calculated as data were not available by hourly timeslot but aggregated across multiple hours. 3.6 | Food advertising by parent company Table 6 presents the parent companies contributing to at least 1% of overall food and beverage advertisements across the sample. Of the 43 306 food and beverage advertisements captured across the 23 datasets from 22 countries, the top 10 advertising companies contributed to one-third (34%) ( Table 6). These companies also contributed to 40% of all food advertisements that should not be permitted. Nine of these top 10 advertising companies were present across most markets.

| DISCUSSION
This study provides a global representation of children's exposure to food and nonalcoholic beverage advertising on television, the nature of this advertising in terms of the promoted products, and the techniques used that have a strong persuasive appeal, the main advertising companies, and the relationship of advertising to government regulation and industry self-regulatory programmes. We identified three major findings: (1) that unhealthy foods and beverages were promoted four times more than healthier foods and beverages and the rate of unhealthy food and beverage advertisements was even higher during peak viewing times for children; (2) that the bulk of food and beverage advertisements derive from a small number of transnational companies; and (3) that existing regulatory arrangements in countries do not appear to have created more favourable/healthy television food advertising environments compared with countries without any such policies. These key points are addressed below.
We identified that, globally, rates of food and beverage advertisements that should not be permitted were 35% higher during children's peak viewing times overall compared with nonpeak viewing times and were significantly higher during children's peak viewing times in Chile (85%), Malaysia (75% higher), Canada (77%), Guatemala (56%), New Zealand (60%), Costa Rica (50%), and Australia (48%). Of these countries, at the time of data capture, only Australia had government statutory regulations in place to restrict television advertising of unhealthy food and beverages to children. It was largely these advertisements for unhealthy foods and beverages that contained persuasive elements that are appealing to children 27 and these persuasive techniques were also most frequently used during children's peak viewing times. These findings are aligned with previous studies measuring children's television food advertising exposures, which similarly found the highest rates of advertising for unhealthy foods and beverages during broadcast periods when the greatest number of children was watching. [28][29][30] The top five most frequently advertised food and beverage categories were "other beverages" (including carbonated soft drinks, mineral water, and flavoured waters), chocolate and confectionery, ready-made food and dishes, breakfast cereals, and cakes, biscuits, and pastries; the vast majority of which were unhealthy versions that exceeded the WHO Regional Office for Europe's nutrient profiling model's threshold criteria for fat, sugar, sodium, and/or energy. Across all viewing times, the rate of advertisements for foods and beverages In our sample, one-third of all food and beverage advertisements were derived from just 10 companies globally. All of these companies were transnational corporations, which had a combined market value of more than US$994 billion in 2017. 31 This suggests the colossal economic power of these food and beverage manufacturing and retail industries and their potential to influence country-level government policies affecting the production, distribution, and promotion of their products. 32 Strong industry resistance to government regulations to limit the sale and promotion of unhealthy products has been documented. 13,33 Retaliatory techniques have included sponsoring and disseminating reports to deny the need for, or impact of, regulation; developing alliances with civil society organizations to promote physical activity interventions to control obesity; building consensus in agreement with industry's agenda; and focusing on the potential for regulations to have detrimental effects to economy and trade, including loss of jobs. 34 Further, all 10 of these top advertising companies were signatories to the International Food and Beverage Alliance (IFBA) global commitment for responsible marketing, which since 2009 has pledged to only advertise "better-for-you" products to children younger than 12 years. 35 Despite this, these companies disproportionately advertised unhealthy products compared with healthier products across all broadcast times.  into an enterprise in another, whereby the foreign enterprise becomes a foreign affiliate of the parent company. 16 Hawkes previously identified that an increasing proportion of foreign direct investment is now entering developing and transition markets, including in Latin America, Asia, and Central and Eastern Europe. 16 This is commensurate with our findings on the contribution of transnational company-owned product advertising in these markets. Food and beverage products owned by these transnational corporations are typically highly processed, energy dense, and high in sugar, salt, and saturated or trans fats. 36 Such foods are hyper-palatable, shelf-stable, and value-added, all leading to their high profitability for food companies. The heavy marketing of these products serves to increase their desirability and normality and, by targeting children, builds brand loyalty that can ensure lifelong product purchases. 37 This study also indicates the ineffectiveness of existing voluntary foods that may be advertised to children, such as the WHO Regional Office for Europe model used in this study.
Related to television, most industry codes of practice only apply to programmes where the audience comprises at least 35% children, which occurs infrequently as identified from television audience measurement data. 40 Alternatively, the definition of children's peak viewing times adopted in the current study was based on the absolute number of children watching. At the time of data capture, five countries-Australia, Mexico, South Africa, Thailand, and the United Kingdom-also had government regulations related to food marketing to children. Despite this, the frequency of advertisements for unhealthy foods and beverages during children's peak viewing times was not significantly different to countries with no policy or with an industry self-regulatory code. Australia had one of the highest frequencies of advertising for unhealthy foods and beverages during children's peak viewing times. This is unsurprising given that these Australian marketing expenditures indicate that expenditure on online media, and particularly social media advertising, will exceed that for television advertising in 2018. 43 As such, comparative studies on children's potential exposures to online media food marketing are needed.
Strengths of this study included the recoding of country datasets according to a standardized coding frame that has been adopted by INFORMAS 23 to allow cross-country comparisons; the application of the WHO nutrient profiling model across countries to explore the extent that this model is appropriate for classifying products as part of food marketing regulations broadly; the use of data on child television viewing patterns to discern broadcast times when the greatest number of children is exposed; and the identification of parent companies of food and beverage advertisements, an analysis that has not been undertaken previously.

| CONCLUSION
Globally, children are potentially exposed to large volumes of television advertising for unhealthy foods and beverages, despite the implementation of food industry codes of practice for responsible marketing in many countries. Such policies have been ineffective in reducing exposures to this form of marketing. Across all countries, television food and beverage advertisements are predominantly for products that exceed WHO maximum thresholds for saturated fat, sodium, and/or sugar for foods and beverages that are considered appropriate to be marketed to children. Most of these advertisements derive from multinational corporations, which reach children by buying advertising slots at times when the greatest number of children is watching and by using techniques with strong persuasive appeal that are engaging to young people. Monitoring data such as those presented in this study can be used as part of evidence-informed policymaking by identifying the broadcast periods during which the highest frequencies of advertisements occur, the persuasive techniques that are used and identifying considerations for applying nutrient profiling models.