An assessment of compliance with proposed regulations to restrict on-package marketing of packaged foods to improve nutrition in South Africa

Abstract

Background

Ultra-processed products (UPPs), high in sugar, fat and sodium, contribute to increasing burdens of obesity and non-communicable diseases in South Africa (SA). Increased consumption and acceptability of UPPs is driven, in part by aggressive marketing. Front of package labelling laws, particularly warning labels (WLs) coupled with marketing restrictions, have been recommended to improve consumer awareness of unhealthy products, incentivise reformulation and underpin other interventions. The SA government published a draft regulation to introduce WLs and restrict marketing, called R3337. This study aims to analyse on-package marketing and create a baseline to evaluate compliance with the proposed R3337.

Methods

This study is an analysis of a dataset of 6768 packaged food products, collected as pictures of products in four major grocery stores in Cape Town, SA between March and July 2019. A codebook was developed based on R337. The codebook was used to identify and assess child-directed marketing (CDM) and health claims on products potentially subject to front-of-package WLs. Food packages and their nutritional information were coded and analyzed to determine whether the product would be required to carry a WL and whether the package included regulated marketing.

Results

The majority (80.16%) of products analyzed would require WLs under R3337 due to high levels of sugar, sodium, or saturated fat. CDM was prevalent, with 59.26% of products displaying some form of marketing, primarily animated characters and depictions of children. Health claims were also common, with 83% occurring on WL products.

Discussion

Using a WL system in SA could effectively target common forms of on-package marketing of unhealthy products, given the high prevalence of CDM and health claims on products likely to carry WLs. Such interventions may contribute to improving diets and reducing UPP consumption. Some forms of CDM are not regulated under R3337 and we recommend expanding the regulation to include these. This study provides a baseline which should be used for future research to evaluate R3337 when it is implemented.

Introduction

Ultra-processed products (UPPs) which contain excessive amounts of sugar, sodium, and saturated fat, and unhealthy diet are significant contributors to the growing prevalence of obesity and non-communicable diseases (NCDs) globally [1, 2]. This is partly due to the pervasive availability and aggressive marketing of these UPPs [3, 4]. In South Africa (SA), as in many other African countries, there is a growing burden of NCDs [5, 6]. There has also been, over the past two decades, an increasing proliferation of UPPs in the food supply, leading to both increased availability and consumption of UPPs [7]. A recent study indicated that 76% of the packaged food products in SA grocery stores were ultra-processed [8]. Low consumer awareness of the health implications of consuming these products and extensive marketing of these products make reducing consumption a challenge from a public health perspective that requires intervention [9, 10].

Simplified nutrition labelling schemes, such as front-of-package (FOP) warning labels and restrictions on child-directed marketing (CDM) have been identified as evidence-informed measures which may contribute to improving diet and preventing these diseases [11]. Interpretive FOP labelling systems such as the warning label (WL) systems adopted in South America, can target and identify products as unhealthy or harmful as well as inform consumers about the nutritional content of a product [8, 12, 13]. FOP labelling systems have been found to contribute to improved diet and nutrition by improving population understanding of nutritional content, assist consumers in changing purchasing behaviours and incentivize reformulation of unhealthy products [12, 14, 15]. SA is currently in the process of adopting a WL system that aims to achieve these objectives. In 2021, the SA Department of Health announced its intention to adopt a FOP WL system [16]. Draft regulations (R3337) to implement this system were published in 2023 and proposed introducing a context-specific, mandatory FOP WL system that would warn consumers of high levels of key ingredients in packaged foods and beverages [17]. Similar to the WL systems in South America, the SA scheme will require packaged products high in key ingredients (sodium, saturated fat, and sugar) or containing non-sugar sweeteners to carry a FOP WL.

FOP labelling systems can also underpin further diet-related NCD prevention intervention such as marketing restrictions or taxes on unhealthy foods [18]. Many countries, including Chile, Peru and Israel, that adopted WL systems specifically have coupled the labels with marketing restrictions, particularly on CDM [19]. This includes advertising restrictions as well as restrictions on the type of marketing food packages may carry [19]. The draft 3337 includes such measures restricting both CDM and other forms of on-package marketing for products that would carry WLs.

Research has shown that on-package marketing is a key marketing tactic for unhealthy foods and that particular strategies, such as cartoons, games and competitions, are used to target children through food packages [20,21,22,23]. In addition, on-package marketing in the form of health claims may lead parents to believe products are healthier than they are, and can influence parents to purchase unhealthier foods [24,25,26]. Not only is there a higher prevalence of marketing for unhealthy products, but as the scope for traditional advertising becomes increasingly regulated and costly, food packages remain an available avenue for marketing [27, 28]. Packaging restrictions on unhealthy foods have the potential to be effective intervention in improving diet and health, but their success depends on their design. If regulations are poorly crafted, food producers may exploit loopholes or shift to unregulated marketing strategies [29, 30]. Therefore, governments must thoroughly understand the marketing landscape to develop effective restrictions.

Substantial research has been done to inform the design and selection of a labelling scheme as well as the nutrient profiling model (NPM) underpinning the WLs to ensure that R3337 is context-specific and responsive to South African consumers and its food supply [8, 13]. There are a large number of products in the South African food supply which, based on their current nutritional information, would be required to carry WLs under the scheme [8]. However, the marketing-related aspects of the regulation and their potential impact are not as well understood and have not previously been investigated. This study seeks to address this gap.

This study aims to analyse the nutritional composition and on-package marketing of products in the SA food supply prior to the implementation of R3337 to create a baseline dataset which can be used to evaluate R3337 when it is implemented.

Methods

Study setting and data collection

The study took place in Cape Town, SA. This study analysed nutrition fact panel (NFP) dataset collected between March and July 2019. The methods for data collection in the NFP dataset are described elsewhere but in brief, the dataset was collected by fieldworkers capturing photographs of food products in six major, national grocery retailers in Cape Town, Western Cape in both middle-income and low-income suburbs [8]. The stores sampled account for more than 50% of the grocery retail market in South Africa and thus provide a representative sample of packaged foods in SA’s formal food supply [8].

The 2019 dataset was selected as being before the government announced their intention to implement a FOP WL scheme in 2021. The dataset consists of information collected from photographs of products, including the Nutritional information panel (NIP), barcode, ingredients, marketing related information and illustrations and the prices of the foods captured [8].

The dataset was previously cleaned to remove duplicate entries or those with missing data (such as missing or unclear photographs or a lack of nutritional information panel (NIP) [8]. Any product with the same product name, brand, flavour, package size and matching barcodes as photographed in a previous store were considered duplicates but products that were the same except for the size were not considered duplicates as marketing may differ based on purchasing different sizes – such as “value” or “bulk” size discounts. A limitation of the dataset and our subsequent analysis is that a number of products did not have an NIP, which is presently not compulsory on food packages in SA. A total of 6768 products were included in our study. The products were categorized into fourteen (14) categories (see Table 2 below).

Data extraction and coding process

R3337 prohibits certain types of CDM and health claims on products carrying WLs. We drew on legal expertise to create a detailed matrix of the types of regulated marketing and develop an analytical framework to assess for the presence of indicators and compliance with R3337 in the 2019 NFP dataset. We developed a codebook by reviewing and analyzing R3337 to identify prohibited CDM strategies and regulated health claims. Indicators were developed for these aspects of the regulation using CDM strategies and used to modify and supplement an existing CDM codebook developed by Khan et al. (2023) which had previously been applied to assess on-package marketing more generally [22]. This method of content analysis of on-package marketing has been used in other studies of on-package marketing [21, 23, 26, 31, 32].

The term “depictions” was interpreted to include references in text or pictures or pictorial representations. Besides identifying forms of regulated marketing, the codebook was also used to identify common CDM strategies that were not regulated, including the promotion of sports and other events, depictions of children without adults, bundling of packages, references to school and education and whether the product encourages an active lifestyle (as an example of health-washing) [22]. The inclusion of these unregulated forms of marketing was to assess the extent of regulatory escape for marketing under R3337. The regulation of health claims is slightly more complex than CDM. Given the study aims, we limited our analysis to whether a regulated health claim was present on a product that would be subject to a WL.

A REDCap questionnaire was utilized for data extraction and included product name, barcodes, product size and marketing information. A total of 6 data capturers were trained on the questionnaire, which included coding pre-selected sets of products until each capturer’s entries were consistent with a model code set. Initial data entries were double-coded and data entries were reviewed by team members on an ongoing basis to ensure consistency, minimise bias and quality assurance. In cases where there was uncertainty with regards to a product, the product was discussed by the research team to obtain consensus.

Analysis

Content and statistical analysis was undertaken. Analysis consisted of two stages. The first stage established whether the product would be required to carry WLs based on the WL NPM in Draft R3337. The second stage was to analyse the frequency and types of CDM and health claims on products.

Under the regulations, products carrying WLs are prohibited from carrying on-package CDM and health claims and thus, whether a product is subject to carrying a warning label is a proxy for whether the marketing on the product is permitted or not. It is noted that a product must meet additional nutritional requirements to carry a health claim which this study did not consider. The analysis was thus limited to only whether a product would be prohibited from carrying a health claim or CDM. We did not analyse whether, in terms of the regulation, the products not carrying a warning label were eligible to carry these claims under the health claim NPM.

For the first stage of analysis, Stata statistical software for data science version 17 of 2021 was used to run analysis of the nutritional content and apply the NPM The nutritional content of the products was analysed using R3337’s NPM (see Table 1) to identify which products would be required to carry FOP WLs based on the nutritional information captured.

Table 1 R3337 nutrient profile model nutrient cut-off values

The NPM was only applied to products that met the criteria for assessment (containing free sugar, added saturated fat, added sodium or non-sugar sweetener). Products will be required to carry a WL if their contents of sodium, sugar or saturated fat exceed the threshold set in the regulation or if they contain any amount of non-sugar sweetener. R3337 prescribes specific labels applied for each of the key ingredients and so the product may carry between one to four labels depending on its nutritional composition. The Mann-Whitney U test was used to compare the difference in mean nutritional composition of products with, and without, warning labels.

Descriptive analysis was also run using Stata to establish frequency and range of CDM and health claims across different categories. Pearson’s chi-squared test, or Fisher’s exact test (in cases where the cell count was below five), was used to understand and compare differences in on-package marketing in WL products and non-WL products. For all statistical tests a P-value of less than 0,05 was taken as significant.

Results

Overview

We found that a majority of products in our sample will be subject to carry warning labels. The products that would carry warning labels (WL products) had substantially higher levels of sugar, sodium and saturated fat.

We found 13115 instances of marketing across the packages of food products analyzed. This is a median of 1.996 instances of marketing per product, indicating that there is generally marketing present on food packages. Of these, 10 547 instances (80,04%) of marketing were found on WL products and 2568 instances (19,58%) were found on products that would not carry warning labels (non-WL products). The presence and levels of CDM across these categories of products were proportionate in both WL and non-WL products. However, there was a higher prevalence of health claims on WL products than non-WL products, with 83,05% of the instances of health claims occurring on products subject to carrying a WL. Our results will discuss first the presence of WLs on products in the sample and their nutritional composition before then discussing the presence of CDM and health claims on products.

Presence of warning labels on products

Table 2 provides a summary of the product categories, frequency and type of WLs. Of the products analyzed, 5425 products (80.16%) would carry one or more warning labels. Beverages (74.95%) had the highest number of products subject to carrying a WL, followed by confectionary (99.05%) and then cereals & cereal products (82.64%). Legumes and meat and dairy substitutes had the lowest number of products carrying warning labels.

The most frequent WL is that for high sugar content (3022, 36.1%), followed by saturated fat (2439, 29.13%), sodium (1895, 22.63%) and, non-sugar sweeteners (n = 1016, 12.14%) being the least frequent (see Table 3).

Table 2 Products and warning labels

Instances of CDM

As explained above, the regulations will prohibit WL products from depicting certain forms of CDM on their packaging. More than half the products in the sample (n = 4011, 59.26% of the total sample) carried some form of CDM on their packaging – this includes the regulated and unregulated types of CDM we identified (see Tables 3 and 4). Of these, regulated CDM was found on 3146 WL products and 865 non-WL products.

Table 3 Overview of nutritional content of products per 100g/100ml

Table 4 Instances* of regulated and unregulated CDM

We found 5989 instances of regulated CDM of which 4670 (77.98%) occurred on 3122 WL products. This represents the CDM that would be prohibited under the regulations and illustrates that a number of WL products carry more than one form of regulated CDM. The most frequent instances of CDM were animated characters or cartoons, and thereafter, depictions of children in mixed groups. Trademarked illustrations specifically were observed in 1023 products – 795 were WL products and 228 were non-WL products.

There were fewer products carrying unregulated CDM in both WL (412) and non-WL products (119). These instances occurred across 404 WL products and 116 WL products. The most frequent instance of unregulated CDM was the promotion of sporting events, followed by bundling of products and depictions of children without adults. The difference between WL and non-WL products carrying regulated CDM was statistically significant, though the differences between the types of CDM were not consistently significant – see for example cartoons, the use of children and family senarios and “new” or “limited edition” products.

Health claims

As with CDM, the regulations contain prohibitions on certain types of health claims, some general prohibitions and others specific to WL products. This includes the following claims:

• Claim that the product contains sweetener

• Claim that the product has reduced or limited amounts certain ingredients or nutrients

• Claim that the product is “High In” a particular nutrient

• Claim that one may get health benefits from consuming the product

• Endorsements / Health Seals from a medical institution

• A seal regarding NCDs

• Claim that the product is fresh or natural

• Claim that the product had a “purity” claim (such as being organic, non-GMO, no colourants etc.).

There were 3720 products which carried a health claim which would be regulated under R3337 (see Fig. 1). Of these, 2971 were on WL products and 759 were non-WL products.

Fig. 1

Instances of health claims

We observed 6531 instances of health claims on products, of which 5424 (83%) were on WL products.

The most common health claims on products were characteristic claims such as the product being organic, not genetically modified or being free from colorants, preservatives or flavourants (2291 WL products and 520 non-WL). Claims regarding the product being “fresh” or “natural” were common in WL products (1385) and uncommon in non-WL products (35). The claim of a product having reduced or no presence of a particular ingredient, such as sugar, sodium or saturated fat was common in non-WL products.

The regulations introduce stringent criteria for the types of institutes that may provide endorsements or health seals on products. In the sample, health seals were given by the Heart and Stroke Foundation, CANSA, the Diabetes Association, Weigh-less and the Glycemic Index Foundation. We also observed the endorsement that the product was recommended by “doctors or dieticians”. Of the 137 seals or endorsements, 103 of them related to NCDs.

Discussion

Our study shows that packaged products in the SA food supply are overwhelmingly high in the ingredients of concern (sodium, saturated fat and sugar) and would bear a WL if R3337 is implemented in future. This remains the case despite the implementation of other NCD prevention policies such as the tax on sugary beverages and restrictions on sodium and trans-fats. This demonstrates the need for a WL system that will better inform consumers about the presence of unhealthy ingredients in their foods and incentivize reformulation. Our results also show that significant amounts of on-package marketing will be restricted or altogether prohibited by R3337. The results affirm that there is a high proportion of marketing on unhealthy products. In this regard the WL system, if implemented, could be an effective means to reduce the marketing of and on unhealthy products in particular. While instances of CDM were observed across both WL and non-WL products, the biggest impact will likely be in the restrictions of health claims. This shows that the regulation of health claims is an important complement to CDM restrictions as products regulated carry both CDM and health claims.

In addition to the benefits an FOPL system will have to consumer literacy and understanding about the presence of unhealthy ingredients [33], our study demonstrates that R3337 (when implemented) could substantially reduce on-package marketing, particularly on unhealthy products that will be required to carry a WL. Our study found a high incidence of on-packaging marketing and, in particular, statistically significant levels of regulated CDM on unhealthy products that would be required to carry a warning label. Across the sample of healthy and unhealthy products in South African grocery stores, we observed marketing in unhealthy food product categories such as beverages and confectionary. This aligns with findings in other countries and other studies, confirming that the strategies used to market these products are similar to those used elsewhere [34].

The inclusion of health claims in addition to evaluating CDM was a point of difference in our study design and we found that there is substantial use of health claims on products that are unhealthy or contain high levels of ingredients of concern. Health claims are used extensively on a range of products, typically on unhealthy products [35, 36]. Health claims have been found to influence both children’s and adults’ perceptions of a product’s healthfulness and their purchases [37, 38]. Our study demonstrates that, as is the case in many other contexts, there is extensive use of health claims on food products, including unhealthy ones [37]. Health claims have not often been included in analyses of CDM or on-package marketing, notwithstanding that health claims often influence adults and children alike into perceiving unhealthy foods as healthier than they are [25, 37, 39]. However, the results of this study indicate that the inclusion of comprehensive restrictions on health claims are both necessary and likely to be effective in reducing the use of misleading health claims, given their prevalence on unhealthy foods. Notably, health claims were prevalent and often used on the unhealthy products, emerging as even more frequent than the CDM on food packages. The effect of FOP WL system can be undermined by the presence of health claims, as consumers become confused as to which information to rely on. Given the high incidence of health claims on WL products, the complementary regulation of health claims may avoid consumer confusion and further limit the influence of on-package marketing in purchasing decisions. Addition of selective ingredients to create a potential nutrient benefit such as micronutrients does not negate the negative health impact of ingredients of concern.

We also note the increasing use of health-washing as a marketing strategy on food labels [40, 41]. This includes the use of endorsements by sports stars and associating the products with sporting events which create the impression that a product is healthier than it is to children and adults and the use of health claims on unhealthy products [25, 42]. Similarly promoting physical activity on a product’s package can be used to associate the product with healthfulness and to emphasize physical activity over healthy diet as a solution to obesity. However, current regulations only focus on the promotion of certain types of health-washing but omit the strategy of associating physical activity and health with unhealthy products. This is a potential area for expansion of the regulations.

Our results demonstrate high levels of traditional CDM strategies which can also shape purchasing behaviours. In particular, children’s choices are often influenced by CDM [43, 44]. Through “pester-power”, this influence may translate into children influencing caregivers to purchase unhealthier foods than they otherwise would [21, 45]. We found that many of the key on-package marketing strategies described in other contexts, such as Canada and United States, are present on the unhealthy products identified through the SA WL system. The two most common forms of CDM observed on products that will be subject to carry WLs were cartoons and animated characters, and depictions of children; both are forms of marketing that specifically target children [45]. Our results show that the regulation could, when implemented, target these forms of marketing on products high in ingredients of concern.

Our study also indicates that while substantial amounts of marketing are addressed by Draft R3337, there are associated forms of CDM which are omitted. Some products carried both regulated and unregulated forms of CDM which may create confusion as the products will carry both WLs and some forms of marketing. Limiting the restrictions to children in mixed-groups as opposed to children on their own and allowing for school-related CDM allow manufacturers to utilize these strategies and still engage in CDM notwithstanding the restrictions. It is also concerning that the regulations use some broad and vague wording that we were unable to identify indicators for such as the “omission of undesirable features” or “promotion or encouragement of inactive or unhealthy lifestyles”. This poses a challenge to the implementation and enforcement of the restrictions and indicates that more precise wording may be needed to ensure that the regulations is clear and that compliance (or not) is unambiguous.

The results of this study also demonstrate the potential for an FOP WL to be used to underpin other health promoting interventions that can improve the food environment. The study shows that utilizing WLs to regulate CDM allows the government to restrict substantial amounts of marketing on unhealthy foods. Based on the results of this study, we recommend the implementation of the R3337 as an important mechanism to improve the food environment and diet and that the government consider expanding and more comprehensively defining the marketing restrictions contained in R3337.

We also recommend that future research evaluate the impact of R3337 (when implemented) on both the nutritional composition of packaged foods in SA and the use of on-package marketing. This baseline can used to inform this future research and in particular support long term evaluations of the regulation.

Strengths and limitations

This study had some limitations and strengths. A particular strength of the study is its inclusion of an analysis of health claim data alongside CDM as there have been few studies which consider both forms of marketing together. In addition, this study provides the first, comprehensive analysis of on-package marketing in SA and provides a strong baseline dataset for future evaluations of the impact the FOP system will have on on-package marketing.

Some of the limitations of the study include that the data collected was not nationally representative, it was secondary data and data collection was confined to Cape Town, South Africa. However, to strengthen the representativeness of the data, it was collected from the largest, national grocery stores and may be considered a representation of the formal food supply in the country. The focus on the formal food supply was appropriate given that the focus was on packaged foods that would have NIP information. Smaller food suppliers will be exempt from the labelling system when it is implemented. For these reasons, we believe this analysis is broadly reflective of the SA food supply in 2019. In addition, the analyses of the dataset was limited to those products which had NIPs, something that is currently not compulsory in SA but would become compulsory when R3337 is implemented. We also note that the nutritional information captured from product labels may be subject to inaccuracies due to errors in labeling or discrepancies between reported and actual contents. This could impact the classification of products regarding their need for warning labels.

In instances where the wording of the regulation was vague or overly broad, we were unable to identify specific indicators for the CDM strategy concerned – this was applicable to two aspects of the regulation as outlined above. This limits the results of the study to only those aspects of marketing which were specific enough to develop indicators. However, it also indicates a weakness in the enforceability and evaluation of compliance with the regulations.

The evaluation was undertaken using the nutritional composition of packaged products sold in South African supermarkets in 2019. Because of this, it does not take into account reformulation that may occur as a result of the regulation. This enables this study to be used as a baseline assessment, to assess and evaluate future changes to the prevalence of on-package marketing to be assessed once the Draft R3337 is promulgated. An important contextual consideration in interpreting the data is that it reflects instances of marketing but did not situate this within the market share of a particular product. Consequently, the results are not a complete reflection of the impact the marketing restrictions would have. In addition, the study is limited to on-package marketing. Future research should consider other marketing techniques and avenues such as in-store marketing and television advertising.

Conclusion

The proposed marketing restrictions are likely to have a significant impact on the amounts of on-package marketing on foodstuffs, particularly UPPs high in nutrients of concern to limit. A substantial number of products will carry warning labels and be prohibited from utilizing common marketing techniques that, without the regulation, would encourage consumers to purchase these unhealthy products. In this regard, the proposed marketing restrictions will be supportive of improving diet and reducing consumption of UPPs. We recommend that R3337 be expanded to include unregulated forms of marketing identified in this study and that it be implemented as soon as practicable to address the problem on on-package marketing on UPPs in SA.