**Dietary Fibre Consensus from the International Carbohydrate Quality Consortium (ICQC)**

**Livia S. A. Augustin 1,\*, Anne-Marie Aas 2,3 , Arnie Astrup <sup>4</sup> , Fiona S. Atkinson 5,6 , Sara Baer-Sinnott 7, Alan W. Barclay 8, Jennie C. Brand-Miller 5,6 , Furio Brighenti <sup>9</sup> , Monica Bullo 10,11,12 , Anette E. Buyken 13, Antonio Ceriello 14, Peter R. Ellis <sup>15</sup> , Marie-Ann Ha <sup>16</sup> , Jeyakumar C. Henry <sup>17</sup> , Cyril W. C. Kendall 18,19,20, Carlo La Vecchia <sup>21</sup> , Simin Liu <sup>22</sup> , Geo**ff**rey Livesey <sup>23</sup> , Andrea Poli 24, Jordi Salas-Salvadó 10,11 , Gabriele Riccardi 25, Ulf Riserus 26, Salwa W. Rizkalla 27, John L. Sievenpiper 18,19,28,29, Antonia Trichopoulou 30, Kathy Usic 31, Thomas M. S. Wolever 18,19,28, Walter C. Willett <sup>32</sup> and David J. A. Jenkins 18,19,28,29**


Received: 17 July 2020; Accepted: 19 August 2020; Published: 24 August 2020

**Abstract:** Dietary fibre is a generic term describing non-absorbed plant carbohydrates and small amounts of associated non-carbohydrate components. The main contributors of fibre to the diet are the cell walls of plant tissues, which are supramolecular polymer networks containing variable proportions of cellulose, hemicelluloses, pectic substances, and non-carbohydrate components, such as lignin. Other contributors of fibre are the intracellular storage oligosaccharides, such as fructans. A distinction needs to be made between intrinsic sources of dietary fibre and purified forms of fibre, given that the three-dimensional matrix of the plant cell wall confers benefits beyond fibre isolates. Movement through the digestive tract modifies the cell wall structure and may affect the interactions with the colonic microbes (e.g., small intestinally non-absorbed carbohydrates are broken down by bacteria to short-chain fatty acids, absorbed by colonocytes). These aspects, combined with the fibre associated components (e.g., micronutrients, polyphenols, phytosterols, and phytoestrogens), may contribute to the health outcomes seen with the consumption of dietary fibre. Therefore, where possible, processing should minimise the degradation of the plant cell wall structures to preserve some of its benefits. Food labelling should include dietary fibre values and distinguish between intrinsic and added fibre. Labelling may also help achieve the recommended intake of 14 g/1000 kcal/day.

**Keywords:** dietary fibre; labelling; carbohydrate quality; ICQC; consensus

#### **1. Introduction**

Conceptually, dietary fibre is a generic term describing non-absorbed plant carbohydrates and relatively small amounts of associated non-carbohydrate components (e.g., phenolic compounds, waxes, and proteins) that are not digested by endogenous enzymes or absorbed in the human small intestine [1,2]. Some forms of dietary fibre are digested by intestinal bacterial enzymes and utilised as substrates for growth and metabolism. The main contributors of fibre to the diet are the cell walls of plant tissues, which are supramolecular polymer networks containing variable proportions of cellulose, hemicelluloses, pectic substances, and the non-carbohydrate components, such as the phenolic compound lignin (Figure 1). Other sources of fibre in the diet include fructans (e.g., inulins), which are not part of the plant cell walls but are synthesised and stored in the cell vacuole [3,4].

**Figure 1.** Carbohydrate components of a primary plant cell wall. A cartoon of the carbohydrate components of a primary plant cell wall demonstrating the supramolecular nature of the wall and the diversity of the cell wall constituents which contribute to dietary fibre. The cellulose microfibrils are composed of crystallites which are further composed of cellulose chains. The cellulose microfibrils are stacked upon one another to give strength as the skeleton of the wall. Hemicellulose is thought to keep the microfibrils apart. The nature of hemicellulose present varies considerably between plants. Pectin is a mega molecule, used for water transport throughout the plant. There are various different sections within pectin, the proportions vary between plants. The egg box region is shown here where different strands of pectin are bound together by calcium. There is a high concentration of pectins in the middle lamellae which interact with the neighbouring cell walls [4].

However, there is much that is not known about dietary fibre, in part because the structure of the plant cell wall, which makes up the majority of our dietary fibre, has not been fully elucidated. In turn, the overall structure of the polymers, and how they interact with each other within the plant cell wall, is not yet fully understood [3,4]. Added to this, what occurs to the matrix of the cell wall during chewing (Figure 2) and movement through the digestive tract is not clear [5], and a substantial percentage of dietary fibre is digested by the microbes in the colon. The nature and actions of the microbiome are just beginning to be explored [6].

**Figure 2.** Surface of an almond seed post-mastication showing ruptured cell walls (dietary fibre). Micrograph, produced by scanning electron microscopy, of the surface of a masticated particle of almond seed. The cell walls (dietary fibre) have been ruptured (as marked by arrows) by chewing, exposing the nutrients inside the cells of the almond cotyledon tissue. Many of these cells still contain protein and lipid (oil bodies and coalesced oil droplets), which are potentially available for digestion (i.e., bioaccessible). Nutrients in intact cells below the fractured surface are not bioaccessible because the dietary fibre acts as a physical barrier to digestion. The scale bar is 30 μm.

#### **2. Definitions**

There is still disagreement about the definition of dietary fibre and how this very complex array of plant materials should be analysed. Current definitions are typically based around descriptions provided by national and international bodies for food standards, such as CODEX Alimentarius, and have focused on fibre being the non-digested and/or non-absorbed fraction of food carbohydrates derived from plants [7–11]. Dietary fibre definitions around the world have been summarised (10), and countries adopting the CODEX definition include Australia, Canada, China, the European Union, Malaysia, New Zealand, and the USA, among others. The US Food and Drug Administration issued a position paper in 2018 on what constitutes dietary fibre for food labelling purposes [11].

It may be useful to distinguish between dietary fibre, as plant cell walls (the main source of fibre) that are part of the plant food matrix, and fibre supplements that are added to food products for a specific physiological/health outcome (e.g., laxation, cholesterol-lowering, and prebiotic activity) [5]. The term natural fibre may be better described as dietary fibre that is intrinsically part of the cell wall material in edible plants such as fruits, vegetables, cereals, nuts, pulses, and even seaweed in some diets (from now on defined as intrinsic fibre). The intrinsic fibre may be modified when processed commercially and/or domestically and may not have the same physiological and metabolic effects of the original intrinsic fibre. These include the purified fibres derived from cereals (e.g., mixed-linkage β-glucans from barley and oats, among others). Some commercially available types of fibre are semi-synthetic, such as hydroxypropyl methylcellulose, which is a chemically modified cellulose. These may also be called novel types of dietary fibre in certain countries (e.g., Canada).

Another distinction seen in the literature is insoluble versus soluble fractions of fibre, which are classified by chemical analysis but not based on their functional behaviour in vivo [5]. These fractions are based on early attempts to classify fibre according to their dissolution properties in aqueous media in the laboratory. There are different chemical methods used for determining dietary fibre (e.g., the gravimetric AOAC method and GC analysis of non-starch polysaccharides) and values do vary significantly, as do the values for 'soluble' and 'insoluble' fractions. These broad classifications continue to be used in nutrition and public health literature, despite their limited use in providing information about functional properties in the gut, and their specific effects on metabolism. Solubility and viscosity are terms often used interchangeably to describe the same type of fibre; however, a soluble fibre that dissolves in aqueous media may not be viscous. Water-soluble types of fibre have the ability to lower fasting blood cholesterol and postprandial glycaemia [12]. These metabolic effects are linked to the capacity of soluble fibre to increase digesta viscosity and slow down the digestion of starch and other macronutrients. The viscosity-enhancing property of a soluble fibre is highly dependent on its polymer concentration and molecular weight, assuming it has solubilised in the gut.

#### **3. Health Benefits**

Dietary fibre can modify gastrointestinal function from the mouth to the anus. The specific physiological effects depend, crucially, on the physico-chemical properties of individual plant polysaccharides and oligosaccharides, and also on the structural integrity of fibre as cell walls, which is an important part of the architecture of the plant tissue [5]. These effects may include increasing or decreasing salivation, luminal viscosity, the gastric emptying rate, nutrient digestion and absorption, transit time, faecal bulking, laxation, fermentation, colonic pH, microbiota amount and composition, and binding of mucus, enzymes, bile acids and other metabolites, which may also be bioactive [13].

Beyond the gut, the established metabolic effects include the lowering of blood cholesterol and postprandial blood glucose, and fasting blood glucose in patients with diabetes [12]. In particular, these effects have been observed with isolated viscous fibres, such as psyllium, mixed-linkage β-glucans, guar gum (galactomannan), glucomannan, and pectic polysaccharides [14]. Another plant isolate, inulin-type fructans, though non-viscous, can lower fasting glucose and insulin and fasting LDL-cholesterol while increasing HDL-cholesterol in patients with diabetes, and to a lesser extent in overweight and obese persons [15]. A manufactured low-viscosity, digestion-resistant maltodextrin also lowers postprandial and fasting blood glucose from drinks and solid foods [16]. The molecular weight of the extracted viscous polysaccharide influences the effectiveness of the metabolic responses [9]. These observations implicate fibre as capable of modifying metabolism. Moreover, fibre-rich sources of edible plants—such as pulses, nuts, barley, oats, and some vegetables and fruits—have been shown to improve long-term control of established cardio-metabolic risk factors, i.e., blood lipids, blood glucose, blood pressure, and body weight. Many of these beneficial health effects have been attributed to the presence of fibre in these foods.

Prospective cohort studies have reported inverse associations between total dietary fibre intake and body weight, risk of type 2 diabetes, cardiovascular disease, stroke, some types of cancer, and total mortality. These associations have been shown with fibre intake from grains, legumes, nuts, fruit, and vegetables. The associations are independent of the dietary glycaemic index and glycaemic load, the effects of which are additive, at least for reducing the risk of diabetes from both observational and interventional studies [17,18]. However, despite the intensive research on nutritional epidemiology, many questions on the role of fibre in disease remain unanswered, and the contribution of associated substances to causality has been difficult to prove. Thus, the associations with fibre seen in epidemiological studies may be partially due to associated components, such as some amino acids, unsaturated fat, minerals, vitamins, and some phytochemicals, such as polyphenols, phytosterols, and phytoestrogens. In nutrition, a distinction needs to be made between intrinsic sources of dietary fibre and purified or chemically/physically modified forms of fibre, given that the three-dimensional (3D) matrix of the plant cell wall confers benefits above fibre isolates. This is because cell walls, and the 3D matrix of the plant cell walls, affect the functional properties of 'fibre' impacting on the digestibility of the cell contents [5]. This may be part of the reason for the strong benefits seen in wheat fibre in cohort studies, despite the lack of effect seen in the short-term clinical trials for cardiovascular risk factors [19–22]. In randomised controlled trials comparing refined and wholegrain cereal foods, when the particle size of the fibre is made too small, the plant cell wall integrity and tissue architecture may be lost. When tissue and the cell wall 3D matrix are sufficiently intact, it can lead to nutrients being slowly absorbed or even not absorbed. For example, cereal foods with a substantially intact tissue structure can also contribute starch as a source of a slowly and/or non-digestible food carbohydrate [23–25].

Fibre in wholefoods, isolates, and modified forms can be sources of substrate for micro-organisms in the large intestine, affecting the amount and species composition of the microbiota and their collective functional capacity to improve the health of the gut and other organs via modulation of the immune system, production of bioactive metabolites (e.g., short-chain fatty acids), and the reduction of intracolonic pH, with beneficial effects on the colonic mucosa and blood lipid levels [26].

At the population level, we suggest replacing some animal foods, and high glycaemic index foods containing refined starches and sugars, with slowly digestible carbohydrate foods with a low glycaemic index that are rich in fibre. This would have a favourable impact on glycaemic control and, hence, diabetes, cardio-metabolic risk, and possibly some diabetes-related cancers [27]. Minimising the degradation of the plant cell wall structures and tissue architecture is important where slow digestibility of macronutrients, such as starch, is required for the production of healthy foods, and also in the development of low glycaemic index foods. These issues are important, especially in some parts of the world with a high risk of cardio-metabolic disease, where dietary fibre intake tends to be below the recommended intake levels. However, it is recognised that in foods where mineral bioavailability needs to be increased, the rupture of the cell walls may provide a way to improve mineral status, e.g., a higher iron bioavailability through the micro-milling of wheat aleurone [28].

Much research is still required to fully understand the physiological and nutritional effects of dietary fibre. We need to further understand the interaction of fibre with the microbiota, and we also need to understand more about the structure, physico-chemical properties, and composition of dietary fibre. Additionally, we require an improved mechanistic insight into how the components associated with dietary fibre interact with fibre, and the impact on metabolic outcomes. Furthermore, an improved understanding is required on the role played by the 3D architecture of dietary fibre on nutrient release (i.e., bioaccessibility), fermentability by gut bacteria, prebiotic activity, and the roles these have in human health. When these are better elucidated, there will be a need to communicate to food producers, consumers, and health professionals on how to make better food choices [5].

Certain types of dietary fibre affect the amount and composition of microbiota, which has been studied mostly in regard to fermentative micro-organisms in the large intestine. Inulins, found in plants like chicory root and galacto-oligosaccharides, present in or from milk, are prime examples of non-digestible carbohydrate or dietary fibres that, among others, behave as prebiotics [29–31]. A prebiotic was recently defined by consensus as "a substrate that is selectively utilised by host micro-organisms conferring a health benefit" [32]. Putative health benefits include the inhibition of pathogens reaching the large intestine, immune stimulation, improved cardiometabolic status, improved mental health, and support to bone mineralisation, among others [32]. More long-term randomized controlled trials are needed to establish causality, which appears promising, though prebiotic effects may not be seen in everyone, especially in persons already in good health or having a sufficient amount and composition of beneficial micro-organisms. Moreover, not all dietary fibres are prebiotic, but the effect prebiotic fibre has can depend on the amount of other dietary fibre that is consumed [33].

Many chemical/enzyme methods exist for analysing dietary fibre, but those used for labelling are often different from those used in food composition tables. Current analytical methods reflect a heterogeneous mix of chemical entities, with no information on any subspecies of fibre or any information on the structural characteristics of the fibre present. One example of how dietary fibre is measured is by using the AOAC enzyme-gravimetric method, which is intended to simulate the physiological conditions of digestion, and measures all the components of fibre, as currently defined by CODEX Alimentarius. This kind of analysis is limited when being used to interpret mechanistic data on the functional properties of cell walls, individual cell wall polysaccharides and storage oligosaccharides. More informative methods, notably dissolution kinetics, molecular weight of individual polysaccharides, and cell wall porosity are urgently required for characterising dietary fibre in nutritional and epidemiological studies, if food sources of dietary fibre for health are to be optimised.

#### **4. Recommendations to the Public and to Health Professionals**

It is generally agreed that dietary fibre is an important part of a sustainable, balanced, healthy diet [34]. Consumption of dietary fibre is below the recommended intake levels for optimal health in many parts of the world and may be decreasing. We recommend maintaining or increasing dietary fibre intake to the recommended levels.

We support the Institute of Medicine recommendations for the total dietary fibre of 14 g/1000 kcal/day. We suggest that this should mainly come from intrinsic dietary fibre. Data from cohort studies with intakes beyond this amount are limited, but many traditional societies consume larger amounts and have a lower risk of chronic diseases.

#### **5. Recommendations to the Food Industry**

The food industry plays an important role in developing new food ingredients and products that have public health benefits and are also highly palatable. In developing new high-fibre foods, the sensory characteristics are important and will strongly influence whether people consume them. At the same time, if these do not have nutritional benefits then such products would be of little nutritional value, regardless of how technologically innovative they may be. It is important to recognise that increasing the fibre content on the food label does not guarantee any enhanced nutritional benefits in a product.

Recommendations to the food industry would depend on the reasons why particular types of fibre are being added, and how they are processed, given that mechanical and hydrothermal processing can affect their properties. For example, in wheat grain there is an advantage in preserving some of the structural integrity of the cell walls of the starch-rich endosperm, in order to produce flour that is digested more slowly and has a beneficial impact on postprandial glycaemia (23). However, if the health outcome is to improve the iron bioavailability in wheat, then there may be advantages to micro-milling (rupturing) the aleurone cell layer, which has a high iron concentration (28). In producing foods for the general population, the first example would be the most appropriate recommendation while, for populations with nutritional deficiencies, the second recommendation may preferentially apply. Therefore, we generally encourage the food industry to preserve many of the benefits of dietary fibre rich foods by minimising the degradation of the plant cell wall structures and tissue architecture, while maintaining palatability, except in situations of special dietary requirements and specific physiological or clinical outcomes (e.g., the use of prebiotic oligosaccharides and viscous polysaccharides).

Currently, labelling the dietary fibre content of foods in certain countries around the world, including Europe, is not mandatory. This represents a problem for consumers, researchers, and medical staff dealing with patient diets. We support the mandatory use of fibre on food labels.

Labelling should distinguish between fibre that is endogenous to foods and that added as a functional supplement because synthetic or purified fibre will not be accompanied by the micronutrients and phytochemicals in foods and, thus, may not predict the same health outcomes. Functional (or other) supplemental fibre, where permitted, should be listed separately among ingredients. The labelling of dietary fibres could be of the form "FIBRE N g PER 100 g, of which X g is SUPPLEMENTAL".

#### **6. Conclusions**

Dietary fibre and its associated non-carbohydrate components have been inversely associated with disease outcomes. Food labelling should include dietary fibre, and distinguish between intrinsic and purified added fibre, given that the intact plant cell wall may confer benefits beyond fibre isolates. The labelling of dietary fibre may also help to achieve the recommended intake of 14 g/1000 kcal/day for health benefits. To extend these recommendations, further studies on the interrelation of dietary fibre, prebiotics, and health, which aim to optimise both the health potential of foods and related food processing methods, are advised. This would include how the structures and the 3D matrix, composition, and physico-chemical properties of dietary fibre affect digestion, gastrointestinal function, and the role of the microbiome.

**Author Contributions:** All authors have made contributions to the statements and various drafts and read and approved the final manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** No funding was received for this consensus statement. The dietary fibre consensus meeting was held as part of the 4th International Carbohydrate Quality Consortium (ICQC) Meeting, Palinuro, Italy, Sept 12–13, 2019, which was funded through unrestricted educational grants from Abbott, Arla Foods, Barilla, Beneo Institute, General Mills, Global Pulse Confederation, Inquis Clinical Research, International Pasta Organization, Nestle' Research and Development, Pulse Canada, McCain, and Quaker. The meeting was co-organized by the Toronto 3D Knowledge Synthesis and Clinical Trials foundation, Nutrition Foundation of Italy, and the Glycemic Index Foundation.

**Conflicts of Interest:** L.S.A.A. is a founding member of the International Carbohydrate Quality Consortium (ICQC) and has received honoraria from the Nutrition Foundation of Italy (NFI), research grants from LILT (a non-profit organization for the fight against cancer) and in-kind research support from Abiogen Pharma, the Almond Board of California (USA), Barilla (Italy), Consorzio Mandorle di Avola (Italy), DietaDoc (Italy), Ello Frutta (Italy), Panificio Giacomo Luongo (Italy), Perrotta (Italy), Roberto Alimentare (Italy), SunRice (Australia). A.A. is a project director at the Novo Nordisk Foundation, responsible for prevention of childhood obesity. He is in the Scientific Advisory Board/Consultant/Board of Directors of Gelesis, USA; Ferrero, Italy; Groupe Éthique et Santé, France; International Egg Commission/Danske Æg, Denmark; McCain Foods Limited, USA; Novo Nordisk, Denmark; Rituals, USA; and Weight Watchers, USA. A.W.B. is consultant at the University of Sydney and is Honorary Associate of the Glycemic Index Foundation, Allied Pinnacle, Beneo, and Nestle, and has authored/co-authored 5 books about dietary carbohydrate and diabetes. J.C.B.-M. is a co-author of books about the glycemic index of foods. She is President of the GI Foundation Limited, a non-profit company that administers the Australian 'GI Symbol' program and oversees the Sydney University Glycemic Index Research Service (SUGiRS), a non-profit GI testing facility for the food industry. She has received honoraria for speaking engagements on the glycemic index of foods. F.B.: declares no ownership or other investments-including shares-in commercial activities, intellectual

property rights and consultancy/advice to private stakeholders. He served as Deputy-Chancellor for Research at the University of Parma from 2013 to 2017 and as representative of University of Parma of the Steering Board of ASTER, an in-house Public Research organization owned by Regione Emilia Romagna and research institutions of the region. Both positions involved coordination of activities with a large number of companies in order to attract European founds and develop Research and Innovation in the Emilia Romagna region. From 2011 to 2016 he served as legal representative (President) of the Italian Nutrition Society SINU–a not-for- profit scientific society. Associates to SINU include companies operating in the fields of nutrition, body composition, nutrition software, meal distribution and food production. Since 2017 he has been a Member of the Board of the Federation of European Nutrition Societies (FENS). He is the spouse of Silvia Valtuena Martinez, MD, PhD, a Scientific Officer at the NDA Unit of the European Food Safety Authority, the agency of the European Union (EU) that provides independent scientific advice and communicates on existing and emerging risks associated with the food chain. A.E.B.: Member of the ILSI Europe Expert Group 'Carbohydrates based recommendations as a basis for public dietary guidelines' (coordinated by the Dietary Carbohydrates Task Force). Member of the Executive Committee of the German Nutrition Society (Guidelines Committee 'Carbohydrate intake and prevention of nutrition-related diseases', Guidelines Committee 'Protein intake and prevention of nutrition-related diseases', Head Section "Public Health Nutrition"). A.C.: is in the advisory board for BD (Beckton Dikinson), Eli Lilly, Mundipharma; gave lectures sponsored by Astra Zeneca, Berlin Chemie, Boehringer Ingelheim, Eli Lilly, Mundipharma, Novo Nordisk and Roche Diagnostics; and received research grants from Mitsubishi. P.R.E.: some of his almond studies were funded by the Almond Board of California. M.-A.H.: Director of East Anglia Food. J.C.H.: his Clinical Nutrition Research Centre conducts food and nutrition research for several companies including Beneo, Roquette, Tate and Lyle, Wilmar and Nestle. D.J.A.J.: has received research grants from Saskatchewan & Alberta Pulse Growers Associations, the Agricultural Bioproducts Innovation Program through the Pulse Research Network, the Advanced Foods and Material Network, Loblaw Companies Ltd., Unilever Canada and Netherlands, Barilla, the Almond Board of California, Agriculture and Agri-food Canada, Pulse Canada, Kellogg's Company, Canada, Quaker Oats, Canada, Procter & Gamble Technical Centre Ltd., Bayer Consumer Care, Springfield, NJ, Pepsi/Quaker, International Nut & Dried Fruit (INC), Soy Foods Association of North America, the Coca-Cola Company (investigator initiated, unrestricted grant), Solae, Haine Celestial, the Sanitarium Company, Orafti, the International Tree Nut Council Nutrition Research and Education Foundation, the Peanut Institute, Soy Nutrition Institute (SNI), the Canola and Flax Councils of Canada, the Calorie Control Council, the Canadian Institutes of Health Research (CIHR), the Canada Foundation for Innovation (CFI)and the Ontario Research Fund (ORF). He has received in-kind supplies for trials as a research support from the Almond board of California, Walnut Council of California, American Peanut Council, Barilla, Unilever, Unico, Primo, Loblaw Companies, Quaker (Pepsico), Pristine Gourmet, Bunge Limited, Kellogg Canada, WhiteWave Foods. He has been on the speaker's panel, served on the scientific advisory board and/or received travel support and/or honoraria from the Almond Board of California, Canadian Agriculture Policy Institute, Loblaw Companies Ltd, the Griffin Hospital (for the development of the NuVal scoring system), the Coca-Cola Company, EPICURE, Danone, Diet Quality Photo Navigation (DQPN), Better Therapeutics (FareWell), Verywell, True Health Initiative (THI), Heali AI Corp, Institute of Food Technologists (IFT), Soy Nutrition Institute (SNI), Herbalife Nutrition Institute (HNI), Saskatchewan & Alberta Pulse Growers Associations, Sanitarium Company, Orafti, the American Peanut Council, the International Tree Nut Council Nutrition Research and Education Foundation, the Peanut Institute, Herbalife International, Pacific Health Laboratories, Nutritional Fundamentals for Health (NFH), Barilla, Metagenics, Bayer Consumer Care, Unilever Canada and Netherlands, Solae, Kellogg, Quaker Oats, Procter & Gamble, Abbott Laboratories, Dean Foods, the California Strawberry Commission, Haine Celestial, PepsiCo, the Alpro Foundation, Pioneer Hi-Bred International, DuPont Nutrition and Health, Spherix Consulting and WhiteWave Foods, the Advanced Foods and Material Network, the Canola and Flax Councils of Canada, Agri-Culture and Agri-Food Canada, the Canadian Agri-Food Policy Institute, Pulse Canada, the Soy Foods Association of North America, the Nutrition Foundation of Italy (NFI), Nutra-Source Diagnostics, the McDougall Program, the Toronto Knowledge Translation Group (St. Michael's Hospital), the Canadian College of Naturopathic Medicine, The Hospital for Sick Children, the Canadian Nutrition Society (CNS), the American Society of Nutrition (ASN), Arizona State University, Paolo Sorbini Foundation and the Institute of Nutrition, Metabolism and Diabetes. He received an honorarium from the United States Department of Agriculture to present the 2013 W.O. Atwater Memorial Lecture. He received the 2013 Award for Excellence in Research from the International Nut and Dried Fruit Council. He received funding and travel support from the Canadian Society of Endocrinology and Metabolism to produce mini cases for the Canadian Diabetes Association (CDA). He is a member of the International Carbohydrate Quality Consortium (ICQC). His wife, Alexandra L Jenkins, is a director and partner of INQUIS Clinical Research for the Food Industry, his 2 daughters, Wendy Jenkins and Amy Jenkins, have published a vegetarian book that promotes the use of the foods described here, The Portfolio Diet for Cardiovascular Risk Reduction (Academic Press/Elsevier 2020 ISBN:978-0-12-810510-8)and his sister, Caroline Brydson, received funding through a grant from the St. Michael's Hospital Foundation to develop a cookbook for one of his studies. C.W.C.K.: has received grants/research support from Advanced Food Materials Network, Agriculture and Agri-Foods Canada (AAFC), Almond Board of California, Barilla, Canadian Institutes of Health Research (CIHR), Canola Council of Canada, International Nut and Dried Fruit Council, International Tree Nut Council Research and Education Foundation, Loblaw Brands Ltd, National Dried Fruit Trade Association, Pulse Canada, and Unilever; in-kind research support from the Almond Board of California, the American Peanut Council, Barilla, the California Walnut Commission, Danone, Kellogg Canada, Loblaw Companies, Nutrartis, Quaker (Pepsico), Primo, Unico, Unilever and Upfield; travel support/honoraria from the American Peanut Council, the International Nut and Dried Fruit Council, the International Pasta Organization, Lantmannen, Oldways Preservation Trust, and the Peanut Institute. He has

served on the scientific advisory board for the International Pasta Organization, McCormick Science Institute, Oldways Preservation Trust. He is a member of the International Carbohydrate Quality Consortium (ICQC), Executive Board Member of the Diabetes and Nutrition Study Group (DNSG) of the European Association for the Study of Diabetes (EASD), is on the Clinical Practice Guidelines Expert Committee for Nutrition Therapy of the EASD and is a Director of the Toronto 3D Knowledge Synthesis and Clinical Trials foundation. C.L.V.: serves on the scientific board of the ISA (International Sweeteners Association) and has received grants from Soremartec. S.L.: has received consulting payments and honoraria for scientific presentations or reviews at numerous venues, including but not limited to Barilla, Johns Hopkins University, Fred Hutchinson Cancer Center, Harvard University, University of Buffalo, Guang Dong General Hospital and Academy of Medical Sciences, and the National Institutes of Health. He is also a member of the Data Safety and Monitoring Board for a trial of pulmonary hypertension in diabetes patients at Massachusetts General Hospital. He receives royalties from UpToDate. Liu receives an honorarium from the American Society of Nutrition for his duties as Associate Editor. G.L.: holds shares in Independent Nutrition Logic Ltd., a consultancy. He and his wife have benefitted from research grants, travel funding, consultant fees, and honoraria from the American Association for the Advancement of Science (USA), the All Party Parliamentary Group for Diabetes (London, UK), Almond Board of California (USA), BENEO GmbH (DE), Biotechnology and Biosciences Research Council (UK), British Nutrition Foundation(UK), Calorie Control Council (USA), Cantox (CA), Colloides Naturel International (FR), Coca Cola (UK), Danisco (UK & Singapore), Diabetes Nutrition Study Group (EASD, EU), DiabetesUK (UK), Elsevier Inc. (USA), European Commission (EU), European Polyol Association (Brussels), Eureka (UK), Food and Agricultural Organization (Rome), Granules India (Ind), General Mills (USA), Health Canada (CA), Institute of Food Research (UK), International Carbohydrate Quality Consortium (CA), Institute of Medicine (Washington, DC), International Life Sciences Institute (EU & USA), Life Sciences Research Office, FASEB (USA), Nutrition Society of Australia, Knights Fitness (UK), Leatherhead Food Research (UK), LitghterLife (UK), Matsutani (JPN), Medical Research Council (UK), MSL Group (UK), Porter Novelli (UK), Sudzuker (DE), Sugar Nutrition/WSRO (UK), Tate & Lyle (UK), The Food Group (USA),WeightWatchers (UK),Wiley-Blackwell (UK).World Health Organization (Geneva). He is a member of the EASD Nutrition Guidelines Committee. PA: is the President of the Nutrition Foundation of Italy (NFI) a non-profit organization partially supported by Italian and non-Italian Food Companies. J.S.-S.: serves on the board of (and receives grant support through his institution from) the International Nut and Dried Fruit Council and the Eroski Foundation. He also serves on the Executive Committee of the Instituto Danone, Spain, and on the Scientific Committee of the Danone International Institute. He has received research support from the Patrimonio Comunal Olivarero, Spain, and Borges S.A., Spain. He receives consulting fees or travel expenses from Danone, the Eroski Foundation, the Instituto Danone, Spain, and Abbot Laboratories. J.L.S.: has received research support from the Canadian Foundation for Innovation, Ontario Research Fund, Province of Ontario Ministry of Research and Innovation and Science, Canadian Institutes of health Research (CIHR), Diabetes Canada, PSI Foundation, Banting and Best Diabetes Centre (BBDC), American Society for Nutrition (ASN), INC International Nut and Dried Fruit Council Foundation, National Dried Fruit Trade Association, National Honey Board, International Life Sciences Institute (ILSI), The Tate and Lyle Nutritional Research Fund at the University of Toronto, The Glycemic Control and Cardiovascular Disease in Type 2 Diabetes Fund at the University of Toronto (a fund established by the Alberta Pulse Growers), and the Nutrition Trialists Fund at the University of Toronto (a fund established by an inaugural donation from the Calorie Control Council). He has received in-kind food donations to support a randomized controlled trial from the Almond Board of California, California Walnut Commission, American Peanut Council, Barilla, Unilever, Upfield, Unico/Primo, Loblaw Companies, Quaker, Kellogg Canada, WhiteWave Foods, and Nutrartis. He has received travel support, speaker fees and/or honoraria from Diabetes Canada, Dairy Farmers of Canada, FoodMinds LLC, International Sweeteners Association, Nestlé, Pulse Canada, Canadian Society for Endocrinology and Metabolism (CSEM), GI Foundation, Abbott, Biofortis, ASN, Northern Ontario School of Medicine, INC Nutrition Research & Education Foundation, European Food Safety Authority (EFSA), Comité Européen des Fabricants de Sucre (CEFS), and Physicians Committee for Responsible Medicine. He has or has had ad hoc consulting arrangements with Perkins Coie LLP, Tate & Lyle, Wirtschaftliche Vereinigung Zucker e.V., and Inquis Clinical Research. He is a member of the European Fruit Juice Association Scientific Expert Panel and Soy Nutrition Institute (SNI) Scientific Advisory Committee. He is on the Clinical Practice Guidelines Expert Committees of Diabetes Canada, European Association for the study of Diabetes (EASD), Canadian Cardiovascular Society (CCS), and Obesity Canada. He serves or has served as an unpaid scientific advisor for the Food, Nutrition, and Safety Program (FNSP) and the Technical Committee on Carbohydrates of ILSI North America. He is a member of the International Carbohydrate Quality Consortium (ICQC), Executive Board Member of the Diabetes and Nutrition Study Group (DNSG) of the EASD, and Director of the Toronto 3D Knowledge Synthesis and Clinical Trials foundation. His wife is an employee of AB InBev. T.M.S.W.: he and his wife are part owners and employees of INQUIS Clinical Research, Ltd. (formerly GI Labs), a contract research organization in Toronto, Canada. He has authored or co-authored several books on the glycemic index for which has received royalties from Philippa Sandall Publishing Services and CABI Publishers. He has received research support, consultant fees or honoraria from or served on the scientific advisory board for Canadian Institutes of Health Research, Canadian Diabetes Association, Dairy Farmers of Canada, Agriculture Agri-Food Canada, Public Health Agency of Canada, GI Labs, GI Testing, Abbott, Proctor and Gamble, Mars Foods, McCain Foods, Bunge, Temasek Polytechnic Singapore, Northwestern University, Royal Society of London, Glycemic Index Symbol program, CreaNutrition AG, McMaster University, University of Manitoba, University of Alberta, Canadian Society for Nutritional Sciences, National Sports and Conditioning Association, Faculty of Public Health and Nutrition and Autonomous University of Nuevo Leon, Diabetes and Nutrition Study Group of the European Association for the Study of Diabetes (EASD). All other authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **The E**ff**ect of Labelling and Visual Properties on the Acceptance of Foods Containing Insects**

**Klaudia Modlinska 1,\* , Dominika Adamczyk <sup>2</sup> , Katarzyna Goncikowska 2, Dominika Maison <sup>2</sup> and Wojciech Pisula <sup>1</sup>**


Received: 27 July 2020; Accepted: 17 August 2020; Published: 19 August 2020

**Abstract:** Introducing insects as a source of nutrients (e.g., protein) plays a key role in many countries' environmental policies. However, westerners generally reject insects as an ingredient of food products and meals. The aim of our study was to assess if explicitly labelling food as containing insects and/or implying it by manipulating the appearance of food influences the participants' perception of food products or their behavioral reaction to such products. Participants were asked to try a range of foods, none of which contained ingredients derived from insects. However, the experimental conditions varied with regard to food labelling (insect content) and appearance (traces of insect-like ingredients). We observed the participants' non-verbal behavioral reactions to the foods. Next, the respondents filled in a questionnaire evaluating the food's properties. Additionally, we asked the participants to fill in a set of questionnaires measuring other variables (food neophobia, disgust, variety seeking, etc.) The results showed that products labelled as containing insects are consumed with reluctance and in lower quantities despite their appearance. In addition, people with lower general neophobia and a higher tendency to seek variety tried the insect-labelled samples sooner than people from the other groups. Recommendations for marketing strategies are provided.

**Keywords:** food labelling; entomophagy; insect-based foods; edible insects; food sustainability; perception of food; novel food; disgust; neophobia; variety seeking; food technology neophobia; consumer studies; behavior

#### **1. Introduction**

While food shortages mainly affect developing countries, where malnutrition is a problem for millions [1], in highly developed countries the conventional methods of food production burden the environment and pose a threat to animal welfare. Introducing insect-based food products could contribute to solving both issues [2].

Entomophagy is present in many cultures and is the main source of nutritious food for many communities [3]. Insects are a source of protein (and amino acids), fats, vitamins (e.g., vitamin B12), beta-carotene, several minerals, fiber and other valuable substances [4–8]. Insect production requires less space [9] and leaves a significantly smaller ecological footprint than livestock farming, and is more ecologically sustainable [10,11]. Moreover, as insects are evolutionarily distant from humans, they are less likely to carry pathogens that could pose a risk to human health [7,12].

Although nowadays the consumption of insects by humans is regularly discussed in the media, attempts at changing people's eating habits and introducing these new types of food have so far

generally met with individual and social rejection. Although some recent studies show a positive effect of information on the willingness to try insect-based foods (e.g., [13–16]), it seems that, in general, rational arguments stressing the advantages of insect production and consumption are not sufficient to change our eating habits [17,18] (see also [19]). In general, westerners are largely opposed to eating insects, as they are an unfamiliar food source that deviates from cultural norms and are expected to have undesirable sensory properties [20] (cf. [21]).

It seems that the main psychological barriers to consuming this kind of novel food are disgust and fear [22–24]. Most people object not only to eating insects, but also to the very idea of eating insects [22,23], and their negative reactions to insects may be very deeply rooted and automatic [25]. The first explanation for this reaction is disgust. Disgust is an emotion that elicits thoughts and behaviors that result in avoiding the objects that trigger it [26]. Things that elicit disgust differ largely from one culture to another; they are acquired in the learning process and are deeply rooted in social norms. Disgust correlates with the evolutionary need to avoid infection, dirt and disease, and may lead to fear [23]. In the case of food, this fear is manifested in anxious hesitation to consume unknown items, which is called food neophobia [27]. Food neophobia correlates positively with general neophobia [27], which may increase reluctance towards unfamiliarity. Therefore, it is likely that persons with a high level of food neophobia, as well as a high level of general neophobia, would be prone to the feeling of disgust (see [28–30]). Persons who fit this profile are very likely to be the least willing to include insects in their diet [18]. Moreover, in the case of such new commercially available products as insects and insect flour, and their mostly unknown production processes, Food Technology Neophobia [31] may also play a crucial part. On the other hand, it is possible that other characteristics such as the Variety Seeking Tendency [32], which involves an intrinsic desire for variety in many aspects of life, including food consumption, may increase the willingness to try new food products [33].

Although the main psychological variables influencing the acceptance/rejection of insect-based foods seem to have been identified, the problem of a limited willingness to incorporate such foods into everyday diet has not been resolved. Over the past few decades, researchers have been trying to devise a strategy to convince people that insects are indeed edible, safe, and tasty. It may be hypothesized that one such strategy could involve adapting foods containing insects to the consumers' general eating preferences. It is clear, for instance, that consumers prefer fatty and sweet products (e.g., [34,35]). Serving insects in the form of, or as an ingredient of, snacks or sweet foods may increase their acceptance and create positive associations with such types of food, facilitating habituation to insect-based foods. On the other hand, studies by Tan et al. [36] showed that sweet-flavored insects were considered less appropriate than savory insect-based meals, especially when the entire insect body or insect body parts were easily discernible.

Another way of reducing the level of food neophobia may be to add insects to familiar dishes [14] (cf. [37]). However, the form in which they are added also plays a part. It seems that adding ground insects to a meal could reduce reluctance to consume it by reducing exposure to the visual stimulus (e.g., [20]). In the case of unconventional animal-derived foods, evoking the image of the entire animal often elicits strong objection to the food offered and is connected with strong negative emotions [38]. When eating animals, we usually consume pieces that do not resemble living individuals; we are reluctant to eat animal heads, entire limbs, etc. Perhaps in the case of insects, the fact that they are often served whole elicits similar reactions in humans by making it clear what they are about to consume. Moreover, adding entire insects to meals may give the impression that a food is polluted or rotten. Evidence to support this claim has been provided in a study by Gmuer et al. [39], which showed that potato chips which contained insect flour or insect elements were assessed less negatively than chips mixed with whole insects (see also [29,40,41]). Other studies, however, show no such correlation (e.g., [42,43]). In addition, the widely applied marketing strategy of presenting insects as sustainable substitutes for animal protein is also problematic, as it gives rise to an expectation that the products will be similar to meat in texture and taste, as is the case with plant-based products made to look similar to their meat-based counterparts [17] (cf. [30]).

Nevertheless, it seems that the reluctance to eat insects is so strong that the very awareness of consuming them elicits an aversive reaction, which may then be generalized to other accompanying products. This hypothesis seems to have been confirmed by Rozin et al. [44] in a study where participants assessed a drink more negatively if the cup they drank from had been in contact with a sterile insect before it was used. Such a reaction may be explained by the participants' feeling that the cup had been contaminated during contact with the insect, and the disgust triggered by that perception was generalized to the drink. In the case of commercially available products, information on insect content is provided in visual form on packaging or it is provided in the product name, which may have a similar effect. Studies carried out to date have shown that both verbal and visual information on packaging has a significant impact on how consumers evaluate the taste and smell of the substances they ingest (e.g., [45–48]). The very indication of insect content on the packaging may affect consumers' assessment of the product quality. Moreover, it may be assumed that ingredients whose appearance suggests the presence of traces of insects in a product may additionally reinforce the label effect by reinforcing the impression that the product has been contaminated (cf. [23]).

Based on the above, we are convinced that more research is needed to further explore the factors influencing peoples' attitude to eating insect-based foods and to identify marketing strategies and educational campaigns. We would like to propose the study protocol involving an assessment of several variables that have been identified in other studies (the results of which, however, are in some cases inconclusive) or that stem from our predictions. In our study, we intended to check, first of all, whether the mere fact of providing information about insect content (label) would influence the sensory evaluation and acceptance of different foods. In this way, we wanted to replicate the findings obtained by Mancini and colleagues [14]. However, we decided to use sweet foods (pastry, sweets), as we hypothesized that snacks would be more likely to encourage people to try this kind of food, especially bearing in mind the general preference for sweetness among consumers. Secondly, we hoped to broaden the scope of the study by evaluating the effect of visibility of insect parts on the level of acceptance of insect-based foods, as studies that have analyzed these aspects have been inconclusive. We expected that the differences in the results obtained by other researchers could have arisen from a specific "presentation" of insects, but also from an interactive effect of the information on insect content and the appearance of food. In addition, we planned to control the mediating effect of commonly studied psychological variables, such as food neophobia and disgust. However, we also incorporated other measures in the study; we used the food technology neophobia scale and the variety-seeking tendency scale to further explore the mechanism underlying the acceptance of insect-based food.

We intended to analyze these aspects by checking experimentally reactions to food products depending on (a) the information about insect content provided on product labelling, and (b) the appearance of products suggestive of traces of insect-like ingredients. To address these issues, we conducted an experiment whereby we observed participants' reactions to specific products: a cookie, a muffin, and a date ball. The experiment resembled a classic product test in which the consumers tried and assessed three products in terms of taste, smell, appearance, etc. Depending on the experimental conditions (2 × 2), the products differed either (a) with regard to the information about the presence or absence of insect content, or (b) in appearance: they are artificially "dotted" to suggest insect content or not. In reality, none of the products used in the experiment contained insects or insect-derived ingredients.

Another important novel element of our experiment, when compared to many previous studies, was the fact that we not only measured participant reactions in qualitative scale-based terms (declaration of willingness to taste a specific product), but we also observed several non-verbal behavioral indicators (e.g., we measured the time and frequency of sampling products and how much of the product was ingested). What is more, after the experiment, we conducted interviews with the participants, which served as a source of additional information to enrich the discussion section of the paper and allow us to suggest possible marketing strategies.

Based on the results of previous studies, we formulated the following hypotheses:

**Hypothesis 1 (H1).** *Information about insect-based ingredients on the product label will translate into lower product evaluation scores and will have an impact on the behavioral reactions to the product (cf. [44–48]).*

**Hypothesis 2 (H2).** *Product appearance (traces of ingredients suggesting insect content) will translate into lower product evaluation scores and will have an impact on the behavioral reactions to a product (e.g., [20,29,39,41]).*

**Hypothesis 3 (H3).** *The perception of the product and behavioral indicators will be controlled by di*ff*erent psychological features. People with higher food neophobia, sensitivity to disgust, food technology neophobia, and general neophobia are expected to be less willing to try products described as containing insect-derived ingredients (product evaluation and behavioral reactions to the products), while people with a high level of variety-seeking tendency will be more open to this kind of novel food (e.g., [22,23,32,43]).*

#### **2. Methods**

#### *2.1. Participants*

The participants were recruited by means of announcements posted online, on website panels, and individual personal requests. All participants gave their oral consent to participate in the study. They also confirmed that they did not suffer from any form of food allergy or intolerance. One person was excluded from the study after having a strong emotional reaction to the food samples and after disclosing an existing psychiatric condition. The final sample comprised 99 participants (81 females and 18 males). The participants were aged between 18 and 45 years old, while the average age was 22. They were mostly university students who came from big cities (>50,000 inhabitants).

The participants were randomly assigned to four experimental groups. In each group, the male to female ratio was comparable. They were told that the purpose of the experiment was to study food preferences. Before the experiment, they were not informed that the experiment involved tasting insect-based products, so their attitudes and expectations did not affect the answers given in the questionnaires and the participants who tasted products with no indication of insects did not feel they were being misled, or that the products they were supposed to try contained hidden insect parts. The data was collected anonymously. Each person was assigned an individual number to allow the data from all parts of the experiment to be linked.

#### *2.2. Procedure and Methods*

The study consisted of three parts. In Part 1 the participants were asked to fill in a set of questionnaires. In Part 2 a behavioral experiment was conducted. Part 3 involved conducting a short interview with the participants.

#### Part 1—Questionnaires

The participants answered a set of questionnaires preceded by socio-demographic questions. They were also asked about their diet preferences (whether they followed a meat or non-meat diet). This information was collected to control the characteristics of the experimental groups and to avoid any possible impact of this variable on the behavioral data collected in Part 2. All questionnaires were administered in a paper-and-pen form. The time to complete the questionnaires was not limited.

Part 2—Behavioral assessment

After completing the questionnaires, the participants were asked to move to another room, where behavioral assessment was conducted. They were seated at a small table and received instructions about the experimental procedure. The participants were randomly assigned to one of four experimental set-ups (Table 1). Each person received a set of three food products (A—a cookie, B—a muffin, C—a date ball). The products were placed on a paper plate, and a label with the name of the product was placed above each item (Figure 1). In two of the four groups, the labels explicitly stated that the products contained insects and in the two other groups, there was no information about insect content. In addition, half of the participants were given samples that could suggest the presence of mildly

processed insects in the food (e.g., chunks of cranberries or nuts easily discernible in the products), while the other half were served "smooth" products. The experimental group set-up followed the ANOVA 2 × 2 procedure (Table 1).

**Table 1.** Experimental group set-up. The products in groups 1A and 2A (no visual clues suggestive of insect content) and in groups 1B and 2B (presence of visual clues suggestive of insect content) looked identical, but differed in the information provided on the labels.


**Figure 1.** A set of three food products: (from left to right:) a cookie, a muffin, a date ball. Labels attached to food products: **Group 1A:** Rice flour cookie; amaranth flour cupcakes; plain date balls—products labelled as not containing insects. **Group 2A:** Cricket flour cookie; mealworm flour cupcakes; beetle flour date balls—products labelled as containing insects. **Group 1B:** cookies with chunks of cranberries; cupcakes with chunks of walnuts; date balls with linseed—products labelled as not containing insects. **Group 2B:** Cookies with crickets; cupcakes with particles of mealworm larvae; date balls with May beetle particles—products labelled as containing insects.

The participants were asked to try the products in a given order. The questionnaires assessing the properties of each of the three products were placed above the plate in front of each participant. The questionnaire concerned the respondents' willingness to try the product; the visual attractiveness of the product; smell; taste; and willingness to eat more of the product. The respondents were asked to indicate their answers on a 10-point linear scale with descriptive explanations provided at the end-points. In addition, each participant received a napkin and a cup filled with water. The participants were told that they were free to eat as much of any of the products as they wanted, and that they were free not to try any of them if they did not want to.

Each participant's behavior during the food-tasting session was recorded with a video camera placed approx. 2 meters in front of the participant.

Part 3—Interview

After the tasting session, the participants moved to an adjacent room for a short qualitative semi-structured individual interview. The interviews were aimed at broadening the scope of data collected in the study and at providing a context for the quantitative data collected. The scenario comprised a few precisely defined thematic areas, but the questions pertaining to those areas were not pre-determined and they were individually adapted to the respondent and interview dynamics.

The aim of the interview was to examine the opinions and experiences of the respondents with respect to entomophagy; to gain a more in-depth knowledge about the participants' experience of the food-tasting session; to understand the motivations behind the (un)willingness to try the products; and to examine the declared readiness to include insects in the daily diet and the participants' view on the social aspects of insect-based diets. The participants were asked about their previous experience with entomophagy (e.g., whether they had ever heard of this phenomenon before, whether they had ever tried insects, whether they believed that insects could be part of a normal human diet and what potential consequences of such a diet could be). Next, the participants were asked about their impressions from the tasting session and the likelihood of including insects in their own diet. The interviewees were encouraged to elaborate on their answers so that a wider range of information could be collected.

The interviews were recorded and transcribed.

#### *2.3. Ethical Statement*

All subjects gave their informed consent for inclusion before they participated in the study. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Commission of Ethics of Scientific Research of the Faculty of Psychology, University of Warsaw, Poland (No. 21/03/2019). Additionally, the participants signed a consent form agreeing to the processing of their personal data (including audio and video recordings). They were also assured that they could withdraw from the experiment at any point of the procedure and that all the data collected during the study (including the recordings) would only be available to the research team members.

#### *2.4. Questionnaires*

All the questionnaires were translated from English into Polish using the back-translation method [49,50]. First, the questionnaires were translated from English into Polish by a professional translator, then a different translator translated the Polish version back into English. The original English and the back-translated versions were compared by a native English speaker. The text was edited according to the native English speakers' comments, following which the reformulated items were translated into English by a translator not familiar with the content of the comments or the original questionnaire; they were then compared with the original by a native English speaker. The final Polish versions of the questionnaires were then consolidated. The internal consistency of the Polish versions of the questionnaires was determined using Cronbach's alpha.

#### 2.4.1. FNS

The Food Neophobia Scale (FNS), developed by Pilner and Hobden [33], is often used to measure the willingness to try new foods [51]. It consists of ten items on a scale from 1 ("strongly disagree") to 7 ("strongly agree"); five statements are positive (indicative of neophilic attitudes) and five are negative (indicative of neophobic attitudes). Examples of statements include: "I don't trust new foods" and "At dinner parties, I will try a new food". The internal consistency of the Polish version of the scale in our study measured using Cronbach's alpha was estimated at 0.87.

#### 2.4.2. GNS

The General Neophobia Scale (GNS) is a scale measuring the general level of neophobia. It was developed by Pliner and Hobden [33] together with the Food Neophobia Scale. It is an eight-item 7-point Likert scale from 1 ("strongly disagree") to 7 ("strongly agree"). Examples of statements include: "I am afraid of the unknown" and "Whenever I am on vacation, I can't wait to get home". The internal consistency of the Polish version of the scale in our study measured using Cronbach's alpha was estimated at 0.91.

#### 2.4.3. FTNS

The Food Technology Neophobia Scale (FTNS; [31]) is a tool suitable for measuring the willingness to try new food products manufactured using new food technologies and attitudes to new food technologies. It consists of thirteen items scaled from 1 ("strongly disagree") to 7 ("strongly agree"). Examples of statements include: "It can be risky to switch to new food technologies too quickly" and "New products produced using new food technologies can help people have a balanced diet". The internal consistency of the Polish version of the scale in our study measured using Cronbach's alpha was estimated at 0.88.

#### 2.4.4. The Disgust Sensitivity Scale

The Disgust Sensitivity Scale (Version 1; [52]) is a measure of individual differences in sensitivity to disgust. It is a scale with thirty-two items: the first sixteen are binary "true" or "false" questions, the rest assess how disgusting a given experience seems on a scale from 0 ("not disgusting at all") to 2 ("very disgusting"). Example items include: "Even if I was hungry, I would not drink a bowl of my favourite soup if it had been stirred by a used but thoroughly washed flyswatter" and "Seeing a cockroach in someone else's house doesn't bother me". The internal consistency of the Polish version of the scales in our study measured using Cronbach's alpha was estimated at 0.825.

#### 2.4.5. VARSEEK-Scale

The Variety Seeking Tendency Scale (VARSEEK-Scale; [32]) is a scale measuring individual variety-seeking tendency in food choices [53]. The respondents assessed eight statements on a five-point Likert scale (from "completely disagree" to "completely agree"). Examples of statements include: "I prefer to eat food products I am used to" and "When I eat out, I like to try the most unusual items, even if I am not sure I would like them". The internal consistency of the Polish version of the scale in our study measured using Cronbach's alpha was estimated at 0.89.

#### *2.5. Data Processing and Statistical Analysis*

In addition to using the questionnaires measuring specific psychological traits, we also analyzed the behavioral data collected during the experimental (product tasting) phase. This data was analyzed on the basis of video recordings made during the experiment. We used BORIS software [54] to code behaviors on the basis of the recorded material, which made it possible to define selected behaviors and to assess their duration and frequency. We scored the behaviors the participants engaged in during the entire experimental phase. Consequently, we were able to assign specific scores to the duration of separate bouts of behavior, their frequency, and the total time participants spent engaging in specific behaviors. We measured the following variables: latency to pick up food products from the plate; latency to begin eating; amount of food eaten (for each product separately); time spent eating; time and frequency of sniffing food products, time and frequency of looking at the food products; frequency of drinking water during the tasting session. In addition, we analyzed the scores awarded by the participants for the products tasted.

The measurements of different behaviors taken in the course of the experiment (as a response to the exposure to food products) were analyzed using an analysis of covariance (ANCOVA) with Label (2) × Appearance (2) as between-subject factors. The behavioral measurements served as dependent variables, while Label and Appearance manipulations stood for independent variables. Scores from the questionnaires served as covariates. For multiple comparisons, the Bonferroni correction was applied to reduce the likelihood of Type I error. Differences were considered significant for *p* values of <0.05.

For the amount of food eaten, a repeated measures ANOVA was conducted to analyze the effect of possible differences between the three food products (A—cookie, B—muffin, C—date ball) on a given variable. Label (2) × Appearance (2) were used as between-subject factors and Sample as the within-subject factor (3). Scores obtained from the questionnaires served as covariates.

Descriptive statistics are set out in Appendix A.

The audio data collected during the interviews was transcribed and then subjected to the thematic analysis, which is a method for developing qualitative data consisting of identification, analysis and description of thematic areas [55]. In this type of analysis, a thematic unit is treated as an element related to the research problem that includes an important aspect of data. Two interview moderators, i.e., persons responsible for interviewing the respondents, encoded and analyzed the transcriptions. Next, the created codes were cross-checked. An important advantage of thematic analysis is its flexibility, which makes it possible to adopt a research strategy best suited to the phenomenon under examination. In our study, the thematic analysis focused on various aspects of insect consumption and general attitudes towards insects. However, account was taken of the exploratory nature of the study and the novelty of the phenomenon, and thus the low level of the participants' familiarity with the topic.

#### **3. Results**

Characteristics of the participants. First, we checked the characteristics of the study participants, especially whether they were distributed equally between the four groups with respect to the demographic variables and the basic psychological measures included in the study (Table 2). Statistical analysis showed no significant differences between the four experimental set-ups with regard to gender balance, age, level of education, and psychological variables. Even though the number of women was four times higher than the number of men, the ratio of men to women was the same in each set-up.


**Table 2.** Characteristics of the study groups. Abbreviation sd refers to the standard deviation.

#### *3.1. E*ff*ect of Label and Appearance on Food Acceptance (Behavioural Data)*

Latency to pick up food. The analysis of the latency to pick up the food products from the plate showed a main effect of Label (F(1, 89) = 6.456, *p* = 0.013, eta<sup>2</sup> = 0.068). Products labelled as containing insects were picked up later than those with labels not indicating any insect content (t = 2.541, *p* = 0.013, d = 0.507). There was no interactive effect of Label and Appearance, nor a main effect of Appearance. We also found main effects of the covariates General Neophobia (F(1, 89) = 5.825, *p* = 0.018, eta2 = 0.061) and Variety Seeking (F(1, 89) = 6.896, *p* = 0.01, eta2 = 0.072). There was a positive correlation between the latency to pick up food and General Neophobia scores (r = 0.206, *p* < 0.05), while a negative correlation was observed for Variety Seeking Tendency (r = −0.219, *p* < 0.05). This may suggest that participants with a higher level of general neophobia started eating food samples later than those with a lower level of general neophobia. Participants with a higher level of variety seeking tendency started eating sooner than those demonstrating a lower level of this characteristic.

Latency to begin eating. In the case of latency to begin eating, we observed a main effect of Label (F(1, 88) = 5.570, *p* = 0.020, eta<sup>2</sup> = 0.059). Participants began to digest food samples labelled as containing insects later than those with labels not indicating any insect content (t = 2.360, *p* = 0.020, d = 0.503). There was no interactive effect of Label and Appearance, nor a main effect of Appearance. There was no effect of covariates.

Sniffing and looking at products. Analyses of the time and frequency of sniffing the food products and the time and frequency of looking at the food products showed no differences between the groups, which means that participants examined olfactory and visual properties of the samples for a comparable amount of time despite their different labels and appearances. No effect of covariates was found.

Amount of food eaten. A repeated measures ANOVA for the amount of food eaten revealed a main effect of Label for all three food samples (F(1, 90) = 23.918, *p* = 0.001, eta2 = 0.210), but there was no main effect of Appearance or Sample – Figure 2. There were no interactive effects of Sample and Label, nor an interactive effect of Sample and Appearance. This may indicate that the food samples were found to be equally tasty, and the effect of Label was similar for all the samples. Samples labelled as containing insects were consumed in lower quantities then those with labels not indicating insect content regardless of the type of food (sample A: t = 6.282, *p* < 0.001, d = 0.884; sample B: t = 2.637, *p* < 0.001, d = 0.551; sample C: t = 4.646, *p* < 0.001, d = 0.963). A significant effect of the Food Neophobia covariate was also found (F(1, 90) = 4.283, *p* = 0.041, eta2 = 0.045). The correlation between Food Neophobia scores and the amount of food eaten was found to be negative, but only for sample C (r = −0.21, *p* = 0.037).

**Figure 2.** Mean amount of food eaten in each group.

Time spent eating. An ANCOVA analysis conducted for the time spent eating yielded a significant effect for Label (F(1, 89) = 5.922, *p* = 0.017, eta<sup>2</sup> = 0.062), with participants exposed to meals labelled as containing insects eating for a shorter time than individuals who were offered samples with labels not indicating insect content (t = −2.433, *p* = 0.017, d = 0.527). There was no main effect of Appearance, nor an interactive effect of Label and Appearance.

A similar effect was observed in the case of the frequency of eating bouts. The analysis showed only a main effect of Label (F(1, 89) = 10.541, *p* = 0.002, eta<sup>2</sup> = 0.106). Participants who ingested food labelled as containing insects ate less frequently than those who were offered food with labels not indicating insect content. These two results may be correlated, as the shorter consumption time probably involves less frequent bites. No effect of covariates was observed.

*Frequency of drinking water.* Differences between the groups were observed in the frequency of drinking water when ingesting food. An ANCOVA revealed an interactive effect of Label and Appearance (F(1, 89) = 4.625, *p* = 0.034, eta2 = 0.049). A post-hoc analysis showed that participants who ate the samples with labels and appearance not indicating insect content drank water more frequently than those who ate the products labelled as containing insects with a matching appearance (t = 3.214, *p* = 0.011, d = 0.793) and than those who ate the products labelled as containing insects with a "smooth" appearance (t = 2.906, *p* = 0.028, d = 0.121). No effect of covariates was observed.

The above results support the first hypothesis. The effect of Label was found in the latency to pick up food, latency to begin eating, amount of food eaten, and time spent eating variations. Food labelled as containing insects was tasted later and in smaller quantities than food labelled as not containing insects.

However, we found no support for the second hypothesis. There were no differences in the behavioral measures between the foods with traces of insect-like parts and those with a smooth appearance.

Additionally, no interactive effect of Label and Appearance was observed.

The third hypothesis was only partially confirmed. The level of food neophobia correlated only with the amount of food eaten for sample C. General neophobia level and variety seeking tendency level correlated with the latency to pick up food. However, we did not find any confounding effect of disgust and food technology neophobia.

#### *3.2. Influence of Label and Appearance on Food Evaluation (Product Questionnaires)*

We conducted an ANCOVA analysis of food evaluation scores depending on product information (label) and food appearance. For the first question about the willingness to try the products offered, there was a main effect of Label (F(1, 89) = 5.379, *p* = 0.023, eta2 = 0.056) and a main effect of Appearance (F(1, 89) = 4.460, *p* = 0.037, eta<sup>2</sup> = 0.047), but no interactive effect was observed. Participants declared less willingness to try the products labelled as containing insects (t = 2.319, *p* = 0.023, d = 0.471) and the "smooth" products with no easily discernible elements (t = 2.112, *p* = 0.037, d = 0.414). However, there were no differences between the experimental groups with regard to the scores awarded for appearance, smell, taste, and willingness to eat more of the product. There was no effect of covariates.

These findings support the first and second hypotheses, but only as regards the first question of the questionnaire.

#### *3.3. Qualitative Data Analysis—Interviews*

Previous experience of eating insects. Prior to the experiment, all respondents had some experience of entomophagy. In most cases, however, this involved observing insect-eating behaviors in other people (on television, the internet) rather than through direct personal experience of ingesting insects. Insects are perceived as exotic and are associated with Asian (particularly Vietnamese, Chinese, Cambodian, Indian) or African food, as shown on travel or survival TV shows. The few persons who had themselves tried dishes containing insects before participating in the experiment, tried insects bought by their friends as "souvenirs" from far-away countries.

*"For sure, only not in our culture. In Asia, if I'm not mistaken, Thailand, I think. Chinese markets are what I always associate [with insects], like on travel shows, with tonnes of colourful larvae. It is certainly controversial for Europeans—it would most likely be for me, but [I would be willing to try insects] out of curiosity what that would be like (* ... *)."*

Barriers to eating insects. The first reaction to ingesting insects was disgust. In the participants' opinion, insects are slimy and evoke associations with filth, basements, and waste. These associations are further reinforced by the image of insects on TV shows, especially on children's programs.

Because we associate bugs with something disgusting, bugs in food are more often associated with throwing away food and not eating it. Bugs are eaten by wild animals and not by humans.

*I would be afraid that I would feel the structure of this thing and that it would simply be disgusting: limbs or feelers or something like that. I think everyone is afraid that an insect can come alive in your mouth. Out of some internal fear—they are so disgusting and unpleasant.*

The respondents claimed that in our culture "one does not eat insects—it is as simple as that". Some participants, even though they were unable to specifically identify what makes insects so disgusting, pointed to the cultural aspects and the importance of upbringing: "we are simply not used to [eating insects]".

*It is a cultural thing. For us, [insects] are exotic, disgusting, because we have learned to think [about them] this way. [They are] food, as any other type of food, specific for particular regions. And this works this way for us too—we eat pickled cucumbers and sauerkraut, which for some people is simply rotten food. So if we looked at it completely objectively, it seems that eating rotten food is a bigger problem than eating processed insects.*

On the other hand, the participants stated it would be sufficient to "get over oneself and try". They claimed to see a potential advantage in the wider availability of insects and in their potential to become something commonplace and therefore not rejected by the society in general.

*It is all in your head—we have always been told that an insect is just an insect, it is disgusting, it is a pest. I think this has a huge impact. Insects are not soft and flu*ff*y, but they have hard shells or something, so they don't look too good either, to be perfectly honest.*

Perceived advantages of eating insects. Many respondents pointed to insects' high nutritional value as an advantage; they evoked their high protein content and occasionally mentioned other unspecified vitamins and nutrients.

Some respondents mentioned the economic aspects of industrial insect farming. Insects are thought to be inexpensive and easy to produce and transport. Occasionally, the participants described insects as a potential future substitute for meat. Such statements, however, mostly came from vegetarians, who also expressed concerns over the ethical aspects of insect farming. Their answers indicated that they were not certain whether insects were animals. A criterion commonly used for assessing the morality of eating insects was their ability to feel pain. The respondents were not sure whether insects feel pain.

*Perhaps also because, for example, when we kill and eat mammals, they certainly feel more pain than insects—it is as simple as that. Maybe it would be more* ... *humane* ... *this may be the wrong word here* ... *but maybe we could follow that path to reduce the number of animals bred for meat.*

Experiences from the tasting session. The main reason for which the participants decided to try the cookies with insects was curiosity. After seeing a "normal" looking cookie, they were curious whether it tasted different from what they expected from its appearance.

*No, I was wondering if I was going to get anything from the new technologies and if it was going to look strange and resemble God knows what, but I was positively surprised, because it looked tasty. Yes, at first, yes, to some degree, in general, when I saw the labels I thought 'What did they give me? There is no way I'm trying it.' But when I had a look, all looked good and this encouraged me to try it. If I had got an insect on my plate, I would never have tried it. When I tried the first one, I completely switched o*ff *thinking that I was eating insects.*

The decision to try the products was also made easier by the fact that the products looked appetizing and nice. Another safeguard encouraging the participants to try the cookies was the scientific setting; participating in a research experiment guaranteed the safety and hygiene of the products ingested ("you would not give me anything poisonous to eat"). According to the study participants, one of the potential barriers could be the lack of hygiene linked to eating insects, resulting from the aforementioned associations with filth.

The study participants were unable to precisely identify their expectations and assumptions with regard to the taste of the cookies. They expected the cookies to taste strange or different, "like a bug", but they were unable to say what they specifically meant. After trying the products, they referred to their own preferences for desserts rather than to the insect content. Their experience was not particularly positive or negative.

Potential for adding insects to everyday diet. The study participants imagined a potential insect consumer to be a young person who is open to new experiences, with a positive attitude, who eats meat but wants to reduce the amount of meat in their diet or to stop eating it completely. In the participants' opinion, insects could become a fashionable "curiosity" in certain social circles.

The respondents suggested that insect-based food may not be a good idea for the elderly or for vegetarians or vegans (due to the unclear status of insects as animals being able to feel pain). Despite the perceived advantages of insects, their market potential and the positive experiences from the tasting session, most participants claimed that insects were not an appropriate food for them. Yet, some answers suggested a willingness to try insects if they were commercially available. The participants thought, however, that it would rather be a one-off experience than a decision to include insects in their diet on a regular basis. They expressed a wish to reduce the amount of meat and not a need to include other types of meat in their diet. They did not see any direct value for themselves resulting from including insects in their everyday diet.

*It can always be a new taste. I'm curious, I must admit. This can always be a new food form. I doubt that I would be eating [insects] in any large quantities—I'm more interested to just try [them]. I doubt I would try the same form. I have eaten a cookie, maybe [I could eat] an entire cricket, but it only happened once, and that's probably enough. Yes, I would most likely not include [insects] in my diet, but I would only try [them], in small amounts, out of curiosity.*

The participants believed that the way of serving and the manner of presenting insects could help encourage more people to eat insect-based products. It would be best if insect bodies or parts were not discernible, i.e., if they were added in powder form as a ground protein additive. Some insect types evoke more disgust than others, and the participants stated explicitly that their names should not be provided on product labels or should be indicated in another way that does not elicit negative associations (e.g., larvae). Overcoming the barrier involving associations with filth could be facilitated, in the view of some participants, by the mere presence of products containing insects in grocery shops. They believe that making a product widely available on the market makes people perceive it as being tested and suitable for human consumption. Other respondents pointed to the need to create appropriate insect production safety certificates.

#### **4. Discussion**

The analysis of the data collected during the experiment revealed a significant effect of Label on product evaluation. Products labelled as containing insects were ingested in smaller quantities and less frequently, regardless of their appearance. In addition, the latency to pick up and eat products was higher in the case of products labelled as containing insects. The effect of label was also found in a study by Mancini and colleagues [14]. The addition of elements imitating insect parts had no effect on consumption levels. This shows that labelling a product as containing insects is, in itself, sufficient to elicit a reluctance to ingest it and results in a decrease in the amount of food ingested. This effect seems to occur regardless of the form in which the insects are served (insect parts or insect flour). Moreover, the type of insect (larvae, crickets, or cockroaches) specified on the label did not affect the quantity of food ingested, which was comparable for all three products. Additionally, when filling in the food evaluation questionnaires, the respondents stated that they were less willing to try products containing insects regardless of insect type and appearance.

While the effect of Label observed in our study is in line with our expectations and confirms the commonly expressed reluctance and aversion to ingesting insect-based products or products with insect content [22–24], the fact that no effect of appearance was observed raises several questions. It cannot be ruled out that the appearance of the elements imitating insect parts affected the outcome of the evaluation. The elements were sufficiently small that they did not resemble whole insects or discernible insect body parts (limbs, wings, etc.) It seems that the form of the added elements did not elicit associations with contamination or pollution [44]. At the same time, this effect may confirm earlier observations that adding processed insects elicits fewer negative impressions than adding whole insects (e.g., [20,39]). It may be suggested that the required level of processing need not reduce insects to flour—it is sufficient that consumers are unable to discern insect body parts in the product.

It is interesting, however, that although the respondents were less willing to try products containing insects, there was no difference in scores awarded for sensory qualities. All the products received similar scores on the taste, smell, and appearance scales. Moreover, there was no difference between the groups with regard to the respondents' willingness to eat those products again. Possible explanations for this may be found in the interviews carried out after the tasting session. The participants stated that they had expected a specific insect taste, and when it turned out that the products labelled as containing insects did not have a new or an unfamiliar taste, they scored the products in the same way as they would score any other cake or pastry. These results are in line with the findings of Sogari and colleagues [43], who showed that both unprocessed and processed insect-based products generate more positive perceptions after tasting compared to expectations. This may suggest that the absence of an unfamiliar taste decreased the novelty effect, thereby reducing neophobia (cf. [27,56]). This is borne out by the fact that in most of the conducted analyses there was no effect of food neophobia as a covariate (cf. [20,57]). The forms of the products (cookie, muffin) were familiar and their taste did not differ from the taste of regular cakes, which may explain why this factor was not observed in the analysis. The only less-commonly known product was the date ball. In this case, food neophobia was only manifested in the amount of food eaten. Participants scoring higher on the food neophobia scale ingested less of the product labelled as containing insects than persons with lower food neophobia levels. Of the three products used in our study, the date ball is the least common and the least widely available in shops. This reduced availability may have elicited a neophobic reaction to an unfamiliar product in some participants [27,56].

The reluctance to try products labelled as containing insects, measured by the latency to pick up food, revealed an effect of the General Neophobia and Variety Seeking Tendency covariates. Participants scoring high on the GN scale picked up the products labelled as containing insects after a longer time than those with low GN levels. Conversely, respondents with high VST scores picked them up sooner than those scoring low on the VST scale. This suggests that the effect of novelty of the products labelled as containing insects manifested itself immediately when participants came into contact with the products. After the product was assessed as safe, however, this effect decreased. The absence of the novelty effect was also observed in the investigative behavior measurements. There were no differences between the groups with regard to the amount of time spent sniffing and looking at the products.

The frequency of water drinking is difficult to explain. There was an interactive effect of Label and Appearance for this variable. When ingesting products in the case of which no insect content was either explicitly indicated on the label or implied by the additional particles, the participants drank water more frequently than when eating products with explicitly stated or implied insect content. It seems that those respondents who ingested non-insect products ate more of the food and therefore needed to drink more water when ingesting it.

The initial reaction to insect-based food that people commonly express verbally is disgust, as confirmed by previous studies (cf. [20,57,58]) and by the interviews conducted in our study. However, this variable was not observed in the behavioral assessment measures, which may be linked, as mentioned above, to the appearance of the elements added to the products that did not elicit associations with contamination or pollution (cf. [44]) and did not resemble whole animal bodies. Low disgust levels may also stem from the experimental setting; in the interviews, the respondents said that they were convinced that in a scientific experiment they would be given safe and hygienic products. They also mentioned that the products were aesthetically pleasing and their dessert form encouraged the participants to try them. This may indicate that serving insects as ingredients in favorite desserts may be a good strategy (cf. [36]), but it may also depend on the food preferences typical of a specific culture. Moreover, it seems that the feeling of disgust reported when thinking about ingesting insects is replaced by other emotions during contact with an aesthetically pleasing product prepared and served in a safe setting.

This may suggest that it is possible to create such products and eating conditions that could help reduce the effect of disgust. If a product is familiar (cf. [14]), has an aesthetically pleasing form, and if consumers are convinced it is safe, such a product may be accepted despite information on insect content (cf. [58,59]). Outside research settings, product safety is determined on the basis of where it is sold. The respondents stated that a product's widespread commercial availability in shops would encourage them to think that it is safe and that its sales are monitored (cf. [60]). Official safety certificates awarded by appropriate institutions or organizations could have a similar effect. Widespread availability would also convey the impression of a product being widely consumed by others (social proof—[61]). Limited availability, coupled with the fact that they are sold in tourism and pet shops, leads people to still perceive insect-based products to be exotic and foreign.

To conclude, the results of our study confirm the first hypothesis. However, the second hypothesis was not confirmed. The participants, regardless of the appearance of the products, ingested products labelled as containing insects in smaller amounts. In the interviews, the respondents implied that the information about insect content was in itself enough to evoke reluctance and doubts as to whether they should try the product. They pointed out that images of insects on product packaging would also create a negative impression and suggested that certain types of insects (e.g., mealworm larvae) were more disgusting than others.

It seems, however, that it is without consequence whether insects are added in the form of small pieces or flour, as long as whole insects or their body parts are not easily discernible. It is a good strategy to add insects to well-liked products. Insects should be added to familiar products, with familiar taste/smell/texture, and the elements added should not change those properties. The products themselves should have an aesthetically pleasing appearance. All those measures may help convince consumers that such products are safe. In this way, the consumers' preferences for specific products may be generalized to insect-based products.

The study has many implications for management and marketing strategies. First, the results of the study showed that the manner of communicating information on insect-based ingredients has a huge impact on the perception of the product and its future marketing success. The presence of such information is enough to reduce interest in the product. Therefore, placing such products on the market should be preceded by extensive consumer research conducted with a view to selecting the right message and labelling to eliminate that negative effect. Second, to ensure a product's marketing success, it is important to select the right target group. Our research has shown that the group most open to insect-based food are experimenters and variety seekers; on the other hand, the group most reluctant to accept such products are people with high levels of neophobia. This result suggests that an effective positioning strategy of a product containing insect ingredients should refer to "variety seeking" or "experimenting". This study also showed no differences in the evaluation of taste regardless of whether the products had been labelled as containing insects or not. At the same time, unwillingness to try products labelled as containing insects suggests that the problem does not lie in the taste of the product (which has also been confirmed by other studies), but rather in some sort of prejudice against products that contain insects. This is linked to the third implication: it is probably worthwhile to place such products on the market using a "sampling strategy", such as food tasting campaigns in shops, which can help consumers overcome their mental block and try these products.

A very significant aspect of our study is the fact that we used several behavioral measures and not only participants' declarations, as is the case with many studies conducted to date. In doing so, we were able to observe real-life multidimensional reactions to the products used. The interviews conducted after the behavioral assessment phase proved helpful for interpreting the quantitative results and provided ideas for future research and practical solutions.

The study, however, has certain limitations. The study population was characterized by a significant gender imbalance, which prevented us from analyzing the effect of the gender variable. Nevertheless, there was no theoretical basis for assuming such an effect, and we strove to ensure that the gender ratio was identical in each group. The same applied to the remaining demographic data, i.e., level of education and age, as well as dietary preferences—no differences were observed for those variables between the groups. Another limitation was that the majority of participants were students, which may reduce the ability to generalize the results to the general population. However, the participants in our study were full-time as well as evening-course students. The latter represent a broader spectrum of population, as they are often older than full-time students; they often work full-time and have families, which may reduce the effect of the specificity of the study group. Nevertheless, future research should be conducted on a more socio-economically diverse sample to help identify the general mechanisms underlying the phenomenon examined and to ensure the relevance of the findings for marketing strategies. Another important factor that should be considered is personal food preference, especially attitudes towards meat consumption. The results of the interviews show that people following a vegetarian diet are not sure about the appropriateness of consuming insects, as they do not fully understand whether arthropods are able to feel pain, etc.

It is crucial to examine cultural differences between the populations (e.g., participants from different countries), as populations may differ in their food preferences. It is possible that this factor substantially influences acceptance of insects as food and the choice of specific products. Cross-cultural comparative studies (including Asian countries) would also shed light on the universality of the mechanisms underlying the acceptance/rejection of entomophagy.

With regard to the possibility of applying research results in practice, future studies should examine other measures identifying different personal characteristics, such as the tendency to take risks, curiosity, and the need for exploration, as well as variables related to health and moral values.

**Author Contributions:** Conceptualization, K.M., D.A., K.G., D.M. and W.P.; methodology, K.M., D.A., D.M. and W.P.; validation, K.M., D.M. and W.P.; formal analysis, K.M., D.A., and W.P.; investigation, D.A., K.G.; resources, K.M. and W.P.; data curation, K.M. and W.P.; writing—original draft preparation, K.M., D.A., K.G., D.M. and W.P.; writing—review and editing, K.M., D.A., K.G., D.M. and W.P.; visualization, K.M.; supervision, K.M.; project administration, K.M. and W.P.; funding acquisition, W.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the National Science Centre in Poland, grant number UMO-2017/27/B/HS6/01197.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Appendix A**



#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Energy Density of New Food Products Targeted to Children**

#### **Danielle J. Azzopardi 1, Kathleen E. Lacy 2,\* and Julie L. Woods <sup>2</sup>**


Received: 12 June 2020; Accepted: 23 July 2020; Published: 27 July 2020

**Abstract:** High dietary energy density (ED) is linked to childhood obesity and poor diet quality. The Australian Health Star Rating (HSR) system aims to assist consumers in making healthful food choices. This cross-sectional study used 2014–2018 data from the Mintel Global New Products Database to describe the ED of new food products targeted to children (5–12 years) released after the introduction of HSR and examine relationships between ED and HSR. Products were categorised by ED (low < 630 kJ/100 g, medium 630–950 kJ/100 g, high > 950 kJ/100 g) and HSR (no, HSR < 2.5 low, HSR ≥ 2.5 high). Non-parametric statistics were used to examine ED and HSR. A total of 548 products targeted children: 21% low, 5% medium, 74% high ED. One hundred products displayed an HSR: 24% low, 76% high; 53 products with both high HSR and ED. The EDs of products differed by HSR (*p* < 0.05), but both group's medians (HSR < 2.5: 1850 kJ/100 g, HSR ≥ 2.5: 1507 kJ/100 g) were high. A high proportion of new products had a high ED, and the HSR of these foods did not consistently discriminate between ED levels, particularly for high ED foods. Policies to promote lower ED foods and better alignment between ED and HSR may improve childhood obesity and diet quality.

**Keywords:** energy density; health star rating; children; food supply; front-of-pack label; discretionary

#### **1. Introduction**

Childhood overweight and obesity are global concerns. The worldwide prevalence of overweight and obesity in children and adolescents is just over 18% [1]. In Australia, at least one in four children and adolescents aged 5–17 years are currently considered overweight or obese [2]. Measures of overweight, obesity, and adiposity are positively associated with dietary energy density [3–5].

The energy density (ED) of a food is defined as the amount of energy in a specific weight of that food and is usually expressed as kilojoules per 100 g (kJ/100 g). The macronutrient composition and moisture content of the food determine its ED, with foods higher in fat tending to have higher EDs than other foods and water-rich foods tending to have lower EDs than other foods. Food energy density is potentially modifiable by adjusting the macronutrient and/or moisture content of foods. Multiple within-subject crossover design experimental feeding studies have demonstrated that lowering the ED of foods, while maintaining their palatability, reduces children's energy intake (EI) [6–8]. Dietary energy density can be reduced by adjusting food energy density, incorporating more foods that are lower in energy density into the diet or reducing consumption of energy-dense foods. In children, decreasing the ED of the diet is a way to prevent overconsumption of energy without reducing EI below the child's current needs [7] and could contribute to a reduction in rates of childhood obesity.

Diets that are lower in ED tend to be of higher quality [9–11] and more in line with dietary guidelines [9]. They tend to include plenty of vegetables, fruit, wholegrain cereals, low-fat dairy, lean sources of protein and healthy oils [12]. There is strong evidence that higher ED diets are of lower quality [9–11]. Studies involving children and adolescents in several countries have shown dietary ED to be positively associated with the consumption and availability of discretionary foods high in sugar and fat and inversely associated with the consumption and availability of fruit, vegetables, protein and fibre [13–16].

Several population-based surveys have found that Australian children regularly consume discretionary foods [17–19]. The Australian National Health Survey found that just under 40% of the total energy consumed by 4- to 13-year-old children came from discretionary foods [19]. Children in this age group have considerable input into the food products purchased for them by their carers, initially through "pester power" [20] and then through a more collaborative decision-making process [21]. Understanding the retail food supply targeting this age group is important for developing strategies to improve children's dietary intakes.

The Health Star Rating (HSR) system was introduced in Australia in 2014 as a voluntary, front-of-pack label to assist consumers in making healthy food choices in a discretionary food-flooded environment. This system rates the overall healthiness of a product on a scale from a half to five stars, with a greater number of stars indicating a healthier product [22]. The number of stars is calculated based on an algorithm, which scores 'negative' nutrients (energy, saturated fat, total sugar and salt) and 'positive' attributes (fruit, vegetable, legume and nut content and, in some cases, protein and fibre content). The number of stars a product receives should increase as its ED decreases, giving this system the potential of assisting consumers in choosing lower ED options. Being a voluntary system for food manufacturers, only 40.7% had taken it up in 2019 [23]. It is evident that manufacturers are selectively applying the HSR to foods, which score ≥ 3.0 stars (i.e., healthier choices), and have been reluctant to display it on foods with lower star scores, although supermarket own brands have applied it across all products regardless of the score [24].

Despite the current knowledge of dietary ED and its links with increased EI, lower diet quality and obesity, no studies have examined the ED of new food products targeted to children entering the retail food market in Australia, either long-term or since the introduction of the HSR system. Advertising and supermarkets target children and promote the consumption of discretionary foods [25–27]. The high ED of such foods results in a greater likelihood of excess EI and overweight and obesity. Examining the ED and HSR of these products is vital to provide a greater understanding of the food supply and the potential of the HSR system in being able to distinguish between foods with high and low ED. The results of this study could potentially be used to advocate for change in the HSR system, which, in turn, may influence food manufacturers to release products with lower ED.

The aims of this study were to:


#### **2. Materials and Methods**

This cross-sectional study examined the ED and HSR, where available, of all new products targeted to children and launched in Australia from 27th June 2014 to 27th June 2018 recorded in the Mintel Global New Products Database [28].

#### *2.1. The Mintel Global New Products Database*

The Mintel Global New Products Database (GNPD) is an online database of consumer-packaged goods from 62 countries, created and maintained by Mintel, a private international market research company [28]. A network of trained Mintel shoppers frequently monitors the release of new products and updates the database at least monthly. The database captures more than 80 fields of information per item for 17 distinct categories of foods: Baby Food, Bakery, Breakfast Cereals, Chocolate Confectionery, Dairy, Desserts and Ice-Creams, Fruits and Vegetables, Meals and Meal Centres, Processed Fish, Meat and Eggs, Sauces and Seasonings, Savoury Spreads, Side Dishes, Snacks, Soups, Sweets and Gum, Sweet Spreads and Sweeteners and Sugar. The dataset consists mainly of packaged foods and does not generally include fresh, non-processed single foods, such as fresh fruit and/or vegetables.

GNPD fields of information include nutrient data, packaging format, claims made and manufacturing details. The GNPD records if a product is targeted to a particular demographic, namely, babies and toddlers (0–4 years), children (5–12 years), teenagers (13–17 years), females, males and seniors (aged 55+ years). The present study used the children (aged 5–12 years) demographic category. As of 27th June 2018, the database listed 62,066 foods and beverages released in Australia since its launch in 1998. A total of 2683 of these are included in the children 5–12 years demographic category, with 579 products added under this demographic since June 2014 [28].

#### 2.1.1. Search for Products in Demographic 5–12 Years

The GNPD was searched using filters that restricted results from June 2014 to June 2018 in Australia, to foods (not beverages) and for children aged 5–12 years. The GNPD defines this demographic category as foods designed for consumption by children and, more specifically, products which are "also dependent on presentation and format, such as child-inspired graphics like cartoon characters, bright colours and/or pictures of children, or particular language like 'great in lunch boxes' [29]. Data from all 17 GNPD food categories were used in this study; however, some sub-categories were excluded. Beverages were excluded because the grouping of beverages and foods together when calculating ED complicates the interpretation of the results, as beverages are relatively low in ED due to their high water content and can have a substantial impact on overall ED values [30].

#### 2.1.2. Additional Product Searches

To ensure no products were missed, further searches were performed without the demographic filter but with the addition of relevant keywords, such as 'children', in the product description. These searches also used filters that restricted results from June 2014 to June 2018 in Australia.

#### 2.1.3. Data Extraction

In this study, data from 10 of the 80 fields available for each product were extracted from the GNPD: date published, company, brand, product name, category, sub-category, energy (kJ/100 g), demographic, packaging pictures and ingredients list. Company and brand fields were extracted to help exclude duplicates and identify products that display an HSR. Packaging images and descriptions of each product were extracted in order to determine the presence of an HSR, as the GNPD does not routinely include information about HSR. Data were downloaded into Microsoft Excel for analysis.

Sorting was used to remove duplicates from multiple searches. Seasonal products, such as Halloween confectionary and Easter chocolates, were also removed as these products are not available all year and so do not make up the typical range of food items available to children. It is possible for a product to lie within multiple demographics; for example, Bellamy's Organic Apple Snacks are listed with both the 0–4 years and 5–12 years demographics. These records were retained, even if the food category was Baby Food. The description and packaging images for such products were examined, and the product was removed if determined unsuitable (e.g. supplement or formula drinks/foods, such as Ensure). Where a product was a variety pack, that is, two or more flavour varieties of the same or

similar food in the one pack, the record was duplicated for each variety. The overall total of products was 548.

#### 2.1.4. Data Cleaning

Data were checked for accuracy and completeness, and a total of 23 records were found to be missing the value for ED (kJ/100 g). Thirteen of these were variety packs, and the missing data were found on the nutrition information panels from the product images. For 9 of the remaining records, missing data were retrieved from similar products of the same brand (*n* = 2) or different brands (*n* = 7) in the GNPD. Missing data for the final record (Tic-Tac) were obtained from the product website.

#### *2.2. Determination of Energy Density Category, HSR Presence and Core or Discretionary Classification*

Products were categorised into one of three ED categories (low: <630 kJ/100 g; medium: 630–950 kJ/100 g; high: >950 kJ/100 g), according to those defined by the World Cancer Research Fund [31]. Whether a product had an HSR and the number of stars it had was determined by examining packaging images from the GNPD and were added to the relevant record. Each product with an HSR was then classified as discretionary or core, according to the Discretionary Food List produced by the Australian Bureau of Statistics (ABS) in the Australian Health Survey User Guide [32].

#### *2.3. Statistical Analysis*

All statistical analyses were conducted in IBM SPSS Statistics version 23 (IBM, St Leonards, NSW, Australia). The ED data were not normally distributed, and so medians and interquartile ranges (IQRs) were reported and used for analysis. Descriptive statistics (frequency, median, IQR, minimum and maximum) for EDs were calculated for each food category, products without an HSR, products with an HSR, products with a low HSR (<2.5 stars) and products with a high HSR (≥2.5 stars). Mann-Whitney U tests were performed to compare the EDs of foods with an HSR to those without, foods that had low (<2.5) and high (≥2.5) HSRs and foods that were classified as core or discretionary. A Chi-square test was performed to compare the proportions of low, medium and high ED products within the groups of products with and without an HSR. The proportion of products with <2.5 stars and ≥2.5 stars in each of the three ED categories was determined, but inferential statistics could not be performed due to violations of assumptions for non-parametric statistical tests.

#### *2.4. Ethics*

This study did not include an animal or human participants or existing data collected from them and so, in accordance with Australia's National Statement of Ethical Conduct in Human Research [33], is deemed negligible risk and did not require ethical review.

#### **3. Results**

#### *3.1. All Products: GNPD Food Category Distributions and Energy Densities*

The 548 food products targeted to children released into the Australian market between 27th June 2014 and 27th June 2018, were from 14 of the 17 food categories in the GNPD. No products were found from the Sauces and Seasonings, Soup and Sweeteners and Sugar food categories. The greatest proportion of foods (30.7%) was from the Snacks category, and almost half (49.3%) were from the Snacks and Bakery categories combined. The additional inclusion of the discretionary categories Chocolate and Confectionary, Desserts and Ice Cream and Sugar and Gum Confectionery represented 76.5% of the entire sample.

The EDs for the sample ranged from 6 kJ/100 g to 2556 kJ/100 g (Table 1). Aside from Fruit and Vegetables (138 kJ/100 g), the categories with the lowest median EDs were Dairy (377 kJ/100 g) and Desserts and Ice Cream (409 kJ/100 g). Nine of the 14 categories had median EDs that were considered high (i.e., >950 kJ/100 g), although three of these categories had low numbers of foods (*n* < 3). There were 117 (21.4%) products that were categorised as low ED, 28 (5.1%) as medium ED, with the vast majority categorised as high ED (*n* = 403, 73.5%). In particular, more than 86% of the Bakery, Breakfast Cereals and Snacks items had high EDs, and more than 98% of the Chocolate Confectionery and Sugar and Gum Confectionery items had high EDs.

#### *3.2. Products With and Without an HSR: GNPD Food Category Distributions and Energy Densities*

One hundred (18.2%) of the 548 products in the sample displayed an HSR on the packaging (Table 2). Nine out of the 14 categories of foods targeted to children contained products that displayed an HSR. Three categories (Bakery, Breakfast Cereals and Snacks) made up 80% of all items displaying an HSR and had high median ED. All of the Bakery and Breakfast Cereals items with HSRs had high EDs. Although the median EDs of the group of foods without an HSR and the group of foods with an HSR were similar (1490 kJ/100 g and 1594 kJ/100 g, respectively), the variability of the EDs for the group of foods without an HSR was higher (IQR = 1070) than that for the groups of foods with an HSR (IQR = 774). A Mann-Whitney U test comparing mean ranks for the products with an HSR and those without found that the groups were not statistically significantly different (U = 21182, *p* = 0.395). The proportions of low, medium and high ED products within the groups of foods without an HSR and with an HSR are shown in Appendix A. A Chi-square test for independence indicated no significant association between ED category and the presence of an HSR, χ2 (1, *n* = 548) = 2.695, *p* = 0.26, Cramer's V = 0.07.

#### *3.3. Products with a Low or High HSR: GNPD Food Category Distributions and Energy Densities*

The breakdown of products across food categories for items with low (HSR < 2.5 stars) and high (HSR ≥ 2.5 stars) HSRs is shown in Table 3. Only the categories Bakery and Breakfast Cereals contained products with a low HSR, and these two food categories combined made up 24% of all products displaying an HSR. The remaining 76% of products, those with a high HSR, were spread across nine food categories, with the majority falling under Snacks (52.6%) and Breakfast Cereals (14.5%), both of which had high median ED. The median ED of products with a low HSR was 1850 kJ/100 g (IQR = 147) compared with 1507 kJ/100 g (IQR = 1005) for products with a high HSR. Although both of these medians represent high ED, statistically, the median ED of products with a high HSR (M = 42.26) was lower than the median of those with a low HSR (M = 76.58; U = 286, *p* < 0.05).


**Table 1.** Numbers of products in each Global New Products Database (GNPD) food category classified as low, medium and high energy density (ED) and median ED for each GNPD food category.

 products Seasonings, Soup Sugar categories. Percentages may equal rounding. possible duetoalownumberofproductsinthecategory.IQR,interquartilerange.




maydue to a low number of products in the category.


*Nutrients* **2020**, *12*, 2242

1 No products found in the Baby Food, Chocolate Confectionery, Sauces and Seasonings, Savoury Spreads, Soup, Sugar and Gum Confectionery, Sweet Spreads and Sweeteners and Sugarfood categories. 2 Indicates not possible to calculate due to a low number of products in the category.

Snacks

Total

 0 (0)

 24 (100)

 1850

 147

 1600

−−

 −

 −

 2110

 76 (100)

40 (52.6)

 1607

 1507

 1005

 138

 2200

 454

 227

 2200

The median HSR for all 100 products with an HSR was 3.5 stars (IQR=1.5), with a range of 0.5 to 5 stars. A total of 16 of the 100 products with an HSR were from the low ED category, and all scored a high HSR (median 4 stars (IQR = 1.4); range 3 to 5 stars). All seven of the products in the medium ED category scored a high HSR (median 3.5 stars (IQR = 1); range 2.5 to 4.5 stars). The median HSR for the high ED category was also 3.5 stars (IQR=2) but with the full range of 0.5 to 5 stars represented. Figure 1 shows the scatterplot of HSRs by low, medium and high ED categories. Among the 77 products from the high ED category, only 24 (31%) scored a low HSR. However, 53 (69%) of the products, categorised as high ED, also scored a high HSR, with the majority of these categorised as Snacks. The breakdown of all 100 products by ED and HSR categories is shown in Appendix B.

**Figure 1.** Scatterplot of product Health Star Rating (HSR) by low, medium and high energy density (ED).

#### *3.4. Core vs. Discretionary Products: Category Distributions, Energy Densities and HSRs*

The breakdown of products with HSRs as core or discretionary foods across food categories is shown in Table 4. Overall, 30% of products displaying an HSR were classified as core and 70% as discretionary. The categories Bakery and Snacks combined accounted for 81.5% of all discretionary products. The median ED of core products was 971 kJ/100 g (IQR = 1164) compared with 1800 kJ/100 g (IQR = 355) for discretionary products, with both medians in the high ED range. The EDs of core products (M = 28.72) were significantly lower than those of discretionary products (M = 59.84; U = 1703.5, *p* < 0.05). The distribution of core and discretionary products across the three categories of ED is shown in Appendix C.

The median HSR for core products was 4 stars (IQR = 0.5) and ranged from 2 to 5 stars. For discretionary products, the median was 3.5 stars (IQR = 2.0), with the full range of 0.5 to 5.0 stars. Only 3% of core foods displayed a low HSR and 97% a high HSR. On the other hand, only 33% of discretionary foods displayed a low HSR, whereas the majority (67%) of discretionary foods displayed a high HSR. The distribution of core and discretionary products across the two categories of HSR is shown in Appendix D.




 No products found in the Baby Food, Chocolate Confectionery, Sauces and Seasonings, Savoury Spreads, Soup, Sugar and Gum Confectionery, Sweet Spreads and Sweeteners and Sugar food categories. 2 Indicates not possible to calculate due to a low number of products in the category.

#### **4. Discussion**

Between June 2014 and June 2018, the majority of new food products targeted to Australian children had high ED. Less than 20% of products displayed an HSR, and the HSR system did not consistently distinguish between low ED and high ED products. About three-quarters of products with an HSR were categorised as having a high HSR, and the majority of products with an HSR (70%) were categorised as discretionary foods.

These findings are consistent with previous Australian and New Zealand research, which found that the majority of food products available for sale [34] and directed at children [35] were considered 'less healthy' using nutrient profiling criteria. Additionally, several population-based surveys have found that Australian children regularly consume high ED, nutrient-poor foods [17–19]. While it is important to encourage children to meet dietary recommendations and energy needs through healthful food intake and limited intake of high ED, nutrient-poor foods, additional strategies targeting the retail food market have the potential to assist in moderating children's dietary energy density and energy intake. For example, a recent study showed that total and saturated fat reformulation of some UK supermarket bakery items (cakes and biscuits) could result in substantial reductions in product energy density [36]. In the present study, a large proportion (18%) of the products that entered the retail food market during the four years of interest were bakery items, suggesting a large segment of the market that could also be reformulated in Australia. While food reformulation of processed foods is potentially useful to reduce the energy density of the food supply, it must not be used as a way to increase the perceived healthfulness of discretionary processed foods.

An HSR was displayed on 18.2% of products examined in this study. This result is higher than that obtained in a study by Lawrence et al., who found that 10.5% of new products (using Mintel's GNPD) released between 27th June 2014 and 27th June 2017 displayed an HSR, and a study by Dickie et al. using the same database but for the time period 6 June 2014–30 June 2019, who found an HSR on 17.6% of products [37,38]. Differences in the database dates used, target sample and/or the product sample size could explain these differing proportions of foods displaying an HSR. Bakery and Snacks categories were the most prevalent products displaying an HSR, as also found by Lawrence et al. and Dickie et al. [37,38]. The food categories in this study that did not have any products displaying an HSR were mostly discretionary foods, for example, Chocolate Confectionary and Sugar and Gum Confectionery. As the HSR system is currently voluntary, manufacturers can selectively apply the HSR to products receiving higher ratings. For example, Shahid et al. reported that for a number of manufacturers, there was a 1.9 to 2.5-star difference between mean HSR displayed on their products compared with their other products that did not display the HSR [23]. This is also supported by the finding that just over three-quarters of products in this study had an HSR ≥ 2.5 stars with a median across the whole sample at an HSR of 3.5 stars.

Despite the voluntary nature of the HSR and the propensity of manufacturers to apply the HSR to higher scoring foods, there was no significant difference between numbers of products with and without HSRs in each of the three ED categories. It could be hypothesised that if the HSR system was better aligned with ED, there would be a greater proportion of products with an HSR in the low ED category [23]. This is the first study to look at ED and HSRs, so it is not possible to compare this finding to the existing literature.

Among the foods displaying an HSR, all low and medium ED foods displayed a high HSR, as would be expected. However, only 31% of high ED foods displayed a low HSR, with the remaining 69% displaying a high HSR. Some high ED products may deserve high HSR. For example, The Happy Snack Company's Roasted Fav-va Beans in four different flavours have an ED of 1867 kJ/100 g, yet score highly on the HSR algorithm for being high in protein and fibre and containing more than 80% legume. However, there are also products that are clearly discretionary, such as Messy Monkey Strawberry and Apple Snack Bars by Freedom Foods. This snack item has 4.5 stars, yet has an ED of 1410 kJ/100 g, is one-third sugar, and contains mostly dried fruit, which is recommended as occasional by the Australian Dietary Guidelines [39]. If the HSR was classifying foods correctly on the basis of ED, then we would expect a much lower percentage of high ED foods displaying a high HSR. This further adds to the body of literature, demonstrating the shortcomings of the HSR system [23,37,38,40–43], and shows that it does not consistently discriminate between levels of ED, especially when considering high ED foods. The median ED of products with a high HSR was 1507 kJ/100 g, well above the cut-off (950 kJ/100 g), signifying the beginning of the high ED range [31].

The classification of food products into core and discretionary groups seemed to align more accurately with the ED categories, with increasing proportions of discretionary foods in each increasing ED category. This supports previous studies that have shown that high ED is associated with discretionary foods [9,13,15,44]. The results relating to core foods (only 3% displaying a low HSR) indicated good concordance between core foods and HSR. However, the same could not be concluded with regard to discretionary foods, with 67% displaying a high HSR. Consistent with this study, Lawrence et al. found that 57% of discretionary foods had an HSR ≥ 2.5, and Pulker et al. found that 55% of ultra-processed foods carried HSRs ≥ 3 [27,37]. These findings are concerning in that they show that the HSR is likely to have the opposite effect to what Hawkes et al. posit the role of front-of-pack nutrition labels should be—to decrease the perceived healthiness of discretionary products rather than increase the perceived healthiness of healthy products [45]. By not accurately discriminating amongst discretionary and high ED foods, the HSR is effectively allowing these foods to be perceived as healthier than they actually are.

Several studies in Australia and New Zealand have found that consumers prefer HSRs over other packaging labels, such as nutrition information panels or daily guide, although product visuals (for example, artificial or natural looking food, pictures of fresh fruit, images of sport, etc.) were found to be the foremost influence on choice [46–49]. Hamlin et al. performed a longitudinal study on the effectiveness of the HSR and, despite heavy advertising campaigns for the HSR system in New Zealand, found it to be ineffective at influencing the customer in their choice between products in a food category [50]. Likewise, Ares et al. found the HSR to be less effective than Nutri-score and a warning symbol in catching attention, healthiness perception and intention to purchase (Comparison of three systems) [51]. An international comparison of a number of different front-of-pack nutrition labels found that most increased consumer ability to rank food healthfulness but that colour coded varieties, such as Nutri-score and traffic lights, were more beneficial than the HSR [52].

In light of the continuing support for the expansion of HSRs, it is imperative that the system provides appropriate guidance for shoppers in making food choices in line with the Australian Dietary Guidelines [39]. We have shown here that when choosing between two products with HSRs, selecting the food with the greater number of health stars will not always be the "healthier" or lowest ED choice [42]. With most new foods marketed to children categorised as high ED and the majority of those with an HSR considered discretionary, consumers need a more consistent measure of healthiness.

This is the first study to examine new food products targeted at children entering the Australian retail food market and assess their ED and discretionary or core grouping with their HSR. The Mintel GNPD is comprehensive, up-to-date and well suited to this study and its aims, as it focuses on new product activity. This is particularly relevant as new food products represent ways in which manufacturers have responded to the introduction of the HSR system. It should also be noted that the GNPD does not reflect a product's market share, only its existence, and so the product's pervasiveness in the diets of Australian children is unclear.

It is difficult to keep up with innovations and developments in food items, making it difficult for the Australian Bureau of Statistics Discretionary Food List to accurately distinguish between discretionary and core foods. Errors may have occurred in classifying the 100 products displaying an HSR into discretionary or core categories. However, to reduce the possibility of error, the coding into categories was checked by both co-researchers.

It would be of benefit to extend the work of the current study to cover all food products on the Australian market targeted to children and not just new foods. This study raises questions regarding the three-way relationship between ED, discretionary foods and HSRs. Future research that combines the analysis of these three measures using a larger sample of foods would further the knowledge in this area. The present study weighted all food products equally and not by market share. Research to analyse EDs of food products and adjust their impact using their prevalence in the supermarket would help deepen the understanding around foods available to children. It would also be of benefit to undertake similar research for seasonal products, that is, analysing their availability in the existing market and their market share, as well as studies to measure the impact of seasonal foods on children's diets. Further research is needed into the effectiveness of the HSR system on whether it is meeting its objectives for consumers at the point of sale and resulting in the purchasing of healthier food products.

The Australian Government acknowledges the need to take action against obesity in children by improving the food environment and, therefore, individual diets through the introduction of initiatives, such as the HSR System [22]. The Australian food industry is also making attempts to improve the food environment by introducing voluntary guidelines to reduce the levels of saturated fat, sodium and energy in foods targeted to children. However, these initiatives by the food industry and the Government to get children eating healthier foods will likely have difficulty translating into positive results while they remain voluntary and unenforced [53].

#### **5. Conclusions**

A high proportion of new food products targeted to children is of high ED, and the HSR of these foods, when displayed, does not consistently discriminate between levels of ED or between core and discretionary foods. Most new products for children that display HSR are discretionary foods, which are likely contributing to lower diet quality and excess EI. There exist potential opportunities (prompted by food manufacturers wanting to achieve higher HSRs) to reduce the ED of some of these foods to help curb excess EI and improve diet quality. The results of this study support the need to advocate for a food policy change that will result in lower ED foods and improvements to the accuracy and consistency of the HSR system, with the aim to improve the diet quality of Australian children and reduce rates of childhood obesity.

**Author Contributions:** Conceptualisation, D.J.A., K.E.L. and J.W.; methodology, D.J.A., K.E.L. and J.L.W.; formal analysis, D.J.A.; data curation, D.J.A.; writing—original draft preparation, D.J.A.; writing—review and editing, D.J.A., K.E.L. and J.L.W.; supervision, K.E.L and J.L.W. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Appendix A**

**Figure A1.** Proportions of low, medium and high energy density (ED) products without and with a Health Star Rating (HSR).


### **Appendix B**


**Table A1.** Number of products by energy density (ED) and Health Star Rating (HSR) categories.

#### **Appendix C**

**Figure A2.** Core and discretionary products displaying the Health Star Ratings (HSRs) by energy density (ED) category.

#### **Appendix D**

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Consumer Misuse of Country-of-Origin Label: Insights from the Italian Extra-Virgin Olive Oil Market**

#### **Francesco Bimbo, Luigi Roselli \* , Domenico Carlucci and Bernardo Corrado de Gennaro**

Department of Agricultural and Environmental Sciences, University of Bari Aldo Moro, 70126 Bari, Italy; francesco.bimbo@uniba.it (F.B.); domenico.carlucci@uniba.it (D.C.); bernardocorrado.degennaro@uniba.it (B.C.d.G.)

**\*** Correspondence: luigi.roselli@uniba.it

Received: 23 June 2020; Accepted: 17 July 2020; Published: 19 July 2020

**Abstract:** Providing information to consumers through the label is a means for food companies to inform consumers about product's attributes, including the country of origin (COO). In the EU, COO labeling has been made mandatory for several categories of food products, to enable consumers to make informed choices at the point of sale. In particular, Regulation (EU) No 29/2012 has introduced a mandatory country-of-origin labeling system for extra virgin olive oil (EVOO). In the present study, conducted in Italy, we test whether there is a price differential associated with the COO information for EVOO. To this end, we employ a hedonic price model and data about the purchase of EVOO products collected from 982 consumers at the supermarket checkout. Having interviewed these consumers, we also assess the share of EVOO consumers that correctly identify the country of origin of the product purchased. Our findings point out that, in Italy, the EVOO with domestic origin, indicated on the label, benefits of a premium price equal to +35% compared to the product labeled as blend of European EVOOs, while a discount of −10.8% is attached to EVOOs from a non-European origin. A significant share of consumers in our sample (19.04%) is, however, unable to correctly identify the origin of the EVOO purchased. This label misuse mostly occurs among consumers who report that they had purchased Italian EVOO, while they had actually purchased a blend of European EVOOs. Female and more highly educated consumers are less likely to misuse label information about the product's origins.

**Keywords:** consumer choice; food labeling; extra virgin olive oil; hedonic price model; country of origin

#### **1. Introduction**

Extra-virgin olive oil (EVOO) is the superior olive oil category extracted from olives by the mechanical extraction process. EVOO is one of the main components of the Mediterranean diet and it is considered worldwide as one of the healthiest oils. The European Union (EU) is the largest EVOO producer worldwide, and production is concentrated in three Mediterranean countries: Spain produces about 57.5% of EU EVOO output, Italy over 19.5%, and Greece about the 15.8% [1]. Due to increasing consumption in non-producing countries, both within the EU (e.g., UK and Germany) and outside Europe (e.g., China, Japan, Russia, Australia, Brazil, Canada, and the US), demand for European EVOO is expected to rise until 2030 [2,3]. Rising EVOO consumption in non-producing countries leads Mediterranean EVOO producers to export their EVOO to other markets [2,3]. The steadily increasing demand for EVOO in non-producing countries has led Italy to export a large share of domestically produced EVOO, which is no longer sufficient to satisfy the domestic market. Such supply imbalance is constantly re-balanced by importing EVOO from Spain and non-European Mediterranean countries, such as Tunisia and Turkey. This EVOO is usually priced lower than Italian EVOO. Compared to Italian EVOO producers, the Spanish benefit from economies of scale and non-European producers with lower labor costs [3]. Production costs at farm level for a liter of Italian EVOO are, on average, 30% higher than the production costs recorded for Spanish and non-European Mediterranean producers [1]. As a result, Italian EVOO sold on the Italian market competes with EVOO imported from other countries.

Over the last three decades, consumers have placed increasing importance on the product's origin. However, the latter cannot be verified either ex-ante or ex-post consumption and asymmetric information about the products' origins arises between producers and consumers. Providing information about products' origins through the label has become a widely adopted tool to mitigate the information gap between consumers and producers. Country of origin labeling policies, by informing consumers about a product's origin, transform information about the origin of the product, that is a credence attribute, into a searchable characteristic and so alleviate the problem of asymmetric information [4,5] Generally speaking, labeling policies address a market failure, asymmetric information, through costly expenditures borne by a combination of consumers, firms, and taxpayers. First, the industry bear costs of labeling, which likely pass on consumers at higher prices. Second, the government's costs of label monitoring and enforcement system are borne by taxpayers via higher taxes. Third, mandatory label exacerbates other market distortions such as decrease competition or encourage rent-seeking and gaming, as well as introduces trade distortions across countries [6].

In the case of the EVOO market, EU policymakers have introduced Regulation (EU) No 29/2012, a mandatory labeling information system requiring producers to indicate on the label the country of origin (COO) of the EVOO. This Regulation establishes that the labeling of extra virgin olive oil and virgin olive oil must bear a designation of origin. Bottlers can use, on the label, one of the following claims relating to origin: (a) in the case of olive oils originating from one Member State or third country, a reference to the Member State, to the Union, or to the third country, as appropriate; (b) in the case of blends of olive oils originating from more than one Member State or third country; or one of the following statements, as appropriate: (i) 'blend of olive oils of European Union origin' or a reference to the Union; (ii) 'blend of olive oils not of European Union origin' or a reference to origin outside the Union; (iii) 'blend of olive oils of European Union origin and not of European Union origin' or a reference to origin within the Union and outside the Union; or (c) a protected designation of origin or a protected geographical indication referred to in Regulation (EU) No 1151/2012, in accordance with the provisions relating to the product specification concerned [7]).

On the one hand, the introduction of Regulation No (EU) 29/2012, by informing consumers about the origin of EVOO, is potentially beneficial to Italian EVOO producers. Since Italian consumers strongly prefer domestic EVOO over non-Italian alternatives and are willing to pay a premium price for it [3,8–11], the regulation would support Italian producers to ensure fair revenues for their product. In other words, by informing consumers about the COO of EVOO, Italian producers are able to differentiate their products and add value to these.

On the other hand, the measures based on informing consumers through the label implicitly assume that information on the label eliminates asymmetric information by fully informing consumers about the product's features, and so restoring information symmetry between consumers and producers. The existing literature points, however, to many instances where the labeling policy is not be able to restore full information. This occurs because consumers either do not make full use of the label, or the label's information is not clear, or consumers are not fully aware of the information's availability [12]. Furthermore, consumers may not fully trust label information due to the risk of incurring food fraud. Frauds are more likely to occur in products that benefit from premium prices. Indeed, the latter work as an incentive for producers to commit fraud [13].

As a result, the aim of the present study is twofold: (i) to test whether there is a price differential associated with COO information for EVOO by using retail level data collected from Italian EVOO consumers purchases; and, (ii) to assess in what measure consumers correctly use the information about the origin of the EVOO they purchased. To the best of our knowledge, no previous studies investigated to what extent consumers correctly identify the origin of the EVOO purchased by using the origin information on the label. We did so by interviewing EVOO shoppers at the supermarket checkout about the origins of the product they have purchased. By inspecting the labeling of the product these shoppers purchased, we assess whether their understanding of the product's origin matches what is reported on the label. We then analyze consumer groups of differing abilities in identifying the COO of the product according to their socio-economic characteristics, as well as their self-declared interest in the product's label information, the origin of the product, and interest in branded products more widely. Indirectly, this allows us to infer the effectiveness of the mandatory COO label in correctly informing consumers and orienting their food choices. The remainder of the paper is structured as follows: the next section presents a description of survey design, the data and the model used; then, we discuss the empirical results. We conclude by providing recommendations for EVOO producers and policymakers.

#### **2. Materials and Methods**

#### *2.1. Survey and Data Collection*

The survey involved 982 EVOO consumers interviewed at the supermarket checkout counters. Consumers were selected on voluntary basis and did not receive any monetary compensation to participate in the study. Once consumers were approached at supermarket cashiers were asked their willingness to participate in the study or not. For those who accepted to participate, written informed consent according to the national ethical requirement "Italian Personal Data Protection Code" (L.D. 196/2003) was collected. Then, the interviewers asked them to state the origin of the EVOO they had just purchased. Consumers were free to answer by selecting one of the following statements: "I am unaware of the origin of the EVOO I purchased", "It is a blend of non-European EVOOs", "It is a blend of European EVOOs" and "It is a 100% Italian EVOO". The interviewers then inspected the labeling of the product purchased. By comparing consumers' answers and the information on the origin of the EVOO found on the label, they were able to assess the correctness of what consumers stated about the COO of the EVOO purchased. The share of consumers that correctly identified the origins of the product was used as a proxy for the effectiveness of the EU Reg. 29/2012 to inform consumers about such origins.

The interviewers also collected information on the characteristics of the EVOO that consumers purchased, including its COO attribute, used for sizing their monetary value by means of the hedonic price model. Socio-economic information about the consumers, if they shopped for EVOO as a result of price promotion, as well as their interest in labeling information, in the origin of products, and preference for branded products, were also collected. These served as a proxy for consumers' knowledge about the product purchased. To this end, five-point Likert scales were employed, assigning point 1 to "strongly disagree" and 5 to "strongly agree". Table 1 reports the summary statistics and a description of the data collected on both the products and consumers' characteristics. The data collection of was carried out between March and July 2017 and consumers were recruited from a regionally representative sample of 14 hypermarkets and supermarkets, all located in the Apulian region (Italy). The sample of hypermarkets and supermarkets included at least one outlet in each of the six provincial capitals of the region, as well as a selection of the leading retailers in the Apulian region, namely Auchan, Conad, COOP, and Famila.


**Table 1.** Summary statistics related to product and consumer characteristics (982 observations).

<sup>a</sup> For all binary variables the mean represents the percentage of observations, the value of the standard deviation is omitted.

#### *2.2. Model and Statistical Analysis*

To measure the monetary value of the product's features, we used the standard hedonic price model first introduced by Rosen in 1974 [14]. According to hedonic price theory, a product is considered as a bundle of attributes. Each consumer in the market selects the set of features which maximizes his/her utility, subject to a budget constraint. Likewise, manufacturers maximize profits by setting the product's price according to its attributes [14]. In a market for products presenting a unique bundle of attributes, buyers' marginal bids and sellers' marginal offers match at equilibrium and the joint envelope of consumers' bids and sellers' offers generate the hedonic price function [13]. Thus, the price, P, of a product, j, can be described as:

$$\mathbf{P}\_{\mathbf{j}} = \mathbf{f}(\mathbf{Z}\_{\mathbf{j}}) \tag{1}$$

where Z is a vector of product attributes belonging to product j and f (.) is an unspecified functional form. Equation (1) implies that the price consumers pay for product P is a function of the marginal monetary values of j's attributes Z [14–16] and can be obtained by partially differentiating (1) with respect to each attribute. Furthermore, the implicit marginal price a consumer pays for the attribute Z corresponds to the marginal cost which the producer incurs in offering that attribute on the market. Equation (1) was estimated by ordinary least squares (OLS).

Then, to measure the share of consumers able to correctly identify the origin of the EVOO products they purchased and infer the effectiveness of the mandatory COO information in correctly orienting consumers' food choices, a cross-tabulation analysis was performed. A Pearson Chi-square test and a Goodman and Kruskal's gamma statistic were used to assess whether the outcomes from the cross-tabulation were statistically significant, so testing the presence of a positive and statistically significant association between consumers' statements about the origins of the product and the verified origins [17,18]. The work ends by profiling consumer groups based on their ability to correctly discriminate the product's origins. Consumers' socioeconomic characteristics, their interest in the information on the product's label, in the origin of the product, as well as in branded products, were used to profile consumer groups. A Tukey test assessed whether consumers differ according to the characteristics listed above [19].

#### **3. Results and Discussion**

The estimated parameters of Equation (1), using the logarithmic transformation of the price as the dependent variable, are reported in the first column of Table 2, along with their standard errors in parenthesis. The functional form using the log-linear transformation of the price shows the lowest value of log-likelihood function testing it competitively with the linear and the box-cox transformation of the dependent variable. Marginal prices of each attribute (in percentage terms) are also calculated using Kennedy's (1981) adjustment and reported in the last column [20] In Equation (1), we also control for brand fixed effects. For the sake of brevity, the resulting coefficients are not reported in the manuscript, but are available upon request. The baseline product is a non-organic EVOO from EU countries, filtered, and sold in 1 L glass bottle at an average price of 6.08 €/L. Based on the Ramsey's RESET statistics for omitted variable bias [21], the model does not suffer from misspecification, and, since the null hypothesis of homoscedasticity of the Breusch–Pagan/Cook–Weisberg test cannot be rejected, the error terms are homoscedastic [22,23]. Skewness and Kurtosis test indicates the normality of the error terms distribution [24]. The model shows an adjusted R<sup>2</sup> of 0.9694 and a statistically significant value of the F-statistic, suggesting the join significance of coefficients regressors. These statistics confirm that the semi-logarithmic specification of Equation (1) is the most appropriate among the possible functional forms for f (.).

The first notable finding reported in Table 2 is that the "Italian" origin label has a positive and significant effect on the price of EVOO sold in Italy and amounts to a price premium of +35%, relative to the baseline product, which is equivalent to +2.18€/liter. This result is consistent with other studies that found a willingness of Italian consumers to pay more for domestic EVOO products. Compared to the price of a European EVOO, "Non-European" EVOO is instead sold at a discount of −10.8%, equivalent to −0.70€/liter [6–8].

The geographical indication labels "GIs" show a positive and significant effect on price of +7.12 €/liter, or +112%, relative to the baseline product's price. This attribute records the highest price premium among all the considered EVOO's attributes and the result is consistent with several studies which also found that consumers, including those on the Italian market, prefer GIs products over regular ones and are willing to pay higher prices for such products [25–27]. The higher price of EVOO with GIs may, however, reflect the higher cost of GIs products, since farmers/producers seeking to sell their products with a GI label have to meet costly production standards that are frequently regarded as a barrier for the compliance with GIs standards [28–30].

Interestingly, the "Organic" attribute records a positive and significant impact on the EVOO price of +15.1% over the baseline product price, equivalent to a price premium of 0.91€/liter. The premium price associated with this attribute is likely to be the result of consumers' willingness to pay for a "sustainable" product, which has been reported in several studies [31–33]. Products labeled as organic also are often perceived as healthier than regular ones, and, indeed, consumers' primary reason for buying organic foods is their belief that these products support human health [34]. Thus, the premium price attached to organic EVOO may also be due to consumers' willingness to buy products which they regard as supporting their health.


**Table 2.** Estimated parameters and percentage of premium price.

<sup>a</sup> Adjustment made according to Kennedy (1981). \*, \*\* and \*\*\* are 10, 5 and 1 percent significance levels.

The marginal prices associated with the unfiltered attribute, "Unfiltered", is positive and statistically significant. Unfiltered EVOO is sold with a markup of +21.6% or +1.28 €/liter. This suggests that the "unfiltered" claim on the label can be used by consumers to infer a higher degree of wholesomeness/naturalness [34], and a premium is thus associated with this attribute.

The marginal prices associated with the packaging variables show positive and statistically significant coefficients. They indicate that premium prices are associated with EVOO products sold in glass packages smaller than 1 L. Products sold in glass bottles of 0.75 L ("Package Size 0.75 L") benefit from a premium price of 0.912 €/liter, while products sold in 0.5 L glass bottles ("Package Size 0.5 L") secure a premium price of 0.742 €/liter. The estimated marginal prices for products sold in other packaging material ("Other Packaging Material"), such as plastic or tin, as well as sold on promotion ("Promotion") are negative, but not statistically significant.

With regard to Table 3, the data on consumers' reports at checkout shows that 13.1% of EVOO consumers (129) in our sample were not aware of the origin of the product purchased, while a minority of 0.2% (2) reports their having purchased non-European EVOO. A larger share of EVOO consumers in the sample, approximately 28.0% (275), report having purchased a European EVOO and the majority of them, 255 out of 275, correctly identified the product's origins. The largest group of EVOO consumers interviewed, 58.7% of the total sample (576), reports the purchase of Italian EVOO. In the latter group, one out of three consumers incorrectly identified the Italian product's origin since they believed that they had purchased an Italian EVOO, but the product purchased actually was a blend of European EVOOs (165 out 576). The misuse about the product COO label then occurs more often when consumers report their having purchased Italian EVOO.


**Table 3.** Consumer-declared origin of the EVOO products purchased and the actual one.

<sup>a</sup> The share of respondents over the total number of respondents of the column is reported in parenthesis. The number

of respondents who correctly indicated the product's origin is in the highlighted in bold.

A potential cause of the erroneous consumers' identification of foreign EVOO as Italian may be related to the fact that consumers may be not aware of, or may not use, the COO information on the label. Another cause can be related to the fact that many non-Italian companies, after multiple mergers and acquisitions, hold in their portfolio several Italian EVOO brands (e.g., Bertolli, Carapelli, Sasso owned by Deoleo S.A.; Sagra and Filippo Berio by the Bright Food Group Co Ltd.). Companies use these brands to market non-Italian EVOO, so increasing the likelihood that consumers mistakenly infer from the brand name that the origin of the EVOO purchased is Italian. The use of a brand name with a more favorable image (in this case an Italian brand-name) to deliberately lead consumers to associate the origins of brand and product is a strategy previously reported for many other consumers goods markets [35]. If consumers use the brand as a clue to the origin of the product and other information on the origin of the product is not taken into account in the purchasing process, this marketing strategy lowers the ability of consumers to correctly identify the product's origins. As proposed by Zhou, Yang, and Hui (2010, p. 204), *"the origin information for most brands may not be readily accessible either because global marketers have the desire to mask the origins of their brands or the globalization of firms and the cross-border acquisition of brands complicate the nature of brand origin"* [36].) Lastly, another hurdle for consumers in correctly identifying the origin of the product may be the fact that several Italian EVOO producers, aiming to offer consumers a greater variety of products, sell both Italian and non-Italian products under the same brand name. Such choice may further lower consumers' ability to correctly identify the country of origin (COO) of the product during their food purchase.

Overall, the data in Table 3, reported in bold, indicates that 666 consumers, 67.8% of the sample, correctly associate the COO of the EVOO purchased. The positive association between the reported and verified origins of the product, and thus the overall ability to correctly associate the product and its origins, is statistically significant. The Goodman and Kruskal's gamma statistic and the Pearson Chi-square test have a low *p*-values (<0.05). This indicates that the likelihood that consumers' identification of the product's origin corresponds to the correct one is high [17,18].

On the one hand, the results discussed above indicate that, overall, the mandatory COO information on EVOO products (EU Reg. 29/2012) is an effective tool in guiding consumers in the identification and selection of EVOO based on its origins. On the other hand, results show that there is still a consistent number of Italian consumers (187) who do not correctly associate the product with its actual origins and in most of the cases they are consumers who reported, and believed that they had purchased an Italian EVOO. Thus, the misuse of COO label mostly occurs where foreign EVOO products are identified as Italian.

Lastly, Table 4 identifies consumer groups according to their ability to correctly associate the COO of the EVOO purchased, characterizing them in relation to their socio-economic characteristics, as well as their interest in labeling information, in the origin of the product and in branded products more generally.

The first consumer group, reporting that they are unaware of the product's origin, encompasses 13.1% of consumers sampled. These consumers are mostly male, with a lower level of education compared to the average level of EVOO consumers in our sample. Furthermore, consumers unaware of the origins of the product live in a household with less than 3 individuals and their income is lower than in the other groups. These consumers purchase EVOO more often than others when it is sold on promotion at an average price of approximately 5.50 €/liter. Compared to consumers in the other two groups, these consumers also reported less interest in the information on the label, in the product's origins and in brands.

The second group, representing 19.04% of EVOO consumers in the sample, encompasses individuals who incorrectly identify the country of origin of the product at the supermarket checkout. Consumers in this group are mostly male and have a higher level of education than those reporting unawareness of the product's origins. Furthermore, they live in larger households and purchase products on promotion at 5.88 €/liter, which is not statistically different to the 5.50 €/liter paid by consumers that are not aware of the origins of the EVOO purchased. Compared to the previous group, consumers in this group report having a slightly greater interest in branded products.

The third and last consumer group, accounting for 67.82% of the total sample, encompasses consumers who correctly identify the product's country of origin. This consumer group is largely composed of female consumers and has the highest level of education. These consumers purchase EVOO on promotion less frequently and also report their being highly interested in labeling information, as well as in the product's origins and in branded products. On average, they pay 6.24 € for a liter of EVOO, a price that is higher and statistically different from that paid by consumers belonging to the two groups discussed before.

Focusing on consumers who correctly identify Italian EVOO, reported in the last column of Table 4, they are again female EVOO shoppers with a higher level of education, highly interested in labeling information, in the origins of the product, and in branded products. This consumer group pays on average above 6.60 €/liter for an EVOO product. The data in Table 4 shows that consumers who place importance on information reported on the food label, including brand and the origins of the product, are more likely than others to correctly identify the product's origins, if their level of education is higher than the average.



 indicate no statistically

 significant

 differences

 between the groups (columns)

a, b, c: values with the same letter as the superscript

using Tukey Kramer test, *p* < 0.05.

 based on the pairwise mean

comparison

 across groups

*Nutrients* **2020**, *12*, 2150

Table 4 thus highlights how gender and education is likely to play a role in consumers' ability to identify the origin of the EVOO product. This is consistent with findings from the general psychological theory about consumers and food labels, which identifies female and educated consumers as having greater ability to understand labeling information, as well as being more likely to take informed food choices and pay a higher price for their purchases [37,38].

#### **4. Conclusions**

The present study confirms that the COO label is an effective tool to differentiate food products, in this case EVOO. In particular, our findings point out that, in Italy, the EVOO with domestic origin gains a premium price equal to +35% (+2.13 €/liter) compared to a product labeled as a blend of European EVOOs. Thus, on average, the mandatory COO labeling regulation for EVOO (Reg. (EU) No 29/2012) can be an effective tool for consumers to identify the origin of the product and for producers to differentiate their products. There is, however, a share of consumers in our sample, 19.04% (187), that incorrectly identifies the origin of EVOO purchased and this more often occurs among consumers who report having purchased Italian EVOO. On the one hand, this is likely due to producers' branding strategies, which may hinder the effectiveness of COO information on labels in signaling the origin of the product. On the other hand, COO information, while useful for legal purposes, is not necessarily relevant to all consumers of EVOO, since other extrinsic clues like price may sometimes prevail in orienting EVOO choices.

Findings also indicate that education likely plays a role in correctly identifying the origin of the product, since it enhances consumers' ability to process the information reported on the label of the product, including information about the product's origins. Regarding this last point, government bodies, as well as food manufacturers and retailers, could implement signpost colored labeling to more easily communicate the origins of the product. A simple visual symbol to indicate a product's origins may lower the cognitive effort needed to process the information on the label and so facilitate consumers' identification of the country of origin. The use of a visual, color-based symbol has already been identified as a promising policy tool to support consumers in making healthier food choices at the supermarket. Several studies are offering encouraging findings in support of color-based labels as an effective policy approach to guiding consumers' food purchases [39,40].

The present analysis is not, however, free from limitations. First, it does not explain the mechanism preventing consumers from decoding the origin of the product. This can depend on several additional psychological factors as well as on consumer knowledge related aspects, including the individual olive oil knowledge, which are not captured in the present analysis. Second, the study focuses on a consumer sample interviewed in a single Italian region, and on a single product category, EVOO, to test consumers' ability to correctly identify a product's origin. Future research will therefore aim to address these limitations by exploring the psychological mechanism underlying the incorrect association between the product and country of origin as well as exploring to what extent the olive oil knowledge affects such relation. Furthermore, we will expand the list of products against which consumer's ability to correctly identify the country of origin is tested.

**Author Contributions:** Conceptualization, F.B., L.R. and B.C.d.G.; methodology, L.R. and F.B.; writing—original draft preparation, F.B. and L.R.; writing—review and editing, L.R., D.C. and F.B.; visualization, F.B., D.C. and B.C.d.G.; supervision, B.C.d.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work has been supported by AGER 2 Project, grant n◦ 2016-0174 and by the EU through the Puglia Region: "Avviso aiuti a sostegno dei Cluster Tecnologici Regionali per l'Innovazione"—Progetto: "T.A.P.A.S.S.—Tecnologie Abilitanti per Produzioni Agroalimentari Sicure e Sostenibili"—codice PELM994.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

1. International Olive Council (IOC). EU Olive Oil Figures. 2019. Available online: https://www.internationaloliveoil. org/what-we-do/economic-affairs-promotion-unit/#figures (accessed on 5 June 2020).


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **Factors that Influence the Perceived Healthiness of Food—Review**

#### **Brigitta Plasek \*, Zoltán Lakner and Ágoston Temesi**

Department of Food Chain Management, Institute of Agrobusiness, Szent István University, Villányi Str. 29-43, 1118 Budapest, Hungary; lakner.zoltan@etk.szie.hu (Z.L.); temesi.agoston@etk.szie.hu (Á.T.)

**\*** Correspondence: plasek.brigitta@etk.szie.hu; Tel.: +36-1-305-7100 (ext. 6178)

Received: 22 May 2020; Accepted: 22 June 2020; Published: 24 June 2020

**Abstract:** The interest of consumers is the consumption of healthy food, whereas the interest of food manufacturers is that consumers recognize the produced "healthier" food items on the shelves, so they can satisfy their demands. This way, identifying the factors that influence the perceived healthiness of food products is a mutual interest. What causes consumers to consider one product more beneficial to health than another? In recent years, numerous studies have been published on the topic of the influence of several health-related factors on consumer perception. This analysis collected and categorized the research results related to this question. This review collects 59 articles with the help of the search engines Science Direct, Wiley Online Library, MDPI and Emerald Insight between 1 January 2014 and 31 March 2019. Our paper yielded six separate categories that influence consumers in their perception of the healthiness of food items: the communicated information—like FoP labels and health claims, the product category, the shape and colour of the product packaging, the ingredients of the product, the organic origin of the product, and the taste and other sensory features of the product.

**Keywords:** perceived healthiness; product attributes; healthy food; consumer perception; food packaging; consumer behavior

#### **1. Introduction**

Which food can be considered beneficial to health? Science and consumers answer this question differently. According to certain sources, there is no precise definition of what can be considered healthy food, or else existing definitions are not yet appropriate [1–3]. The understanding of the category of "healthy food" differs even among experts; moreover, some treat the words "healthy" and "nutritious" as synonyms [4,5]. What can be considered healthy for whom depends on gender, age, metabolism, obesity, diseases or sensitivities. A nutritious food product generally considered definitely beneficial to health with several positive effects in case of certain diseases can be harmful for consumers suffering from other diseases [6].

Let us illustrate the effort to define healthy food with two examples. In their article, Zaheer and Bach [7] (p. 1) applied the following definition: "*Per the United States Food & Drug Administration (FDA), Healthy foods are defined as those that are "low in fat, low in saturated fat, contain at least 10% of daily value for vitamins A, C, calcium, iron, protein fiber" and are limited in amount of sodium and cholesterol (USFDA).*" Rodman and his colleagues [5] (p. 83) employed the following definition for their research: *"Foods that provide essential nutrients and energy to sustain growth, health and life while satiating hunger; usually fresh or minimally processed foods, naturally dense in nutrients, that when eaten in moderation and in combination with other foods, sustain growth, repair and maintain vital processes, promote longevity, reduce disease, and strengthen and maintain the body and its functions. Healthy foods do not contain ingredients that contribute*

#### *to disease or impede recovery when consumed at normal levels. (University of Washington Center for Public Health Nutrition (UWCPHN) 2013* [8])".

Dieticians argue that there is no such thing as healthy or unhealthy food; instead, there is only appropriate or inappropriate diet (e.g., [1]). However, since consumers consider certain foods healthy, while others unhealthy, it is important for us to know how they make this distinction. Mai and Hoffman [9] (p. 8) use the term perceived healthiness, which, based on Howlett et al. [10], they define as "*Perceived healthiness is a consumer's expectation of a product's influence on his or her state of health*". The importance of "perceived healthiness" is also supported by the research findings on health claims by Steinhauser and colleagues [11] that the higher the level of perceived healthiness of a product is, the more likely it is that the product will be purchased. All this becomes a factor that also increases the willingness to pay and purchase if it takes into account what influences the credibility of the health benefits of a product [12].

The effects of food on health is a widely researched topic, which gets attention from various aspects, thus our knowledge-base related to its consumer perception is also expanding. In their review, Niebylski and colleagues [13] examined the effects of taxation, subsidies and easy access on the consumption of products considered healthy. According to the results of Provencher and Jacob's review [14] specifically on perceived healthiness, cognitive factors—among them, brand and type of product—have an effect on the perceived healthiness of food, but such features do not influence the choice and intake of food. The reviews of Alba and Williams [15], and Krishna [16] highlight the topic that continues to be researched ever since, namely that the perceived healthiness of food has an effect also on the assessment of the taste of food (e.g., [17,18]). However, research attests that the perception of the healthiness of food is not influenced by one factor only, but by a combination of factors [19,20], so we can state that this topic is highly complex and important both for consumers and companies.

The aim of our literature review is to assemble earlier research and survey the factors that influence consumers in their perception of the healthiness of food.

#### **2. Research Methodology**

In an attempt to access the articles related to the perceived healthiness of food, we employed several search engines—Science Direct, MDPI, Emerald Insight, Wiley Online Library—in our literature analysis. In recent years, numerous review-type articles touching on the topic of healthiness have been published (e.g., [12–18,21,22]), but they only fleetingly mention the issue. The present literature review, however, specifically approaches the topic from the consumers' point of view and so examines the factors which, according to research literature, influence consumer perception of the healthiness of food.

Between 2012 and 2016, several review articles touched on the topic of perceived healthiness of food [13,15,16,21,22] or chose it as their main topic [14]. However, it has remained a widely researched area ever since, so we focused on the time period that followed. Articles published between 1 January 2014 and 31 March 2019 were selected using the following terms:


We looked for the terms in the title, the abstract or among the key words; naturally, because of the way they work, there were slight differences when using the different search engines.

In the I. case, on the ScienceDirect surface we looked for the exact term "perceived healthiness" in quotation marks in the "title, abstract or keywords" fields, while "food" appeared in the "terms" field. On the MDPI page, a very similar method was used, "perceived healthiness"—again in quotation marks—was searched for in the abstract, while "food" was searched for in "all fields". Between the two terms specified in quotation marks, we used the AND relationship to make sure that the search results include both terms. On the Emerald Insight surface, we looked for the complete terms in the abstract and the title, while with Wiley Online Library, in the abstract only, without quotation marks.

In the II. case, on the ScienceDirect search field first "evaluating healthiness", then "evaluation of healthiness" in quotation marks was in the "title, abstract or keywords" field, while "food" was in the terms field. Very similarly to this and point I, on MDPI, the previously mentioned terms were searched for in the abstract, while the term "foods" was searched for in all fields. Just like in the first case, we ran the search with the AND relationship between the search terms. With Wiley and Emerald Insight, we collected the articles in a similar way, looking for the terms in the abstract only and in the title and the abstract, respectively. The search results and the filtering of hits are illustrated in Figure 1.

**Figure 1.** The search hits and the steps of their filtering.

In our analysis, we specifically focused on the products of the food industry, so we did not include research on restaurants, catering establishments, and those on various casseroles, boiled and fried foods served on plates. Moreover, articles on children's dietary habits and on healthy food provision were also not included. The accessed full-length articles were evaluated by two authors (B.P. and Á.T.). Any contested issues were resolved by three authors (B.P., Á.T. and Z.L.).

#### **3. Results**

The main question of our research is what influences consumers in their perception of the positive effects of a given product on health, which, for the sake of simplification, we will refer to as the healthiness of the product. We provide a comprehensive display of the main results of the articles on the topic in Table 1, then we analyse them, reviewing the points of agreement and opposition.


**Table 1.** The articles included in the literature analysis and their main Claims.


**Table 1.** *Cont.*


**Table 1.** *Cont.*


**Table 1.** *Cont.*


**Table 1.** *Cont.*


**Table 1.** *Cont.*


**Table 1.** *Cont.*


**Table 1.** *Cont.*


**Table 1.** *Cont.*


**Table 1.** *Cont.*

Table 1 clearly shows that numerous factors influence consumers when assessing the healthiness of a product. In our literature analysis, we categorized these factors as follows:


The perceived healthiness of a food product is influenced by numerous factors. For bigger clarity, the main points of the research results are summarized in Figure 2.

**Figure 2.** Factors influencing perceived healthiness.

*3.1. The E*ff*ect of the Communicated Information on the Perceived Healthiness of a Product*

When companies provide consumers with information related to nutritional value or to health effects in some way on the product packaging, it has a positive effect on perceived healthiness [26,34–36,47]. At the same time, care must be taken that consumers comprehend this information correctly, so that they do not evoke undesired associations [62], and consumer scepticism related to health claims must also be taken into account [74], as in this case information may even have a negative effect on assessing healthiness.

All this also entails how much of the messages communicated through the product a consumer will comprehend and thus how healthy they will perceive the product. This is supported by the previous knowledge of the consumer, which influences perceived healthiness to a great extent [27,69,78]. Moreover, perceived healthiness of a product is further improved by adding a picture of the product to the communicated information [36] and is also affected by FoP labels and health claims [26,34,38,53]. Although several studies show that FoP labels help consumers choose healthier foods [26,34], due to the diversity of FoP labels, it cannot be clearly stated that their use always helps greatly in increasing the perceived healthiness of the product [53].

The health motivation of the consumer also influences the assessment of the product; Machín et al. [26] maintain that it plays a pivotal role in the way front of package information is used. In contrast, according to the results of Rebouças et al. [47], consumer interest in healthy nutrition does not influence the acceptance and perceived healthiness and nutritional value of the product they examined ("cashew nut beverage").

#### *3.2. The Influence of the Shape and Colour of the Product Packaging*

Research results confirm that the shape and colour of the product packaging influence the perceived healthiness of the product, but from certain aspects the results contradict each other. Whereas Marques et al. [23] maintain that a product is perceived to be healthier in a rounded packaging, other researchers [28,61] claim that consumers perceive a product healthier in angular shaped packaging. A further result related to the shape of the packaging is that packaging resembling a slim human figure is perceived healthier [25].

The influence of the colour of the packaging has also been examined by several researchers. According to the results of Marques et al.'s research [23], buttered products were perceived healthier in a red and yellow coloured packaging. The effect of the colour red is mentioned by several other studies. W ˛asowicz et al. [66], along with yellow, green and blue, mention the colour red as a colour referring to health. At the same time, Reutner et al. [71] assert that the colour red can have a significant effect on the refusal of unhealthy foods. Certain colours and hues, however, can imply that the product is less healthy: research participants considered dark glass [45] and colours hinting at artificiality ("heather", "pink", "celadon") [66] to be referring to unhealthy or less healthy products.

Contrasts resulting from the perception of colour can be attributed to the variety of products and their different packaging investigated in the studies; so it is possible that in the case of a buttered product, the red & yellow colour combination found by Marques et al. [23] was perceived healthier, while it can be different for other products. Therefore, when discussing the effect of colour, it is important to acknowledge the influence of product category. Differences between countries are also important; so for example, in Denmark, paler, whereas in the United States, balanced colour tones are more standard on healthy products [28]. Moreover, the effect of colour can differ according to the age of the consumer: for example, with young people, colours have a stronger effect than health messages [32].

#### *3.3. The E*ff*ect of the Ingredients of a Product on its Perceived Healthiness*

Results related to ingredients show that consumers mostly pay attention to the ingredients that nutrition experts emphasize in relation to healthy nutrition. The majority of the research studies we have examined address sodium- and fat content as well as omega-3 content. According to the results of Lazzarini et al.'s [59] research, the fat content of a product is an indicator of perceived healthiness for consumers.

Whether it is Bolognese sauce, frankfurter sausages, or other processed meat products, consumers prefer a reduced sodium- and fat content, so that is how a company can make consumers perceive these products healthier [24,29,39]. Moreover, while there are consumers who rely on the fatand fibre content of the product for its perceived healthiness [64], others ignore the protein-, sodium-, and saturated fat content when making decisions [76].

The other ingredient featuring in numerous studies was omega-3. In several cases, consumers would choose to change an ingredient other than the fatty acid [24], or they did not consider the product suitable for the addition of omega-3 [29]; at the same time, they prefer if omega-3 fatty acid is added to the product rather than if nothing is added [40].

#### *3.4. The E*ff*ect of Product Category on Perceived Healthiness*

Perceived healthiness is also influenced by the product or product category [75,78]; in fact, in Orquin's [78] research, product category emerged as one of the two main factors based on which consumers perceived the healthiness of a product. In the studies, products assigned to different categories according to different criteria were compared. Fenko et al. [61] compared consumer perception of cereal- and buttered cookies, of which consumers perceived cereal cookies healthier. Vasiljevic et al. [70] compared muesli bars and chocolate, and their results show that, regardless of label, consumers perceived chocolate tastier and muesli bars healthier. According to Maehle et al.'s [75] surprising result, consumers are less concerned about the healthiness of the product if they consume it for the nutritional value (utilitarian products), than in the case of products consumed for pleasure (hedonic products).

#### *3.5. The E*ff*ect of Organic Origin on Perceived Healthiness*

Results of numerous studies have confirmed a positive effect of organic origin on the perceived healthiness of a product [42,45,48,51,68]. In addition, organic origin also facilitates the understanding of the communication of "healthy food" [5]. Health-conscious consumers also tend to show openness towards bio foods and generally ignore the health-related messages of functional foods [46].

#### *3.6. The E*ff*ect of the Sensory Features of the Product on Perceived Healthiness*

The sensory features of the product also play a role in its perceived healthiness. The taste and other sensory features of the product may dominate over the perception of healthiness [41,52,79], and if the sensory features of the product do not satisfy the consumer, then communicating the nutritional value is not enough to make the product accepted [44].

#### **4. Discussion**

The aim of our literature analysis was to explore the factors that influence the perceived healthiness of food products. Numerous studies set out to discover what influences the perceived healthiness of individual products in the time period we focused on. In the present article, we only considered research results related to foods.

Based on the research results, we identified six categories that influence perceived healthiness of a product: the effect of the communicated information, product category, the shape and colour of the product packaging, the ingredients of the product, the organic origin of the product and the taste and other sensory features of the product.

The effect of the communicated information clearly influences perceived healthiness; at the same time, previous knowledge clearly affects how this information influences perception. Product category is a main factor in the perceived healthiness of a product. In recent years, a diverse range of product categories has been tested, which makes generalizations difficult.

The most numerous contradictory research results were related to the shape and colour of the product packaging, which calls for further investigation. Research results are ambiguous concerning whether angular or rounded packaging is more suitable to communicate healthiness. One of the most researched colours is the colour red. Nevertheless, results related to the colour red do not point in the same direction.

Research results related to the ingredients of the product confirm that reducing the sodium-, sugarand fat content increases consumer acceptance of the improved product in terms of health; at the same time, research does not give a definite answer regarding consumer perception of the possible enriching ingredients.

The organic origin of the product positively influences perceived healthiness. Health halo effect emerged in several studies in connection with bio products.

Basically, the taste and other sensory features of the product dominate over perception of healthiness. A common result of the examined studies showed that the unsuitability of sensory features cannot be balanced out by favourable perceived healthiness.

Our collected results and their juxtaposition can help with the proper planning of product development and marketing communication, and they also raise further research questions related to the inconsistent results. Our conclusions can serve as a baseline from several aspects when devising packaging of a new product. They can help with the proper design of the packaging, both in terms of shape and the used colours, and with choosing the right FoP labels. The choice of the labels used on the packaging requires special care. The type of health claim communicated by the company has to be considered carefully, provided that the use of a health claim is effective in the first place. At the same time, it is also important to consider that communicating the different ingredients may be an effective method to reach its goal.

Within the categories, we have found several conflicting results, as well as unanswered research questions, which call for further research. The most important aim of further research may be to gauge the effect of the discovered aspects relative to each other, even comparing all the aspects.

Further research may also aim at clarifying the emerged controversial results, as the research results are not uniform for example in connection with the shape of packaging and colours that evoke a healthy feeling, keeping in mind that these factors may change according to product category. Apart from clarifying discrepancies, further research may take on the task of testing specific features on different food products. Treating the constructed system as a complex entity, it is worth examining whether a different colour and shape of packaging is justified to communicate the health benefits of each food.

#### **5. Limitations**

In the course of our literature analysis, we encountered several barriers that have to be taken into account when evaluating the results. There has not been a review article on the topic since 2016, even though several new studies have been published since then. As we reviewed only the 2014–2019 time period, we can only report the results of the most modern research. The surfaces used for data collection are also important to mention: during the research, we had no access to the surfaces covering the whole literature, therefore we chose the above described search surfaces, where we could access the full-length articles. Our options were limited by the year-to-year changes in the agreement between the Hungarian Electronic Information Service National Programme (EISZ) and Elsevier [81].

**Author Contributions:** Conceptualization, B.P. and Á.T.; methodology, B.P.; writing—original draft preparation, B.P.; writing—review and editing, Á.T.; visualization, B.P.; supervision, Z.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** The Project is supported by the European Union and co-financed by the European Social Fund (grant agreement no. EFOP-3.6.3-VEKOP-16-2017-00005).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **The E**ff**ect of an Online Sugar Fact Intervention: Change of Mothers with Young Children**

#### **Yi-Chun Chen 1,2,\* , Ya-Li Huang 2,3,4, Yi-Wen Chien 1,5 and Mei Chun Chen <sup>1</sup>**


#### Received: 21 April 2020; Accepted: 19 June 2020; Published: 22 June 2020

**Abstract:** Research indicates that high sugar intake in early childhood may increase risks of tooth decay, obesity and chronic disease later in life. In this sugar fact study, we explored whether an online intervention which focused on comprehensive and useful information about nutrition labels impacted mother's choice of low sugar food. The intervention was developed on the basis of the theory of planned behavior. In total, 122 mothers were recruited. Mothers were divided into an online-only group and a plus group. Knowledge of sugar and nutrition labels, behavioral attitudes, perceived behavioral control, behavioral intentions and behavior towards purchasing low-sugar products with nutrition labels were collected. After the intervention, both groups exhibited significantly enhanced sugar and nutrition label knowledge, perceived behavioral control, behavioral intentions and behavior. Compared to the online-only group, knowledge, perceived behavioral control and behavior of the plus group significantly improved. After the intervention, about 40% of the plus group and 80% of the online-only group still did not know the World Health Organization (WHO) sugar recommendations. Understanding sugar recommendations and using nutrition labels are crucial to help people control calorie and sugar intake. Further research with a larger sample is warranted to evaluate the effects of the intervention on long-term changes in shopping behavior. More efficient and convenient nutrition education is required to increase public awareness of sugar recommendations and help people control calorie and sugar intake.

**Keywords:** online nutrition intervention; theory of planned behavior; nutrition labels; sugar; consumer behavior; consumer attitude; consumer perception

#### **1. Background**

Being overweight and obese increases the risks of many health problems, including diabetes, heart disease and certain cancers [1–3]. An examination of the 2015 data of the World Obesity Federation reveals that overweight rates of children in Taiwan were the highest in Asia [4]. A long-term follow-up study in Taiwan found that approximately 90% of young children consume sugary drinks and snacks once per day, and one-third of 5 year-old children more than 10% of their caloric intake from refined sugar [5]. Studies have confirmed that sugar promotes a high energy balance and children who consume more sugar have higher obesity rates than those who consume less sugar [6–8]. Thus, the World Health Organization (WHO) strongly recommends that it is good to reduce the sugar intake to <10% of total energy intake for both adults and children [9].

Products with low- or no-sugar-related claims, such as "sugar free", "no added sugar" and "reduced sugar", may be particularly appealing to parents who want to manage their child's sugar intake [10,11]. However, a Canadian study found that half of 3048 prepackaged foods with sugar-related claims contained excessive sugar, and a greater proportion contained sweeteners than did products without such claims [12]. A survey in Taiwan also found that more than 90% of popular snacks and drinks with no-added-sugar claims consumed by children were high in sugar [13]. A study in Australia and New Zealand found that 28% of consumers misunderstood the meaning of the claim of "no added sugar", believing that products with such a claim contained no sugar [14]. In addition, 95.7% of mothers know that excessive sugar intake increases future health risks in children, but only 21.9% know the WHO sugar recommendations [15].

Nutrition labeling is an important tool to help people choose healthy foods. A previous study has discovered that although most consumers trust nutrition labels, they perceive that the information on them is difficult to understand and confusing, including information on recommended daily allowances, percent daily values and servings [16]. Because of limited time, consumers normally only read one or two of the facts on nutrition labels, such as calories and fats [17]. The Health Information National Trends Survey in the US suggested that although most people have difficulty interpreting nutrition labels, they cannot effectively utilize such facts to make informed dietary choices they possess insufficient reading comprehension and calculation skills [18]. Education can help the public improve those skills required to understand nutrition labels, thereby allowing consumers to effectively purchase suitable foods according to nutrition labels [19]. A study in the US applied a multimedia intervention to participants in an experimental group and significantly improved their comprehension of nutrition labels [20]. Another study in the US demonstrated that online courses helped the public improve skills required to effectively use nutrition labels to buy healthy foods [21]. In brief, effective education can improve people's comprehension of nutrition labels, thereby enabling the public to select healthy foods.

Despite the desire of parents to maintain their children's optimal health, they may not provide healthy foods to their children if they lack nutrition knowledge. A survey of Taiwanese mothers with young children demonstrated that the less sugar-related knowledge they had, the more positive their attitude was toward no-added-sugar infant cereal and the higher their purchase intention was for these infant cereals [15]. Moreover, some parents who lack this knowledge consider providing healthy diets to their children a challenge [22]. Under such circumstances, desirable nutrition education can improve parents' beliefs and behaviors concerning their children's diets, allowing them to provide a healthy diet for their children. The theory of planned behavior (TPB) has been widely adopted in studies that explore or predict various health behaviors as well as those that explore health promotion. This theory posits that behavioral intentions are a pivotal antecedent that affects actual behaviors; furthermore, attitudes toward a behavior, subjective norms and perceived behavioral control are possible controlling factors that affect behavioral intentions [23]. Increasing one's relevant knowledge and developing relevant skills help to change attitudes toward a specific behavior and in turn enhance individuals' intentions to perform such a behavior, thereby encouraging individuals to improve that behavior. A study of Australian mothers with preschool-aged children demonstrated that perceived behavioral control and intention were positively associated with mothers' healthy feeding behavior perceived behavioral control was the only variable positively associated with the mothers' perceptions of their children's fruit and vegetable consumption [22]. A study in South Australia indicated that a parent-focused nutrition intervention can affect maternal feeding practices, which reduced growth-related indicators of future obesity risk in young children [24]. Similarly, an Australian study regarding breakfast consumption indicated that people's attitudes and perceived behavioral control significantly affected their intention to eat breakfast; such an intention affected an individual's breakfast consumption [25]. A podcast-based research study in the US determined that listening to a podcast about omega-3 fatty acids in the grocery store enhanced customers' attitudes and perceived behavioral control toward purchasing foods rich in omega-3 fatty acids [26]. According to the aforementioned studies, it was assumed that nutrition educational interventions based on the TPB can promote the public's health behaviors.

In Taiwan, nutrition labeling regulations regarding sugar content apply to the products manufactured after July 2015 [27] and sugar recommendation was available on 2018 [28]. Encouraging people to use nutrition labels is crucial. Traditionally, interventions targeting parents with young children which required face-to-face classes had low attendance and high dropout rates [29,30]. Online interventions are an efficient and cost-effective way to provide nutrition education. Previous studies confirmed that online nutrition educational programs should be considered to expand outreach and decrease barriers to attending traditional face-to-face classes [31,32]. According a 2017 Taiwanese Internet usage survey, there was an 80% Internet usage rate across the nation [33]. Therefore, online education is feasible in Taiwan. In the present study, the TPB was used to develop an online intervention program to enhance mothers' use of nutrition labels to buy low-sugar foods for their children.

#### **2. Methods**

#### *2.1. Design and Participants*

This Sugar Fact intervention was a quasi-experimental trial, which was conducted from December 2017 to August 2018 in Taiwan. Online videos and a small-group discussion were used to encourage mothers to use nutrition labeling to buy low-sugar foods for their children. The intervention was designed for the senior high school level because 95% of Taiwanese women aged 25–44 years have at least a senior high school education [34]. Mothers who live in Taiwan, communicate in Chinese, had a child aged 1–6 years, and were their child's primary caregiver and the family's food purchaser were eligible for the study. Various parenting social networks (e.g.; BabyHome, a mothers' groups on Facebook and BabyMother on a bulletin board system) were approached and those who agreed to distribute or advertise the study posted a link to the online recruitment questionnaire on their network. Participants were classified into an online-only group or a plus group according to their intentions. An official Line account (a social networking app popular in Taiwan) was used to contact and follow-up participants. A notification message by Line was sent once per week to remind participants to continue the intervention. The intervention for the online-only group consisted of watching two online videos. The intervention of the plus group included watching two online video and participating in one small-group discussion.

In total, 236 mothers were recruited, and 185 mothers were enrolled in the intervention. Finally, 90 mothers in the online-only group and 32 mothers in the plus group completed the intervention and posttest questionnaire (Figure 1). The completion rate was 62.1% for the online-only group and 80% for the plus group. Participants who completed the intervention received a commercial voucher as an incentive (NT\$100 for the online-only group and NT \$300 for the plus group; about US \$3 and \$9, respectively).

#### *2.2. Ethical Considerations*

The study was approved by the Taipei Medical University—Joint Institutional Review Board (N201711059) and written informed consent was obtained from all mothers.

#### *2.3. Developed Educational Intervention*

The educational intervention was developed based on results of a previous study [15]. Considering the difficulty of attending classes by mothers with young children [35], online video courses were used. The purpose of the online video session for mothers was to increase their positive attitudes and perceived behavioral control of using nutrition labels to buy low-sugar foods for their child (Figure 2). The purpose of the first video was to increase the mother's perceptions of "sugar and health" and "sugar and nutrition". The purpose of the second video was to improve the mothers' understanding of "sugar recommendations", "sugar-related claims" and "nutrition labeling on food packages". All of

the contents of the videos were reviewed by three public health nutrition professionals. Five mothers eligible for recruitment were included in a pilot intervention. They were asked to watch the first version of the two videos and give comments. The two videos were uploaded on YouTube after being revised.

**Figure 2.** Intervention concept.

For the online-only group, the nutrition intervention included two 15 min online video sessions. For the plus group, the nutrition intervention included two 15 min online video sessions and one small-group discussion led by the researchers (for 2–3 h). Additional educational materials included a booklet for the online videos and a pamphlet for the small-group discussion. The aims of the small-group discussion were to help mothers clarify the content of the online videos and improve their perceived behavioral control structure. The small-group discussion session focused on barriers that mothers encounter when selecting low-sugar foods and controlling their children's sugar intake. From January to May 2018, the small-group discussion sessions were held 10 times. There were three to five mothers in each small-group discussion. In total, 32 mothers participated in small-group discussions, and all group discussions were recorded. Several topics on a list were discussed to gain an in-depth understanding of participants' reactions to the online video courses and barriers when the mothers practiced their skills of choosing low-sugar foods for their children. The mothers were also allowed to practice using nutrition labels to choose low-sugar foods.

#### *2.4. Questionnaire*

A theory-based questionnaire containing the mother's demographic characteristics was used to collect data. All participants completed the questionnaire before the intervention, and they also completed a posttest questionnaire within 2 weeks of finishing the intervention.

The theory-based questionnaire was prepared by reviewing other questionnaires applied in similar studies [15,26,36,37]. Three nutrition and statistical professionals reviewed and revised the questionnaire. The questionnaire was also tested by 15 mothers who had a child aged 1–6 years. Questionnaire items were tested for consistency and comprehensibility. The questionnaire included the following sections; demographic characteristics of the mothers, including age, education (≤high school, undergraduate or ≥graduate), medical background (mother were health professional, such as medical doctors, dietitians or nurses: yes or no), parity (child was firstborn or non-firstborn), household income (≤NT \$50,000 or >NT \$50,000) and the child's age. Second, data on the mother's behaviors and intentions were collected. Behaviors: I used nutrition labels to buy low-sugar foods for my child/children during the past week? (always, usually, sometimes or seldom); Behavioral intentions: I am going to use nutrition labels to buy low-sugar foods for my child/children in the coming week. The third part was about the mothers' sugar-related knowledge. It included "sugar and health", "sugar and carbohydrates", "daily sugar recommendation", "no-added-sugar claims" and "nutrition labels". The last part included mother's attitudes, subjective norms and perceived behavioral control. There were four questions about attitudes (e.g., I believe that using nutrition labels to buy low-sugar foods for my child/children is very important), three question about subjective norms e.g., My family members expect me to use nutrition labels to buy low-sugar foods for my child/children) and four questions about perceived behavioral control (e.g., It is difficult for me to use nutrition labels to buy low-sugar foods for my child/children). A Likert scale was used to score the data collection instruments. In order to avoid neutral feedback, this section used a Likert 6-point scale divided into "very disagree", "disagree", "disagree a little", "consent a little", "agree" and "very agree". These were scored 1–6 points, respectively [38].

#### *2.5. Data Analysis*

A statistical software package (SPSS, Chicago, IL, USA) was utilized. The Kolmogorov-Smirnov test was used to examine the normal distribution of all continuous variables (the test of normality was present at Supplementary Table S1). Nonparametric statistics were used due to most variables did not match normal distribution. Baseline data for demographic characteristics and TPB variables of the two groups were analyzed using a chi-squared and Mann-Whitney U test. A Wilcoxon signed-rank test was used to study changes before and after the educational intervention and a Mann–Whitney U test was used to evaluate the mean of changes and compare the mean of study variables in the two groups. Spearman's rank correlation was used to examine the correlation among the difference of TPB variables. Statistical significance was set as *p* < 0.05.

#### **3. Results**

Table 1 shows the demographic characteristics of participants. There was no significant difference in demographic characteristics between the two groups (*p* > 0.05). The mothers' average ages were 35.3 ± 4.5 years in the online-only group and 34.9 ± 4.5 years in the plus group. The children's average ages were 2.7 ± 1.6 years in the online-only group and 2.9 ± 1.6 years in the plus group. Most mothers of the two groups had more than one child, had a university/college degree, worked full-time. The average family monthly income was NT \$30,000–50,000 and more than 80% of the mothers had no medical background, such as medical doctors, dietitians or nurses.


**Table 1.** Demographic characteristics of participants by the intervention condition 1.

<sup>1</sup> Data are presented as the number (percentage) or mean ± standard deviation; <sup>2</sup> participants of the plus group finished online videos and a group discussion; <sup>3</sup> the average exchange rate in 2018 was US1.00 ≈ New Taiwan (NT) \$30.

Table 2 presents the results after the educational intervention, including changes in nutrition knowledge and TPB variables in the two groups and differences between the online-only and plus groups. After the educational intervention, sugar and nutrition label knowledge in both groups significantly improved (*p* < 0.001). The change in the plus group was greater than that in the online-only group (4.3 ± 2.4 vs. 2.0 ± 2.3, *p* < 0.001). No significant difference was observed between the groups regarding the mean scores for behavioral attitudes, perceived behavioral control or subjective norms before the intervention. Mean changes in scores for behavioral attitudes, perceived behavioral control, and subjective norms were significant in the plus group (*p* < 0.05), but the change in behavioral attitudes in the online-only group was not. The mean change in perceived behavioral control in the plus group was greater than that in the online-only group (0.3 ± 0.7 vs. 0.7 ± 0.9, *p* = 0.026). After the intervention, the mean changes in intentions and behavior significantly improved in both groups. The improvement in behavior for the plus group was significantly greater than that of the online-only group (1.8 ± 1.7 vs. 0.8 ± 1.7, *p* = 0.005).


**Table 2.** Changes in knowledge and theory of planned behavior (TPB) before and after the intervention in both groups. 1.

<sup>1</sup> Data are presented as mean and standard deviation (SD); <sup>2</sup> participants of the plus group finished online videos and a group discussion; <sup>3</sup> difference between the online-only group and plus group; <sup>4</sup> difference between the before and after scores; \* *p* < 0.05; \*\* *p* < 0.001 by Mann-Whitney U test and Wilcoxon signed-rank test.

Figure 3 presents the correlation among changes in TPB constructs. The figure indicates significant predictive associations with mothers' intentions to use nutrition labeling to buy low-sugar food and the positive associations between mothers' intentions and behaviors.

Table 3 presents changes in the mothers' sugar-related knowledge in both groups after the intervention. The correct rate of all questions had increased in the plus group, but not in the online-only group. Both before and after the intervention, more than 90% of mothers in both groups knew of the association of sugar with health. After the intervention, the change in the correct rate of "No-added-sugar claim" had increased 25% in the online-only group and more than 50% in the plus group. In particular, the correct rate of "comparing sugar contents of foods with 'no added sugar' and those without 'no added sugar' on the label" had increased 68.7% in the plus group. In the small discussion group session, most mothers expressed that they really cared about the sugar content of foods and they would buy the food with "no added sugar" claim, since they did not realize the difference between "sugar-free" claim and "no added sugar" claim. During the small-group discussion session, they practiced reading several child food packages to understand the difference and found it was important to read the sugar content on nutrition labels.

**Figure 3.** Constructs of the theory of planned behavior in this study (*n* = 122).



<sup>1</sup> Data are presented as the number (percentage); <sup>2</sup> a popular yogurt drink; T: True; F: False.

After the intervention, the correct rates of "sugar and nutrition" exceeded 80% in both groups, and the correct rates of "brown sugar is healthier than white sugar" were still lower than 40% in both groups. The "sugar recommendation" had the lowest correct rates in both groups. After the intervention, changes in the correct rate of "calories from daily sugar intake" was 59.4% in the plus group and only 15.6% in the online-only group. Even after the intervention, correct rates of the other two questions about "sugar recommendations" was still <40% in both groups. During the small-group discussion, most mothers expressed that they knew that the recommended sugar intake differed by age, but it was difficult to memorize different sugar recommended gram for different ages.

After the intervention, the correct rates of "nutrition labeling" were >80% for the online-only group and >75% for the plus group. During the small-group discussion, some mothers expressed confusion about the information presented by nutrition labels. Mothers were confused why some products contained only one serving while other products contained two servings (Figure 4) and they had trouble calculating sugar contents or calories for all products and comparing or choosing products.


**Figure 4.** Nutrient labeling information for knowledge part E. nutrition labeling.

#### **4. Discussion**

#### *4.1. Application of the TPB in a Low-Sugar Educational Intervention*

This study employed the behavior attitude and perceived behavior control elements of the TPB to design an online educational intervention program. The study results revealed that at one week after participation in the intervention program, both the plus group and online-only group showed increases in intentions and behavioral frequencies toward using nutrition labels to buy food products with lower sugar content for their children. Moreover, changes in behavioral attitudes and perceived behavioral control were positively associated with mothers' intentions. According to the TPB, behavioral attitudes and perceived behavioral control are crucial factors influencing the behavioral intentions and actual behaviors of people [23]. By applying this theory in the context of this study, it was possible to use an educational intervention to increase the health-related behavioral intentions of participants, which in turn promoted positive behavioral attitudes and ultimately increased actual positive health behaviors. The results of this study supported that educational interventions could influence a person's behavioral intentions; the stronger the personal behavioral intention is, the greater is the frequency of health behavior execution. Approximately 10% of Taiwanese women aged 25–44 years have at least a master's-level education [34]. Compared with women's education level in Taiwan, more participants in this study, particularly in the plus group, had a master's-level education. Future interventions should be modified to enable more mothers to benefit from the online class.

#### *4.2. Advantages of the Online Nutritional Education Intervention*

In considering both the pervasiveness of Internet services and mothers with young children may have difficulties attending face-to-face learning sessions, this study used online videos to deliver the intervention program; some mothers who were interested could participate in the small-group discussion sessions. The online learning program seemed to be a suitable educational intervention tool for the mothers who participated in this study were highly educated (90% had greater than a college educational level). A study conducted in the US revealed that online teaching media (e.g., emails and websites) are low-cost intervention tools for nutrition education, and more important, these online teaching media can enhance the willingness of learners to participate [39]. The completion rate of this study was 62.1% for the online-only group and 80.0% for the plus group. This finding indicated that although online videos could increase the mothers' convenience to participate in the intervention program, some mothers still had insufficient time available to finish watching two 15-minute online videos. As for the plus group, the participants may have had stronger intentions and motivation to learn and were therefore willing to overcome obstacles to complete the intervention program. The online videos used in the present study should be modified to increase mothers' participation rates; for instance, the videos should be split into shorter, more-manageable videos, and also the content could be modified according to results of the group discussions in the present study.

This study revealed that compared to the online-only group, the plus group consciously enhanced their perceived behavioral control and frequency of using nutrition labels to buy low-sugar food products after participating in the intervention program. This indicated that the small-group discussions can improve the ability of mothers to read and use nutrition labels. A study conducted in the UK revealed that group interventions are useful for teaching mothers about child feeding techniques and helping them achieve desired dietary targets for their children [40]. In another study analyzing parents as nutritional education participants, group discussions had a positive role in remedying parents' difficulties with feeding their children and improving their feeding techniques [41]. Although the small-group discussions are a useful intervention tool, it was difficult to identify a group meeting time that suited three or more participants in this study, because more than half of the participants were working mothers. Therefore, eight participants (20%) of the plus group could not even complete one required face-to-face discussion session. Because an increasing number of web conferencing software programs, such as Skype business, are becoming popular, future studies may use this kind of software to conduct the small-group discussions and increase the participation rate of mothers.

#### *4.3. Improvement of Mothers' Knowledge of Nutrition Labels*

Results of the current study revealed that after the intervention, the knowledge, behavioral attitudes, perceived behavioral control, intentions and behaviors of the plus group participants had significantly improved. Specifically, their knowledge, perceived behavioral control, and behaviors significantly improved compared to those of the online-only group. Regarding the plus group, in addition to watching the online educational videos, they participated in a small-group discussion to clarify doubts regarding the videos and engage in nutritional-label-use practices. Both their behavioral attitudes and perceived behavioral control in terms of using nutrition labels to shop for low-sugar foods improved after the intervention. A previous study found that nutrition knowledge can improve the ability of participants to shop for healthy food items [42], while another study demonstrated that knowledge influences purchasing behaviors through its influence on attitudes [43]. Berg et al. [44] asserted that behavioral attitudes and perceived behavioral control are crucial factors that influence personal food choice intentions. A study in Iran that employed the TPB to increase participants' fruit and vegetable intake revealed that compared to pre-intervention levels, participants' post-intervention behavioral attitudes, perceived behavioral control, and behavioral intentions significantly improved [45]. However, the researchers asserted that even if a person had positive behavior attitudes toward such food items, if he or she did not possess the ability to implement autonomous control, then they would still be unable to modify their behavior [45]. Results of a study about fruit and vegetable intake also confirmed that perceived behavioral control is a fundamental factor in predicting food choice intentions and behaviors [46]. Results of the present study revealed that after the educational intervention was administered, mothers in both groups had a significantly improved ability to consciously shop for low-sugar foods using nutrition labels. This indicates that enhancing perceived behavioral control is an important factor in increasing participants' positive nutrition-label-use behaviors.

Although the present study did not incorporate subjective norms into the design of the educational program, the subjective norm scores of both groups significantly increased from the pre- to post-intervention stages. It is possible that during the study period, participants may have engaged in additional discussions related to sugar, health, and/or nutrition labels with their family and friends, which subsequently led to an increase in support from friends and family with regard to controlling the sugar intake of their children. This would ultimately have led to an increase in subjective norm scores. To further increase the support of family and friends towards participants, in future studies, intervention sessions could be designed in which the participants could engage in discussions with their friends and family.

#### *4.4. Understanding No-Added-Sugar Claims and Changing Behavioral Attitudes*

In terms of behavioral attitudes, this study demonstrated that before the educational intervention was implemented, fewer than 50% of the online-only group participants and fewer than 40% of the plus group had correct perceptions about the "no-added-sugar" claim, whereas more than 50% of the total participants believed that food items with the "no-added-sugar" label contained less sugar than food items without it. During the group discussions, several participants reported that they would buy the "no-added-sugar" food items for their children, but they did not know considered the actual sugar content. A survey conducted in Taiwan reported that more than half of mothers believed that infant cereals with "no-added-sugar" claims contained less sugar than products without such claims [15]. After the educational intervention was implemented, the mean percentage of correct responses improved by 25.0% for the online-only group and 56.3% for the plus group. A study conducted in the UK revealed that approximately 40% of responding parents took the initiative to search for food items with the "no-added-sugar" label when shopping for food for their children [47]. Furthermore, a snack and beverage marketing survey conducted in Taiwan also revealed that more than 90% of children's popular snacks and beverages with "no-added-sugar" or "no-artificial-sweeteners" claims actually had high sugar contents [13]. If parents do not possess sufficient knowledge about sugar claims, they may be easily misled by the "no-added-sugar" claims and end up choosing food items that actually have high sugar contents. In the future, stricter sugar claims regulations should be implemented to prevent the public from being misled. Additionally, the present study revealed that mothers' incorrect perceptions could be effectively modified using online videos and/or group discussions. Furthermore, group discussions can enhance mothers' positive attitudes toward paying attention to nutrition labels. To promote the effects of online videos, future interventions could include explanations (with relevant examples) about "no-added-sugar" claims and sugar contents of common foods.

#### *4.5. Awareness of Sugar Recommendations*

In this study, although 90% of mothers in this study had a college education, after the intervention, about 40% of the plus group and 80% of the online-only group did not know the WHO sugar recommendations. These findings are consistent with the previous study that even highly educated people are rarely aware of sugar recommendations [48]. An online survey in Northern Ireland found that even about 90% of participants were college level, there were still 65% of them were unaware of the WHO guidelines for sugar intake, only 4% of respondents correctly classified sugar and artificial sweeteners [48]. Another online survey among Canadian young people also found that only 4.8% of participants correctly identified Canadian recommendations for sugar intake [49]. The present study also found that during the small-group discussion, mothers mentioned the difficulty of remembering WHO recommendations or were confused about the different acceptable daily sugar intake for different age groups. Understanding sugar recommendations and using nutrition labels to choose foods are important to help people control their sugar intake. Therefore, more nutrition education or campaign are needed to raise people's awareness of the sugar recommendations and control their sugar intake.

#### *4.6. Di*ffi*culties with Nutrition Labels*

The study found that after the intervention, about 15% of both groups still had difficulty interpreting nutrition labels. During the group discussion, the mothers expressed difficulties in using nutrition labels to calculate sugar contents and selecting foods with low-sugar contents. Mothers were confused why some products contain only one serving, but other products contain more than two servings thus they had trouble using nutrition labels to compare or determine sugar contents of products. An experiment about the efficacy of nutrition label formats found that the percent daily value information could help consumers correctly identify the relative amount of total sugar, and the added sugar information could help people correctly identify the relative amount of added sugar in a food [49]. A previous web-based label-reading training intervention also found that web-based practice led to improvements in nutrition label-reading skills; however, consumers may not bother using nutrition labels if reading them is too difficult or time-consuming [21]. Previous studies showed that graphic symbols, such as a traffic light, were easily understood by people with different education or income levels [50–52]. In Taiwan, the percent daily value of sugar or graphic symbols are not available. Further studies are needed to develop the simple label to help parents to choose the low sugar food.

#### *4.7. Limitations and Further Developments*

First, although the online video course made it convenient for mothers with young children to participate, it is not possible to confirm whether all the mothers watched and understood all of the online videos. Second, the importance of knowledge on sugar and intention to use food labels for mothers' food purchasing decisions remains unclear. Third, because the groups were assigned by participant preference, the motivation of those in the plus group may have been higher than that in the online-only group. This may have been a factor causing differing results between the two groups.

The results suggest that an online intervention increases mothers' intentions to use nutrition labeling to buy low-sugar foods for their children. Online video s should be improved using the feedback from small-group discussions; therefore, more mothers can benefit from online classes. Furthermore, more research with a larger sample size is warranted to evaluate the effects of the intervention on long-term changes in mothers' real shopping behaviors. Purchases over one week after the intervention may be an inadequate reflection of change. Therefore, an examination of purchases over time, such as one or two months later, is required.

#### **5. Conclusions**

This study demonstrated that after the online sugar educational intervention, mothers in both the online-only and plus groups exhibited increased intentions and behavioral frequencies of using nutrition labels when selecting low-sugar food products for their children. In addition, after the intervention, approximately 40.0% of the plus group and 80% of the online-only group still did not know the WHO sugar recommendations. Awareness of the WHO sugar recommendations and using nutrition labels to select foods are instrumental in helping people control their calorie and sugar intake. Therefore, more efficient and convenient nutrition education is required to increase public awareness of sugar recommendations and help people control their calorie and sugar intake.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6643/12/6/1859/s1, Table S1: Test of normality for knowledge and constructs of TPB.

**Author Contributions:** Y.-C.C. conceived and designed the intervention; Y.-C.C. and M.C.C. performed the intervention; Y.-L.H. and Y.-W.C. analyzed the data; Y.-C.C. was responsible for data interpretation writing original draft preparation. The datasets generated and/or analyzed during the current study are not publicly available because new manuscripts are in preparation. They can be available from the corresponding author on reasonable request. All authors have read and agree to the published version of the manuscript.

**Funding:** This work was supported by the Ministry of Science and Technology [Grant Number MOST 107-2511-H-038 -001 -MY2].

**Acknowledgments:** The author wishes to acknowledge the help of Shwu-Chen Kuo and Shwu-Huey Yang in commenting on an early draft of the manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
