INTRODUCTION

Eighty-five percent of countries worldwide have implemented regulations to include nutritional information on pre-packaged foods on a compulsory or voluntary basis, mainly through nutrient declarations placed on the back side of pre-packaged foods¹. However, the extension of nutrition labelling to unpackaged foods or foods made and packaged at the point of purchase is gaining increasing attention worldwide given the growing contribution of foods prepared outside of home to overall diet^2,3. The increase in the consumption of foods away from home has been associated with an increase in the intake of total calories, total fat, saturated fat, trans-fatty acids, and sodium, and a decrease in the intake of fruits, vegetables and fiber^4,5.

To date, the implementation of nutrition labelling on unpackaged foods has been mainly focused on calorie labelling on restaurant menus². So far, no consistent results have been found for the effectiveness of calorie labelling on triggering changes in food choice and nutrient intake. Some systematic reviews have shown that this approach may be effective in reducing the total energy content of foods ordered and consumed in the away-from-home environment^6,7, whereas others have pointed out that displaying calorie labels alone may have a limited impact on consumer behavior⁸.

Research conducted in the United States has shown that the majority of consumers do not use menu calorie labelling when making their food choices in fast-food or full-service restaurants^9,10. In this sense, one of the main concerns of posting calories on restaurant menus is that the average consumer may not be able to understand the information¹¹. This matches results from several studies reporting that numeric nutrient declarations included on food packages are difficult to understand for consumers¹². For this reason, front-of-package (FOP) nutrition labelling schemes providing a simple and visible summary of the nutritional quality of foods have gained increasing attention worldwide¹³.

Nutritional warnings have recently emerged as a FOP nutrition labelling scheme to discourage consumption of foods with excessive content of nutrients associated with non-communicable diseases (NCDs)¹⁴. This FOP nutrition labelling scheme has been implemented as black octagons highlighting pre-packaged foods with excessive content of calories, sugar, fat, saturated fat and sodium, in several Latin American countries (Chile, Peru, Uruguay, and Mexico)^15,16,17,18. An increasing body of evidence shows that nutritional warnings are easy to understand, facilitate the identification of foods with excessive content of nutrients associated with NCDs, and discourage consumers from choosing them^19,20. In addition, warnings have been reported to be efficient at triggering changes in food choices, even immediately after their implementation^21,22.

Considering the increasing consumption of foods away from home, the extension of nutritional warnings to establishments selling unpackaged foods or foods made and packaged at the point of purchase should be foreseen. In this sense, it is worth highlighting that the restriction of nutritional warnings to pre-packaged foods was among the main concerns raised by the food industry during the public consultation held before the implementation of this public policy in Uruguay²³. Industry representatives stated that restricting warnings to pre-packaged foods was discriminative as they would be regarded as less healthy than those sold at retail food establishments. In response to the comments received during the public consultation for the regulation, the Chilean government identified the extension of nutritional warnings to retail food establishments as a relevant topic for the medium-term policy agenda²⁴.

The extension of nutritional warnings to retail food establishments is particularly challenging as it requires deciding how to display information that can capture consumers’ attention and serve as a relevant cue for making choices²⁵. Retail food establishments such as bakeries, delicatessen, and fast food restaurants, offer a wide range of alternative approaches to display warnings. For this reason, considering consumer perspective on how to implement nutritional warnings in those establishments can reveal innovative and useful insights that may not be obtained by solely relying on experts²⁶. In particular, an important advantage of this approach is the opportunity to get closer to the needs and expectations of consumers.

So far, the only application of warnings in this setting is a regulation from the New York City Department of Health and Mental Hygiene that requires the inclusion of sodium warnings on the menus of chain restaurants, next to any food item containing more than 2.300 mg of sodium²⁷.

METHODS

An online cross-sectional study was conducted in April 2019, during the period granted to the food industry to include nutritional warnings on food packages. Four types of retail food establishments were explored: bakeries, delicatessens, fast-food retail and online food ordering. These establishments aimed to capture a diversity of settings selling unpackaged foods or foods made and packaged at the point of purchase in the presence of consumers. The decision of including online food ordering grounded on the accelerated growth in the availability and use of these platforms in the country in the last years. It is worth noting that sit-down restaurants were not considered in the present study because the alternatives for the inclusion of warnings are mainly limited to the menu. Indeed, so far, the inclusion of nutrition labelling on unpackaged foods has mainly focused on menu calorie labelling.

The study protocol was approved by the Ethics committee of the School of Chemistry of Universidad de la República, Uruguay.

Participants. A convenience sample of participants was recruited using social media. This methodological decision was made based on the fact that most Uruguayan citizens of all socio-economic groups had internet access and the widespread penetration of social media among the Uruguayan population. In 2019, 80% of Uruguayan citizens older than 14 years old accessed the internet on a daily basis³¹. Facebook was the most popular social media in the country, being accessed daily by approximately 50% of Uruguayan adults between 20 and 64 years old³¹. Meanwhile, Instagram was mainly popular among adolescents (14-19 years old) and young adults (20-34 years)³¹.

To increase the external validity, participants were recruited using an advertisement on Facebook and Instagram targeted at adults (>18 years) living in Uruguay, selected by the Facebook software. The advertisement included the text “Help us think how we can have more information in establishments that sell prepared foods”, accompanied by a picture of a bakery shelf. To encourage the participation of people not largely involved with food, participants were given the chance to enter a raffle for a voucher worth 70 US dollars.

The advertisement was delivered to 31,400 social media users, randomly selected by the Facebook software. A total of 2,300 people clicked on the link, 703 accepted to be part of the study and 547 completed the whole study. As shown in table 1, the sample was diverse in terms of age, gender and socio-demographic characteristics. However, the sample cannot not be regarded as representative of the Uruguayan population as it underrepresented young people, males and people with low education^32,33.

Questionnaire. Participants clicked on the study advertisement and were re-directed to the website where the online questionnaire was hosted (Compusense Cloud, Compusense Inc., Canada). They provided informed consent through an online form at the beginning of the study. Then, participants were introduced with the following text: ‘In August 2018, Uruguay approved a law that states that packaged processed food products containing excessive amount of sugar, fat and sodium should feature warnings (as those shown in the figure below) on the front of their packages’, accompanied with an image showing the warnings design (black octagons with white borders, including the word “Excess”, followed by the name of the nutrient and the initials of the Ministry of Public Health). Participants were asked to indicate to what extent they agree this information to be displayed at each of the following retail food establishments selling unpackaged foods: bakeries, delicatessens, fast-food restaurants and online food ordering. A 5-point Likert agree-disagree scale (1= completely disagree, 5= completely agree) was used for this purpose. The wording of the question was: “What is your degree of agreement with the inclusion of this information in the following establishments selling foods and prepared meals? (In Spanish: ¿Cuán de acuerdo estás con que esta información también aparezca en los siguientes puntos de venta de alimentos y comidas preparada?). After that, participants were requested to provide ideas on how this information could be included in each of the 4 retail food establishments mentioned above by answering the following open-ended question: “How would you include the nutritional information we showed you before in the foods sold at (name of the establishment)?” (In Spanish: ¿Cómo incluirías la información nutricional que te mostramos antes en los alimentos que se comercializan en (nombre del establecimiento)?). No word limit was implemented. The order in which the different retail food establishments were presented to participants was randomized, following a Williams’ Latin square experimental design to avoid order effects. Finally, participants indicated their age, gender and educational level and answered a series of questions to estimate socio-economic status according to a standard Uruguayan methodology³². Participants took between 4 and 26 minutes to answer all questions. The study was conducted in Spanish and translated to English for publication purposes.

Data analysis. Responses to the Likert scales were analyzed using descriptive statistics. The percentage of participants providing responses indicating agreement (i.e., scores higher than 3 in the 5-point Likert scale) was calculated. For each establishment, a binary variable was created indicating if each participant agreed with the inclusion of warnings (i.e., provided a score higher than 3; no: 0 and yes: 1). A generalized linear model (glm) was used to evaluate the influence of gender (male/female), age group (18-30/31-45/46-60/older than 60 y), educational level (primary education/secondary education/technical education/university degree/postgraduate studies) and socio-economic status (low/medium/high) on the likelihood of agreeing with the inclusion of warnings in each of the 4 types of establishments. A significance level of 5% was considered in the analysis and when the effects were significant, post-doc differences were calculated using Tukey's test.

Responses to the open-ended questions asking how nutritional warnings could be implemented in each of the establishments were analyzed through content analysis based on inductive coding³⁴. Two researchers were involved in the coding process. For each retail establishment, researchers coded each of the responses into categories, as they emerged from the text. Categories corresponded to participant insights on how to implement the warnings in each type of establishment. After they had individually coded the data, a meeting was held to define the final categories by consensus. In order to identify the most relevant suggestions for the inclusion of nutritional warnings in retail food establishments, the percentage of participants who mentioned responses within each of the identified categories was calculated. Categories were grouped into dimensions following the same procedure described above. While data analysis was performed in Spanish, categories, dimensions and exemplar responses within each category were translated into English for publication purposes.

RESULTS

Agreement with the inclusion of nutritional warnings in the 4 types of retail food establishments. Results from the Likert scales showed that the inclusion of nutritional warnings in retail food establishments selling unpackaged foods was positively perceived by participants. Regardless of establishment type (bakery, delicatessen, fast-food retail or online food ordering), the majority of the participants (93-94%) provided scores that indicated agreement with the inclusion of warnings. For the 4 types of retail establishments, results from the glm model showed that agreement with the inclusion of nutritional warnings was not significantly influenced by age, gender, educational level or socio-economic status (all p-values higher than 0.074).

Insights on how to implement nutritional warnings on retail food establishments. Responses to the open-ended questions provided consumer insights on how warnings can be implemented in different types of retail food establishments. As shown in tables 2 and 3, the majority of the responses were coded within dimensions that implied that warnings should be visible before making a purchase decision (e.g. in a visible in-store placement, next to products). Responses proposing the inclusion of warnings in places that would only be visible after having bought the products were less frequent for the 4 types of establishments, as evidenced by the frequency of responses within dimensions such as Take-home information or Food delivery.

For bakeries and delicatessens, the majority of participants indicated that warnings should be included next to products, on the retail display case where products are exhibited (Table 2). Visible in-store placements such as menu boards were the second most frequent response for these establishments.

Regarding fast food restaurants, participants identified menu boards as the best place to include nutritional warnings (Table 2). Moreover, the inclusion of nutritional warnings on the wrapping paper or take-out packaging was mentioned by 19% of participants.

In the case of online food ordering, participants highlighted that warnings should be visible while exploring available options. As shown in table 3, most participants suggested that warnings could be included next to the product description, name or picture.

DISCUSSION

Nutritional warnings are gaining popularity in Latin America as a policy option to discourage consumption of foods with an excessive content of nutrients associated with NCDs^15,16,17,18. Although warnings have been restricted to pre-packaged foods, their extension to unpackaged foods or foods made and packaged at the point of purchase deserves consideration by policymakers. Participants from the present study showed a high level of support for the inclusion of nutritional warnings in 4 types of retail food establishments: bakeries, delicatessen, fast food restaurants and food ordering websites. This result is in line with the high acceptance of nutritional warnings as front-of-package nutrition labelling scheme in pre-packaged foods^35,36, as well as with results from previous European studies showing that nutrition labelling is one of the policies with the highest public acceptance^37,38.

Gender, age, educational level and socio-economic status did not introduce significant variations in participant agreement with the implementation of nutritional warnings in retail food establishments. This suggests that support of this policy seems not to depend on socio-demographic characteristics. Similar results have been reported for public support for the implementation of warning labels on pre-packaged foods^35,36.

The present work also gathered consumer insights on how to implement nutritional warnings in different types of retail food establishments. Participant responses highlighted that warnings should be highly visible at the point of purchase, so that they become a relevant cue for making purchase decisions. In particular, the retail display case was identified as a key location for the inclusion of warnings in bakeries and delicatessens, whereas menu boards were the most frequently mentioned location for fast food restaurants. In a similar vein, when considering online food ordering, most participants posited alternatives that make sure warnings are visible and front of mind for consumers when deciding what to eat. It is worth highlighting that beyond the inherent differences between physical and online food retail establishments, results provide further insights on the relevance consumers attach to having access to nutritional warnings at the time of ordering, so they can take this information into account for deciding. Capturing consumer attention has been extensively recognized as a key determinant of nutrition labelling efficacy³⁸. In the short time consumers invest making their purchase decisions²⁵, nutritional warnings need to easily capture their attention so that they can incorporate the information they provide in their decision-making process. Nutritional warnings have the potential to make the high content of sugar, fat and sodium of foods salient in the mind of consumers, discouraging them from selecting these products³⁹. Based on the reported advantages of nutritional warnings over other front-of-package nutrition labelling schemes¹⁹, it can be hypothesized that they could be more efficient at encouraging healthier food choices than calorie labelling.

Although consumers were positive about the inclusion of warnings, policy makers need to be aware of the many operational challenges that the food service industry would face for their implementation, including difficulties for accessing accurate nutrition information and the loss of flexibility for menu changes⁴⁰. Moreover, the complexity of implementing such a policy may vary depending on the type of retail food establishment. In this sense, the standardization of recipes, a laborious but needed step to provide accurate nutritional information, may be even more complex when considering products involving numerous ingredients (e.g. menu options in fast food restaurants) or when ‘variable’ menu items (e.g. different pizza toppings) are offered⁴⁰. Policy makers should develop consistent implementation strategies and provide support to relevant actors to overcome these obstacles⁴¹.

Although exploratory in nature, the present work provides valuable insights on how to implement the extension of nutritional warnings to unpackaged foods or foods made and packaged at the point of purchase. Further research on the topic is needed to assess the impact of implementing nutritional warning (taking input from the present work) on consumer ordering decisions under different settings.

Limitations of the study. The main limitation of the present work is that participants cannot be regarded as representative of the Uruguayan population as they were recruited using social media. However, it should be noted that the use of social media is widespread in the country and that a diverse sample of participants in terms of gender, age, educational level and socio-economic status was considered. Secondly, participants received no direct compensation for their participation, suggesting that the sample may overrepresent citizens with a high involvement with food and nutrition. Finally, although the study suggested a high level of support for the implementation of nutritional warnings in retail food environments, it is worth highlighting that responses to the Likert questions may be affected by social desirability bias.