Imagine that one day the European Union decided to assign a simplified label to the books we buy. A green color for books that are absolutely essential reading. A yellow color for moderately interesting books, and a red color for books of poor literary quality. The assignment of these colors would be based on criteria previously defined by a group of literary experts and made publicly available. The use of this algorithm and the mention of this label and its respective color would not be mandatory, but publishers or authors who wished to could mention it. The great advantage of using this nomenclature would be to facilitate the choice at the time of purchase, preventing many poor-quality books from being bought (In fact, we buy many books based on the aesthetics of the cover, on what a literary critic tells us whose "algorithm" we don't know, or on the mention of some award whose criteria we also don't know). On average, our bookshelves would be filled with higher-quality books, and the books marked in red would tend to reduce their sales, especially since bookstores that value themselves would not put them up for sale. On the other hand, publishers and writers would be tempted to identify what makes a book worthy of bearing a green mark and would try to create works that followed the established algorithm, realigning their production according to the criteria previously publicized for the algorithm's construction, such as creativity/imagination/innovation, or the coherence and cohesion of the text, or even adherence to the characteristics of the genre in question. In this sense, new or old works should include these ingredients in the appropriate doses so that the book, submitted to the algorithmic test, could bear the mention of "absolutely worth reading" and have its green dot on the cover. In the long term, most books for sale would have the green mark or simply no symbol at all, as some authors or publishers would not adhere to this voluntary scheme for improving "literary quality".
We now leave the fantastical world of literature (where reading a bad book doesn't, in principle, cause serious health problems) for the real world of "supermarket food," where up to 18,000 food products are piled up to be chosen and purchased on any given weekend, in a short span of one to two hours. Inadequate diet or inadequate choices each weekend are the main determinants of the years of healthy life lost by the Portuguese. In this very short fraction of time, the quality of life we want to have in the medium term is decided. Excess salt, sugar, and saturated fat, a lack of dietary fiber, or the absence of fruit, legumes, or vegetables are examples of consumption patterns that can "kill us prematurely" or, conversely, can give us more years of healthy life or greater resilience to disease.
And how do we choose? How do we decide to buy the most nutritionally appropriate food product? Reading labels has been unanimously pointed out by experts as a good way for citizens to learn about the composition of foods and make more informed choices (we know, however, that price, convenience, taste… should also be considered in this equation that conditions food choices). Despite the importance of reading labels, when we analyze the nutritional information, mandatory at the European level and currently present on all packaged foods, we easily conclude that it is very difficult for the average citizen who buys large quantities of food in a short time to read and understand. Furthermore, it requires a high level of mathematical literacy to convert, for example, the amount of salt per 100g of food to the weight of the portion purchased and to the individual daily limit of 5g. This calculation is almost impossible at the time of purchase and, moreover, it is a factor that contributes to health inequalities. In fact, the majority of the Portuguese population, older and with lower levels of education, has greater difficulty in understanding and using this information. In other words, the population that already has greater difficulty accessing healthcare, that generally has a higher proportion of chronic disease, and that has to spend proportionally more on illness and food, is also the one with the greatest difficulty in understanding key information for preventing disease or halting its progression.
To overcome this difficulty and create a simplified nutritional labeling system that everyone can easily understand, regardless of their social status, a nutritional algorithm has been proposed and is already being used in several countries such as France, Belgium, and more recently Spain. When applied to packaged food products, this algorithm assigns a color ranging from green (for food products considered very nutritionally interesting) to red (for food products considered not nutritionally interesting). In reality, it is a rectangular logo placed on the front of the packaging, divided into five colors (green, light green, yellow, orange, and red), linked in turn to the letters A to E. This scale aims to briefly describe the nutritional quality of foods and, at the same time, encourage easy and quick interpretation by consumers. Multinational companies such as Nestlé, Danone, Auschan, Leclerc, Intermarché, Bonduelle, and McCain support this information model and are already labeling and advertising the foods they produce with these symbols. According to officials, in the medium term, most food products sold by these companies will be reformulated and will present significant nutritional improvements. From biscuits to breakfast cereals and even chocolate products. Many packaged food products may eventually have a green or similar label. A large part of the processed products sold by these multinationals will be easily reformulated – simply by sparingly removing or adding 3 or 4 nutrients, and they will sport a label that will certainly not be red. Other companies or producers who do not wish to participate, especially since they may sell fresh products without the right to labeling or others with limited reformulation capacity, should not want to join this voluntary scheme to improve nutritional quality through these good practices of "food literacy".
The process is underway, and in the near future, supermarket shelves may increasingly feature packaged food products with color-coded labels, making everyday consumer choices simpler, regardless of social status. This will be one of the benefits of this information model. Most likely, it will increase the availability of packaged foods with lower amounts of salt and sugar and higher fiber and protein content. However, if these processes are not carefully monitored, it could give the green light to the consumption of many food products that do not fit the nutritional profile model recommended by nutritionists, as we will see below:
1. The current nutritional labeling system, which has been intensively studied in the Portuguese population in recent years thanks to the efforts of the Directorate-General of Health and the National Program for the Promotion of Healthy Eating (PNPAS), does not seem to fulfill its role of helping consumers make informed decisions. In this sense, studies suggest the need to adopt simplified nutritional labeling systems, such as those already adopted in Australia, Chile, and other countries. Studies conducted for the Portuguese population do not highlight any simplified labeling scheme. All the models studied appear to perform well compared to the current system.
2. In the case of Nutriscore, which we analyze in greater detail here because it is the model suggested for adoption in the European Union , the assignment of a rating from A (green) to E (red) to a food depends on an algorithm, that is, a mathematical calculation, where the presence or absence of certain nutrients has implications for the final color assigned. Thus, by adding fiber and removing fat, it is possible to increase the value of a food, even if it contains considerable amounts of sugar. This is the case, for example, with yogurts and breakfast cereals. The same can happen with salt or other nutrients. In this context, one of the questions to consider is whether we currently have sufficient evidence to define complex nutritional profile models that allow us to globally classify a food without incurring significant risks. Do we have scientific evidence to support the idea that a food product with 29g of sugar and 7g of fiber per 100g, classified as C (yellow), could be classified as B (green) after 4g of sugar is removed and 1g of fiber is added per 100g? This aspect certainly deserves reflection from professionals.
3. Another issue closely related to the previous point concerns the impact of this labeling model on consumer perception, regardless of any food education campaigns that may be carried out. In other words, the process will improve the overall quality of many packaged products, but it may give a misleading indication, since a food classified with a green color will be perceived by consumers as a healthy food, encouraging its consumption. For example, for some food categories, the line separating a C (yellow) product from a B (green) product may be very thin in terms of the product's nutritional quality, while the impact on consumer perception could certainly be significant. In this respect, a risk-benefit analysis would be interesting.
4. Unfortunately, these flaws identified in the Nutriscore algorithm are more common in the yogurt and breakfast cereal categories . These categories are primarily aimed at children and often have their products labeled and advertised with nutritional claims. Evidence has shown that parents of young children are particularly susceptible to nutritional claims on food labels or in advertising, as they are generally motivated to ensure a healthy diet for their children.
5. It is also important to reflect on the reasons that have led large food companies to adopt this model, after some initial resistance. Certainly, it is a matter of social responsibility, modifying the composition of many foods for the better, which is great. On the other hand, the process will allow, with small modifications in the composition of 3 or 4 ingredients, to rewrite the nutritional history of the food and the public perception of it, transforming products previously perceived as nutritionally uninteresting into nutritionally interesting products. This is a subject that should deserve the attention of nutritionists. We will have minimally processed cereals like oat flakes with the same classification as highly industrially processed cereals with intense manipulation of their ingredients and perhaps with a completely different nutritional profile (with about 24 times more sugar, for example) so that they have an equally positive result. That is, chocolate-flavored cereals with the same classification as oat flakes. This creates a great opportunity for the marketing strategies of these companies. The experience of Australia and New Zealand , which have already implemented simplified nutritional labeling models that have some similarities to Nutriscore, has shown precisely this. These models have frequently called into question dietary recommendations and can promote the indiscriminate marketing of foods that may not fit into a healthy eating pattern. And with that comes another major challenge for those who provide nutritional education and for educators in general, and one that deserves attention.
6. An analysis we conducted on December 17, 2020, concerning a sample of 91 food products available on the Portuguese market already labeled with Nutriscore, shows that 57% are classified with the letter A or B (“green”). If we look specifically at the two most critical categories, where we consider Nutriscore to be less efficient, this value rises to 80% and 87%, for yogurts and breakfast cereals, respectively. For breakfast cereals, food products labeled A or B have an average sugar content of 17.3g (min=9g and max=27g). And for yogurts, products labeled A and B have an average sugar content of 7.1g (min=0.7g and max=13g).
7. These reflections become even more relevant when we consider that, in parallel with the discussion on the adoption of Nutriscore within the EU, there is also discussion about the possibility of establishing a common nutritional profile for the regulation of food advertising aimed at children. A nutritional profile model based on the Nutriscore algorithm could reveal many food products whose advertising is currently prohibited following the nutritional profile criteria of the World Health Organization.
labeling is a public health measure that can represent very important gains for the literacy and health of populations , and we have always been in favor of its implementation in Portugal (including the Nutriscore option). However, these tools require improvements in the algorithm and constant reflection on the part of nutritionists, so that one of the basic principles of implementing any public health measure can be guaranteed from the outset – “first do no harm” . Our value as professionals also depends on our ability to contribute to an in-depth discussion based on quality scientific evidence, on these and other topics.
PS The authors declare no conflict of interest in this matter. Maria João Gregório is the Director and Pedro Graça is a consultant for the National Program for the Promotion of Healthy Eating at the Directorate-General of Health.
Written by
Pedro Graça, Director of the Faculty of Nutrition and Food Sciences at the University of Porto
