An evaluation of Chile’s Law of Food Labeling and Advertising on sugar-sweetened beverage purchases from 2015 to 2017: A before-and-after study – PLoS Blogs

Introduction

In recent decades, consumption of sugar-sweetened beverages (SSBs) has rapidly increased across the globe [1,2]. Excess intake of these beverages has been linked to increased weight gain, glucose dysregulation, and development of noncommunicable diseases such as type 2 diabetes [3–7]. Compared to foods, SSBs are uniquely harmful because they contain large amounts of calories, which can be rapidly absorbed and are less satiating, leading to inadequate caloric compensation at other eating occasions and contributing to overall positive energy balance [8]. Around the world, public policies are an important and increasingly common strategy being used to reduce consumption of these beverages and prevent continued increases in obesity and related diseases [9–12]. Recently, fiscal policies such as taxes on SSBs have been the predominant approach for reducing intake [11]. Over 42 countries and six United States cities have implemented SSB taxes or increased an existing SSB tax, with most of these implemented in the last decade [13]. Although impact on SSB purchases depends on tax design and rates, these policies have generally led to reductions in SSB purchases equivalent to the price increase on SSBs [14–18]. Evidence also suggests that SSB taxes affect low-income groups and high-SSB consumers the most [16,17,19].

Additional policy strategies to reduce SSB consumption include mandatory front-of-package (FOP) warning labels, governmental restrictions on marketing SSBs, and bans on SSB sales and promotion in schools [20]. In particular, FOP warning labels have become the focus of many health scholars and advocates, gaining favor over the heretofore more common positive FOP labels (e.g., a health seal or stamp) and other voluntary FOP systems such as traffic-light labels [21–26]. Despite global momentum behind FOP warning-label policies, virtually no evidence exists yet on changes in SSB purchases following implementation of a warning-label policy [27]. Similarly, little is known about the effects of national school sales bans or mandatory restrictions on unhealthy food marketing to children and subsequent changes in food and beverage purchases [28]. Finally, despite calls for implementation of more comprehensive “packages” of obesity prevention policies [10], the joint impact of a set of policies on SSB purchases is unclear, as most national-level food and beverage policies been implemented one at a time.

Chile, a high-income country with high levels of SSB intake [1,29], has implemented a uniquely comprehensive set of obesity prevention policies regulating how SSBs and other energy-dense, nonessential foods are packaged, marketed, and sold. The first of these regulations, implemented October 2014, increased Chile’s existing tax on SSBs from 13% to 18% for high-sugar beverages and decreased the tax from 13% to 10% for low-sugar beverages. One evaluation found that in the first year postimplementation, this modification resulted in only small price increases and 3.4% declines in SSB purchases [30]. In June 2016, Chile implemented the Law of Food Labeling and Advertising, which included the first national system of mandatory FOP warning labels for SSBs and energy-dense, nonessential foods [31]. Similar warning-label policies have since been adopted in Peru, Uruguay, and Israel in 2018, and policies are currently under public review/potential finalization phases in Brazil and Mexico, among others. Chile’s Law of Food Labeling and Advertising also exacts comprehensive restrictions on child-directed marketing of SSBs and nonessential, energy-dense foods to children under 14 years of age, as well as restrictions on the promotion and sales of these products in schools [31]. Chile’s marketing regulation, in particular, is more comprehensive than other countries’ in that it restricts unhealthy food marketing on more products across a wider range of media, and it prohibits use of more marketing techniques [28]. Understanding how Chile’s policies on labeling and marketing—and, to a lesser extent, the school sales ban—are linked to changes in household purchases is critical for developing evidence-based obesity prevention policies across the globe.

It is also important to understand whether these policies had a differential effect by socioeconomic status (SES) to ensure that such policies do not inadvertently increase SES-related diet disparities, especially in regions such as Latin America, where the burden of obesity is shifting to lower-SES individuals and households [32–34]. Education is a particularly important measure of SES when evaluating labeling policies, as education may influence purchasing decisions by affecting how well an individual is able to understand the information communicated in nutrition education messages or on food labels [35–37]. Education is also correlated with higher financial and nonfinancial resources [38], which can likewise influence food choices. Moreover, it is currently unclear whether low- versus high-educated households would have differential responses to other policies, such as marketing restrictions or school sales bans on high-in beverages. Thus, it is important to examine whether low- versus high-educated households show greater changes in beverage purchases after Chile’s labeling regulation was implemented.

This study’s objective is to use longitudinal data on household beverage purchases made in stores to examine changes in the volume, calorie content, and sugar content of beverage purchases following implementation of the Chilean labeling and marketing regulation, overall and by household educational attainment.

Methods

This study was reviewed and approved by the University of Chile Institutional Review Board (IRB) and is exempt from review by the University of North Carolina, Chapel Hill IRB as the study uses secondary, de-identified data. This study and protocol were registered with the Open Science [39] and protocols.io [40].

Results

Discussion

The main finding of this study is that following implementation of the Chilean Law of Food Labeling and Advertising, household purchases of high-in beverages decreased 23.7% compared to the counterfactual, or what would have been expected based on preregulation trends. This translates into roughly 12 fewer calories and 2.7 fewer grams of sugar purchased per capita per day from high-in beverages. In contrast, volume of not-high-in beverage purchases increased by 4.8%, translating into 5.7 calories and 0.7 grams of sugar purchased per capita per day. This increase in not-high-in beverage purchases was not commensurate with the decrease in high-in beverage purchases. Indeed, there was a relative decline in total beverage purchases of 2.2% for volume, translating into a decrease of 7.4 calories and 1.7 g sugar of beverage purchases per capita per day. Among beverage subcategories, we found the largest reductions in consumption among high-in fruit drinks and dairy (42.9% and 28.8% percentage point reductions in the percent of households consuming these drinks, respectively).

Households with higher educational attainment had larger relative reductions in high-in beverage purchases than did households with less education. This may be because higher-educated household had lower baseline purchases of high-in beverage purchases than did low-educated households, creating a larger relative difference compared to the counterfactual in the postregulation period. However, the interaction term for education, regulation period, and time was also statistically significant for high-educated households versus low-educated households, suggesting higher-education households had a differential response to the regulation. This is consistent with previous literature finding that higher-educated individuals have higher levels of health literacy [54] and may be more likely to use and understand nutrition labels [55–58]. Absolute reductions were similar in amount between highest- and lowest-educated households. These results may be concerning, since similar absolute reductions in high-in beverage purchases could lead to an increase in the relative differences in SSB consumption between high- and low-educated households, since low-educated groups had higher baseline consumption levels. This potential for an increasing disparity in SSB consumption is problematic considering that in Chile, individuals with less education already have higher levels of obesity [59,60]. Future research will be needed to further understand whether these policies increase or reduce disparities in diet and health by educational attainment. Additional research will also be needed to examine changes in purchases among those who were top consumers of high-in beverages prior to the regulation, since previous SSB policy evaluations have found top consumers typically reduce purchases more than do low consumers [47].

Although the Chilean law included a package of policies implemented at once, it is interesting to consider how the effects observed here compare to results from evaluations of previous Latin American policies focused on SSB reduction, primarily SSB taxes. Compared to their respective counterfactuals, SSB purchases in Chile declined by 23.7% under the Chilean Law of Food Labeling and Advertising (22.8 mL/capita/day). This relative decline is larger than or similar to previous estimates of reductions in SSBs after Chile’s earlier SSB tax modification, which raised the tax on high-sugar drinks from 13% to 18% (one SSB tax evaluation study found a 3.4% decline in SSB purchases [30], whereas a second study found a 21.6% decline [61], though both studies found only an approximately 2% increase on paid price for SSBs, making the latter study’s large drop in SSBs unlikely based on price elasticity [62]). The absolute and relative decline in SSB purchases after the Chilean Law of Food Labeling and Advertising was also larger than the declines in SSB purchases observed after Mexico’s 10% SSB tax, which was associated with an average 7.6% posttax reduction in SSB purchases (12 mL/capita/day) 2 years after the tax [17]. The observed decline in the present study was also larger than the average decline in SSB purchases after a 10% SSB tax, as estimated by a recent meta-analysis [63].

On the other hand, it is unclear how these results will compare to larger SSB taxes, which, as expected, are likely to have larger effects on purchases, as indicatd by recent evaluations of Philadelphia’s beverage tax [64]. More generally, it is complex to draw comparisons between SSB policies implemented in different settings and times and with different study designs, at least in part because these real-world evaluations rely on pre-post observational data and cannot assess causality. Based on available evidence, however, the changes in SSB purchases relative to counterfactual estimates following implementation of Chile’s Law of Food Labeling and Advertising are larger than those observed after most Latin American SSB taxes. The relatively large effect observed here is not surprising, given that the Chilean Law of Food Labeling and Advertising included a package of policies targeting different aspects of consumer behavior, whereas most SSB taxes have been implemented as standalone policies. These results suggest that policymakers and public health advocates should consider a package of policies, including FOP warning labels, marketing restrictions, and school sales policies alongside SSB taxes as important strategies for reducing population-level purchase and intake of SSBs.

Although there have not yet been any evaluations of a real-world mandatory FOP warning-label policy, the estimated decline in high-in beverage purchases relative to the counterfactual is also larger than what might be expected based on recent meta-analyses of food labeling policies. For example, Shanguan and colleagues found that labeling was only associated with a 6.6% decline in calories purchased [65], whereas Crockett and colleagues largely found a null effect of labeling interventions across outcomes [27]. However, these studies examined a diverse array of labeling systems, including voluntary FOP systems, such as the Guideline Daily Amounts (GDAs) and traffic-light labeling, as well as back-of-package nutritional labeling, in a variety of settings (stores, schools, vending machines, cafeterias, and restaurants). Recent experimental studies in Latin American populations have found that the style of FOP warning labels used in Chile is easier to understand and more likely to discourage consumption than other types of FOP labels, such as the GDAs or traffic-light labels [66–68]. In addition, a recent US-based randomized experiment found that nutrient-based FOP warning labels reduced SSB purchases by 31 calories per interaction, or 22% [69], which is similar to the effect observed in this study.

In addition, along with the warning-label component, the Chilean law includes marketing and school sales restrictions, which likely contributed to the larger effect found here. A recent evaluation found that schools reduced the percent of products that were high-in from 90.4% before implementation to 15.0% after implementation [70], which may have influenced children’s preferences and the subsequent household purchase made by their parents. Indeed, a separate qualitative study found that schools were a key promotor of behavioral change relating to the labeling component of the law: children learned about the regulation at school and then encouraged their mothers to purchase nonlabeled foods and beverages for them [71]. In addition, an evaluation of Chile’s food marketing regulations found that children and adolescents’ exposure to unhealthy food and beverage advertisements on television was reduced by 44% and 58%, respectively, in the year following implementation of the law [72], although it is unclear how reductions in exposure translate to changes in purchasing behavior. Another consideration is that as these regulations were being enforced, the Chilean Ministry of Health launched a mass media campaign to inform consumers about the meaning of the warning labels and encourage them to choose products with fewer labels, potentially further strengthening the effect of the law. Unfortunately, we are unable to disentangle the effects of each of the labeling, marketing, and school sales ban components of the law, here. Future research will be needed to understand the individual impact of each policy as well as how the policies and media coverage and promotion interacted to lead to reductions in high-in beverage purchases.

This study has important limitations. As previously mentioned, the main limitation is that this is an observational pre-post study and thus unable to assess the causal impact of the law or disentangle the drivers of the observed reductions in high-in beverage purchases. The reductions in purchases of high-in SSBs found in this study likely reflect a combination of changes in consumer behavior (e.g., consumers choosing not to purchase a high-in beverage) and industry behavior (e.g., product reformulations that could shift products from high-in to not-high-in status, or other industry actions such as changes in marketing strategy or pricing changes). For example, after the policy, we observed that there was a large decline in the percent of households who purchased high-in fruit drinks and a sizeable increase in the percent who purchased high-in fruit drinks after the policy. Currently, it is not clear whether this is because consumers were choosing to switch from high-in fruit drinks to not-high-in fruit drinks or whether they purchased the same or similar beverages, but these beverages were reformulated under the nutrient thresholds and thus no longer subject to the regulation (and thus no longer classified as “high-in”). Future research to understand the mechanisms of how this regulation changed beverage purchases will be important for designing future policies. For example, regulations intended to incentivize reformulation may be quite different than those designed to influence consumer behavior.

It is also important to note how different analytical approaches may have affected results. For example, the difference between observed and expected purchase amount depends on a comparison of the slope of the preregulation trend with the slope of the postregulation trend. Underestimating the preregulation trend could, therefore, lead to an underestimated counterfactual and, ultimately, an overestimation of the difference between slopes in the postregulation period.

Similarly, the specified length of the preregulation period can affect the preregulation trend, thereby affecting the counterfactual comparison. One potential limitation of this study is that we have a relatively short preregulation time period (January 2015 to June 2016). Notably, results were consistent even in sensitivity analyses in which we expanded the preregulation period to January 2014 to account for Chile’s October 2014 SSB tax modification (i.e., we account for the fact that there may have been a steeper preregulation trend between October 2014 and June 2016 due to the change in SSB tax). It will be important for future policy research to consider the effects of including different preregulation time periods on results, especially in countries where multiple policies are implemented over time.

Our analyses also included only a year and a half of postimplementation data. The results show a significant reduction in purchases of high-in beverages immediately after the regulation was implemented, with no change in the postregulation trend (i.e., there was no attenuation or increase in effect over time). These results could be due to the relatively short postimplementation period. Over time, we might expect to see an increase in the effect as social norms shift and as people learn more about the health harms of high-in beverages. Alternately, we could see the effect decrease if consumers grow accustomed to the warning labels and the “novelty effect” wears off, as has been shown in tobacco labeling [73]. Moreover, the nutrient thresholds became stricter (i.e., more products became high-in and were subject to the warning label and marketing and school restrictions) in June of 2018. Longer-term analyses will be needed to understand the effects of the law as it gets stricter over time, as well as to understand whether purchases of high-in beverages remain low, continue to decline, or rebound over time.

Another limitation is that our work evaluates changes in high-in beverages without regard for which specific nutrient(s) exceeded thresholds (sugar, saturated fat, sodium, and/or calories). Although for beverages, the vast majority were regulated as high-in because of excess sugar content, future research should explore potential differences in purchasing according to specific high-in nutrient or the number of high-in nutrients, since the more nutrients that are regulated, the more warning labels a product will carry.

The study also has several limitations relating to generalizability. For example, the sample is limited to only urban-dwelling households. However, 90% of Chile’s population is urban, suggesting that the results are generalizable to most Chilean households [41]. Regardless, it will be important for future research to understand whether policies affect purchases differently in rural areas versus urban cities, as well as by geographical region. Similarly, the study only includes beverage purchases made at stores, including supermarkets, grocery stores, and convenience stores. Store-bought beverages account for roughly 88% of nonalcoholic beverage sales in Chile, and this proportion has remained consistent before and after the regulation [74]. However, future research should examine whether there have been any changes in the type or quantity of beverages purchased from other sources (e.g., in restaurants, schools, or homemade), as well as whether actual dietary quality of beverages across sources has changed. For example, if the healthfulness of store-bought beverages improved, but the healthfulness of restaurant beverages declined, this could offset improvements in diet quality and blunt any subsequent health gains. Additional work will also be needed to understand changes in food purchases and food intake after these regulations, as well as how consumers potentially substituted between foods and beverages. Examining the effects of these regulations across total purchases and diet will be essential for understanding whether this package of policies is likely to achieve its goal of obesity prevention.

It is also important to note that this study evaluated beverage purchases only after the first phase of implementation in 2016. In July 2018, the second phase was implemented, including stricter nutrient thresholds as well as a new marketing law that restricts from 6:00 AM to 10:00 PM all television and cinema advertising for high-in foods and beverages and requires health promotion messages to be included when products are advertised outside those hours. This 6:00 AM–10:00 PM restriction is considered to be the most comprehensive marketing restriction on unhealthy foods and beverages implemented to date [28]. In July 2019, the third and final phase of the regulation was implemented with the most stringent nutrient thresholds. Future research on both foods and beverages will be needed examine the impact of the Chilean regulation following implementation of the new marketing law and the final set of nutritional thresholds in 2019 [31] and whether the effects of these regulations wear off or accelerate over time.

This evaluation has important policy implications. Three other countries (Israel, Uruguay, and Peru) [75] have adopted FOP mandatory warning labels, and Mexico and Brazil are among countries considering warning labels and are in consultative stages. However, the majority of these adopted or proposed regulations focus on warning labels only and do not include other components of the Chilean regulation, such as the restrictions on marketing or school sales and promotions. Multicountry evaluations may be helpful to disentangle the effects of each of these regulations, as well as to understand how the regulations work in different populations.

In conclusion, this study describes changes in SSB purchases following introduction of Chile’s policy package that includes FOP warning labels, child-directed marketing restrictions, and restrictions on sales in schools of unhealthy foods and beverages. After implementation of this policy package, purchases of high-in beverages declined by nearly 24%; these reductions are larger than those observed after standalone SSB reduction policies in Latin America, such as taxes. Future research should examine the differential effects of the labeling, marketing, and school policies; whether changes in purchases were driven by industry or consumer behavior; and the effects of these policies on SSB intake.

Supporting information

Acknowledgments

We would like to thank Ms. Emily Busey for her contributions proofreading and organizing files and Dr. Donna Miles for data organization and descriptive analyses.