Abstract
Food Composition Databases (FCDs) are the resources providing detailed information on the nutritional composition of foods, while they are fundamental tools for multiple stakeholders. FCDs should always evolve to answer to the needs of their users. The change in the population’s dietary habits observed the last decades born the need of studying processed foods. However, while the existing generic FCDs, were not capable of imprinting the nutritional variability of processed foods, an international movement towards the development of Branded Food Composition Databases (BFCDs) is being observed. In Greece, the only FCD available contained 300 foods and its last update was conducted in 2004. Therefore, in 2019, at the Laboratory of Food Chemistry and Analysis, we launched a coordinated effort for the development of a Greek BFCD, which was named “Hellenic Food Thesaurus”, HelTH. The aim of HelTH is to provide detailed information towards the nutritional composition and the quality indicator ...
Food Composition Databases (FCDs) are the resources providing detailed information on the nutritional composition of foods, while they are fundamental tools for multiple stakeholders. FCDs should always evolve to answer to the needs of their users. The change in the population’s dietary habits observed the last decades born the need of studying processed foods. However, while the existing generic FCDs, were not capable of imprinting the nutritional variability of processed foods, an international movement towards the development of Branded Food Composition Databases (BFCDs) is being observed. In Greece, the only FCD available contained 300 foods and its last update was conducted in 2004. Therefore, in 2019, at the Laboratory of Food Chemistry and Analysis, we launched a coordinated effort for the development of a Greek BFCD, which was named “Hellenic Food Thesaurus”, HelTH. The aim of HelTH is to provide detailed information towards the nutritional composition and the quality indicators of the Greek branded food products. In its first edition, the structure of HelTH was composed by 4 files; the description, the nutrients, the claims files and the photobook. Food labels are always the data source of HelTH . Photographs of the products’ packaging are selected via the major super-market webstores. In 2020, HelTH contained ~4000 branded foods with >100 variables for their description. The aim of the current PhD thesis is to convert HelTH’s preliminary structure into a dynamic, up-to-date, and granular national BFCD. Its overall objective is to demonstrate that a Hellenic BFCD is a crucial tool for food and nutrition policy, academia, food industry, public health, legislative bodies, and consumers, to enhance public health. The specific objectives of the current PhD thesis are the following:a) Establishment of the database structure and targeted expansions (chapter 2),b) Data exploitation as a tool for food and nutrition policy (chapter 3),c) Data exploitation for the mapping of the Greek branded foodscape (chapter 4),d) Data exploitation to provide food-level nutritional guidance (chapter 5), ande) Dataexploitationforthefacilitationofsuitablefoodchoicesandthepromotionof personalised nutrition through digital applications (chapter 5).Chapter 2For the evolvement of HelTH’s preliminary structure into a dynamic, up-to-date, and granular BFCD, a series of modifications and additions in the preliminary structure, as well as targeted expansions to specific subcategories, were considered crucial. STUDY I. The Expansion of the Hellenic Food Thesaurus; Allergens Labelling and Allergens-Free Claims on Greek Branded Food Products. Recently, allergies and/or intolerances have been recognised as an emerging concern for various stakeholders. The primary aim of the study was the expansion of HelTH, to include allergen information. Its additional objectives were to document the total amount of allergens in Greece's branded food market, whether they are present as ingredients or indications, to map the allergen-free claims made on-pack, and to compare the nutritional content of foods carrying allergen-free claims with this of their equivalents. HelTH's structure was expanded with a new file, and data were carefully selected to include label information. One or more allergens could be found in the ingredient list of 68.4% of products and the warning statement of 38.9% of products. In the ingredient list, milk (38.8%), gluten (32.7%), and soybeans (17.4%) were most frequently stated; in precautionary statements, nuts (18.3%), eggs (13.1%), and milk (12.2%) were most frequently declared. Only 5.3% of the products had allergen-free claims, most of which included gluten and milk. The dietary composition of products that were claimed to be allergen-free and their counterparts did not, in general, differ from one another. This research presents an extended BFCD with organised and thorough allergen information. In addition, it yields fresh insights on the occurrence of food allergens in branded foods, as well as concerns about allergen declaration that need to be addressed to enhance label information. Related to HelTH’s structure, 5 more variables were added (FVNLO%, tMDP, sMDP, NOVAname, NOVAingr.), while 3 were modified (product’s name, brand, manufacturer). Related to the number of HelTH’s products, two targeted expansions were conducted; the first was targeted to branded pulse products (n=333) from 27 online retailers and the second one targeted to plant-based imitations. Chapter 3 Regarding HelTH’s exploitation as a tool for food and nutrition policy, data were used for the application of Nutri-Score and the NOVA Classification System for the extraction of valuable outcomes related to the nutritional quality and the degree of processing of branded foods, as well as related to the suitability of these systems for Greece. STUDY II. Performance and discriminatory capacity of Nutri-Score in branded foods in Greece. In Europe, concerns over food and nutrition policy revolve around the uniformity of front-of-pack nutritional declarations. Several countries have suggested the Nutri- Score system as a viable option, but its applicability for use throughout Europe is currently being investigated. The objective of the current study was to assess Nutri- Score's performance and discriminatory ability in Greece and determine whether or not it aligns with the country's food-based dietary recommendations. Both a continuous variable (FSAm-NPS Score) and a categorical variable (Grades (A)–(E)) were derived from the Nutri-Score algorithm. There were notable variations among the different food categories (p < 0.001), with the median FSAm-NPS score for all categories being 10. While the majority of liquids (59.2%) were scored (E), more than half of the solid foods (53.8%) were graded from (A) to (C). Cheeses were primarily assessed with (D), while a large percentage of milk and yoghurt products were graded with (A) or (B). Seafood and eggs received lower scores than livestock products. The primary grades for fine bakery goods were (D) and (E). There was at least one product in each Nutri-Score grade in eight of the thirteen food categories, indicating a strong discriminating ability. The algorithm demonstrates good conformity with the national recommendations by evaluating vegetables, legumes, and low-fat dairy products positively and sweets and processed meats unfavourably. Nutri-Score demonstrated a strong ability to guide consumers towards making healthier dietary decisions, in accordance with national recommendations. It also identified areas for improvement in the food supply and demonstrated the potential to steer manufacturers and consumers towards less energy-dense and more nutrient-dense choices. STUDY III. Precision Food Composition Data as a Tool to Decipher the Riddle of Ultra-Processed Foods and Nutritional QualityUltra-processed foods (UPFs), according to the NOVA Classification, are industrially formulated products characterized by the presence of food additives and suboptimal nutrient profiles. The consumption of UPFs has been linked with various non- communicable diseases (NCDs). However, BFCDs, a prerequisite for having access to products’ ingredient lists, have been rarely employed in UPFs’ research. The purpose of this study was to determine whether having access to branded food data may affect: (i) UPF identification; and (ii) the relationship between composition and processing. HelTH's availability of food (n = 4851) was categorised as NOVA1-4 either with or without the use of ingredient lists (using the generic and branded approaches, respectively). More than half of the items were categorised as UPFS in both methods. Nevertheless, UPF identification was boosted by 30% when the ingredients list was used. While almost all meat products were categorized as UPFs using branded food data, the classification of eggs and ready meals remained unaffected. Using ingredient lists, almost 30% of foods that were previously thought to be minimally processed (NOVA1-plain dairy, frozen vegetables, etc.) were instead categorized as UPFs. However, compared to the products already present in the UPFs’ group, these reclassified foods had reduced contents of calories, fat, sugar, and sodium (p < 0.001). The Nutri-Score performance of UPFs discovered with access to branded food data was improved, and they demonstrated a range of nutritional compositions appropriate for all Nutri-Score grades [A (~10%) - E (~20%)]. When NOVA is used with the relevant branded food composition database, the presumption that all UPFs have the same negative nutritional composition is questioned. Chapter 3. HelTH can also be used for the detailed description of the braded foodscape in Greece. Particularly, the granular and up-to-date character of this BFCD allows to fully describe specialized food groups available in the marketplace, such as the biological/organic foods, products bearing claims towards naturality, and plant-based imitations. STUDY IV. Nutritional Composition of Natural and Organic Branded Food Products: A Cross-Sectional Analysis of the Greek Foodscape Searching for healthier options, modern consumers resort to products promoted as "natural/organic." There is, however, little data connecting the claims made on pack with better nutritional composition. The objective of the current investigation was to determine the frequency of packaged products sold with a "natural/organic" claim and compare their nutritional content with standard equivalents that were matched by food group. In total, 22.2% of all foods made a claim about being natural, 4% claimed to be biological or to include no genetically modified organisms. The most often used claim was "free from" (12.3%), which was followed by "fresh" (4.6%), "natural/pure" (9.1%), "bio/organic" (3.3%), and "GMO-free" (1.2%). There were variations in the nutritional composition of natural/organic foods and conventional foods in just five of the thirteen food categories. For prepared foods and yogurts, being classified as natural or organic was associated with superior nutritional composition; however, there was a mixed effect for breakfast cereal, with reduced carbohydrate content and higher calorie and fat content. Organic pasta had less protein, whereas fine bakery wares marked as "free-from" had less total sugar. Overall, there was very little evidence that being branded as natural or organic was associated with having a better nutritional composition. STUDY V. Nutritional Quality of Plant-Based Meat and Dairy Imitation Products and Comparison with Animal-Based Counterparts. While consumers are increasingly adopting plant-based meat and dairy imitation products, the nutritional quality and adequacy of those foods to act as a substitute is still under discussion. This study aimed to the expansion of HelTH to map currently available meat and dairy imitations in Greece, to fully describe them in terms of their main ingredient used, nutritional composition, and quality indicators, and also to compare their nutritional composition and quality with their animal-based counterparts. A total of n = 421 plant-based imitations were analyzed; meat imitations were made primarily of wheat (38%), pulses (37%) or soy (25%), while dairy imitations were made primarily of nuts (34%), grain (19.8%) or vegetable oil (21%). The majority of plant-based imitations carried an extensive on-pack communication that included nutrition claims (65%), other claims such as vegan (62%), allergen-free (40%), bio/organic (37%), naturality (25%). Imitation sausages had higher protein and lower total and saturated fat content than their animal-based equivalents. Milk, and yogurt products had lower total fat and saturated fat content compared to their animal-based counterparts, while plant-bases cheeses were lower in protein and higher in saturated fat. Plant-based meat imitations were graded favourably by the Nutri-Score algorithm compared to their animal-based counterparts. In contrast, the nutritional quality of imitation cheeses was graded as D–E, under the Nutri-Score system, compared to A– C for the animal-based cheese. Plant-based imitations have variable composition based on their main ingredient, and the substitution of specific food groups with plant-based alternatives may not support an equivalent or improved diet. Chapter 4HelTH can also be exploited as the means to provide food-based nutritional guidance. HelTH’s data were used to test the alignment of Nutri-Score and NOVA Systems with the Mediterranean Diet (MD) recommendations, with the aim to investigate another dimension of the suitability of these systems for Greece and the provision of the nutritional guidance to enhance public health. STUDY VI. Alignment of Nutri-Score with Mediterranean Diet Pyramid: A Food Level Analysis MD is being promoted worldwide as a healthy dietary pattern. Testing the alignment of Nutri-Score with the MD Pyramid is a key step in ensuring that future food-level policies will not be conflicting with existing dietary guidelines. The study’s objectives were the classification of foods as eligible or not for inclusion in the MD and the comparison between their Nutri-Score grade with the MD tier they were ranked to. Two MD Pyramid schemes were used; the 1995 traditional and the 2020 sustainable. Only 25% of all foods were eligible under the traditional MD, while the sustainable MD covered ~58% of all foods. In both scenarios, foods that were ranked at the bottom tiers of the MD Pyramids, that are proposed for daily consumption, were commonly graded as A by the Nutri-Score algorithm. In contrast, at the top tiers of the Pyramids, that foods are recommended to be consumed rarely, the most common Nutri-Score grades were D or E. This fact suggests a good alignment between MD and Nutri-Score. Good discriminatory capacity was also seen within each tier. STUDY VII. Ultra-Processed Foods in the Mediterranean Diet according to the NOVA Classification System; A Food Level Analysis of Branded Foods in Greece. While the MD is championed as a healthy and sustainable dietary pattern, the NOVA classification is discussed as a tool to identify ultra-processed foods and further specify healthy food choices. This study aimed to test the alignment of the NOVA Classification System with the MD recommendations as presented in the MD pyramids. NOVA identified 70% of all foods as UPFs, and 59% or 41% of foods included in the sustainable and the traditional MD, respectively. In both pyramids, foods from all the NOVA groups were identified, while in nearly all the MD Pyramid tiers, foods from at least two NOVA groups were identified. In both pyramids, all processed meat products, and sweets, which are ranked at the top tiers of the pyramids , were classified as UPFs. However, NOVA identified > 50% of foods in the MD base as UPFs, too. Only 22–39% of foods in the MD base were not UPFs and of high nutritional quality (Nutri-Score A-B). Only approximately 30% of the products now available at supermarket shelves and less than 60% of the foods within the suggested MD tiers are not classified as UPFs by NOVA. NOVA has low discriminatory capacity across the MD tiers, and it restricts food choices to <30% of foods currently available in supermarkets and <60% within the recommended MD tiers. NOVA has low discriminatory capacity across the MD tiers and cannot always identify suitable food choices in line with the MD recommendations in the modern packaged food environment. Chapter 5In addition, HelTH’s exploitation can facilitate consumer’s choices and promote personalized nutrition through the development of digital applications. HelTH’s data were used to test the associations between the degree of food processing and the presence of allergens, in an effort to facilitate the food choices of people living with allergens in the modern food environment. STUDY VIII. Food Allergens in Ultra-Processed Foods According to the NOVA Classification System: A Greek Branded Food Level Analysis. UPFs’ consumption has been positively linked to the presence of allergic symptoms, but it is yet unknown whether this is linked to their nutritional composition or allergen load. This study aimed to estimate the prevalence of allergens among different NOVA Groups per food subcategory, to investigate the variability of allergens in each NOVA Group per food subcategory, and test whether processing, as assessed by the NOVA Classification System, is associated with the introduction of allergens as an ingredient or as a contaminant per food subcategory. Overall, 72% of branded foods declared the presence of at least one allergen, while 71.4% were classified as UPFS. Differences in the allergens’ prevalence was identified only in 4 out of 41 food subcategories (rice, non-chocolate confectionary, juice, condiments). Even though NOVA4 foods were more likely to contain allergens than unprocessed foods, the products’ recipe/matrix was more strongly linked to allergens’ presence with NOVA4 foods declaring 1.3 allergenic ingredients vs. 0.4 allergenic ingredients in NOVA1 foods (p < 0.01). Exposure to trace allergens was more common for NOVA4 than NOVA1 foods (45.4% vs. 28.7%), but the extent of contamination was similar (2.3 vs. 2.8 trace allergens). Overall, UPFs are more complex mixtures with higher numbers of allergens per food and are more prone to cross-contamination. Although the degree of processing was not considered a sufficient indicator for the identification of allergen-free choices within the same subcategory, HelTH has all the specifications to promote personalized nutrition. As demonstrated by the collaborations already conducted, HelTH can be used as the data source for the development of digital applications. Therefore, the development of a mobile app specialized for HelTH would be a milestone in the promotion of personalized nutrition in on a mass level Greece, it would augment its impact and also would promote nutrition and health research, by collecting app’s consumption data. In conclusion, it is worth highlighting that no one from the studies presented herein would be implemented without access to branded food data and thus, to emphasize the importance of the development of national BFCDs. The present doctoral dissertation delivers HelTH, the first and unique Greek BFCD and validates the hypothesis that such a granular, up-to-date and dynamic infrastructure can be used a) to guide data-driven policy decisions, b) to fully describe the branded food environment in Greece in terms of products’ nutritional composition and quality, their degree of processing, allergens’ presence and quality indicators, c) to provide food- based nutritional guidance, d) to facilitate consumers’ food choices, and e) as the data source for the development of digital applications that promote personalized nutrition, with an ultimate goal that its dissemination will impact in the enhancement of public health.
show more
