2. Methods and Materials
Labeling data were collected from pre-packaged foods and drinks sampled in five EU countries: the UK, the Netherlands, Germany, Slovenia and Spain. The selection of countries aimed at a geographical spread across Europe. Three types of stores were selected in each country (with a total of 15 stores overall), with the aim of covering a range of different retail outlets, in order to map consumer exposure to food and drink products across a variety of shopping places: a large supermarket/national retailer, a discounter and a neighborhood store. Differences in the penetration of various food labeling information in different retail outlets were reported recently [19
], confirming the rationale of the approach used in this study. Stores were selected based on the accessibility of their network within the respective country, as well as comparable store characteristics (e.g., store size). A food or drink (henceforth only referred to as “food”) was defined as a single item available for sale in the selected store. This definition meant that the same food in different sized packaging could be included in the survey, on the basis that the packaging for the same food in different sized packages may carry different health-related claims. Within-country, exact duplicates, however, were removed from the database.
Only pre-packaged foods were considered. The EU Regulation on the provision of food information to consumers (EC) 1169/2011 defines a pre-packaged foodstuff as “any single item for presentation as such to the ultimate consumer and to mass caterers, consisting of a foodstuff and the packaging into which it was put before being offered for sale, whether such packaging encloses the foodstuff completely or only partially, but in any case in such a way that the contents cannot be altered without opening or changing the packaging” [25
2.1. Data Collection
In each of the five countries, approximately 400 foods were sampled i.e., purchased using a randomized sampling approach. A power calculation was conducted to estimate the sample size required. An assumed 50% prevalence rate for NHC was selected in order to ensure an adequate sample size, for the detection of NHC prevalence with 5% precision. The sample size needed for an estimation of prevalence is at maximum when the measured prevalence is 50%. Hence, after adjustment for a finite population and assuming a prevalence rate of 50% for NHC, 400 foods for each country would produce confidence intervals of ±5%, sufficient for distinguishing a 10% difference in prevalence between countries.
In each of the five countries approximately 250 foods were sampled from the supermarket/national retailer, 75 foods were sampled from the discounter and 75 from the neighborhood store. Details of the whole sample are provided in Table 1
Retailer’s store/stock lists were converted to Microsoft Excel spreadsheets, if not already in that format. The following types of product were then excluded: (a) non-food items; (b) food supplements; (c) alcoholic drinks; and (d) unpackaged foods. The remaining foods were assigned an ID number and the appropriately sized sample randomly selected using the RAND function in Excel. The sample of foods was then purchased.
For four of the stores, a store/stock list could not be obtained. In these cases a floor plan was created which mapped the layout of the store (including the location of promotional stands or other non-aisle displays). Each section/aisle was then assigned a number and the number of foods in each section/aisle was estimated using a tally counter. The total number of foods in the store was then estimated and each food location assigned an ID number. Again, an appropriately sized list of food locations was randomly generated using the RAND function in Excel. The researchers then returned to the store and purchased the foods using the list of locations.
Exclusion of products also occurred post sampling. Post-sampling exclusion occurred for unpackaged products selected inadvertently. Foods were selected and purchased in the same time frame (July–August 2013) for all countries. For perishable foods, the packaging was removed. Photographs were taken of all sides of packages carrying at least one claim.
Two pilot studies were carried out: one in Germany at a large supermarket and the other in the UK at a neighborhood store. Feedback from the pilots resulted only in minor changes to the final protocol for data collection.
2.2. Data Extraction
Labeling data were taken from packaging and entered into a database. All NHC were identified and characteristics of the claims were recorded, i.e., whether the claim was worded, pictorial or a combination (both worded and pictorial) and whether it qualified as a symbolic claim. Worded claims were recorded verbatim and translated into English. Pictorial claims (or claims which were a combination of words and picture(s)) were briefly described. The type of health symbol was noted.
Other labeling and food composition data were also collected (e.g., whether the package had front-of-pack nutrition labeling and information about nutrient content from the nutrient declaration).
Additional characteristics of the NHC recorded included their position on pack and the number of times they appeared on one packet. For health claims, it was also noted whether it was a specific health claim, meaning whether a specified nutrient, other substance or health-related ingredient was stated to have a specific health effect, or a non-specific health claim.
Once identified, claims were classified into nutrition claims (including nutrient content claims and nutrient comparative claim), health-related ingredient claims and health claims (including reduction of disease risk claims, nutrient and other function claims, general health claims and children’s development and health claims).
All claims were categorized as being either symbolic or non-symbolic. A symbolic claim was defined as a health-related claim (whether a nutrition, health-related ingredient or health claim) that was pictorial or combined words and pictures and for which the criteria for use of the symbol have been published. An example of a symbolic general health claim is the Dutch “Choices” logo [26
]. An example of a symbolic nutrient and other function claims is the “Toothfriendly” logo [27
]. see Figure 1
Color-coded nutrition labels (e.g., traffic light or % GDA labeling) did not qualify as a symbolic claim because they present nutritional information but do not suggest a health relationship [28
]. All NHC definitions are consistent with EU Regulation:
A nutrition claim was defined as “any claim which states, suggests or implies that a food has particular beneficial nutritional properties due to (a) the energy (calorific value) it (i) provides; (ii) provides at a reduced or increased rate; or (iii) does not provide; and/or (b) the nutrients or other substances it (i) contains; (ii) contains in reduced or increased proportions; or (iii) does not contain” [1
Nutrition claims were classified as either a nutrient content claim when it “describes the level of a nutrient contained in a food (or its energy value)” (e.g., “high in fibre” and “low fat”) or a nutrient comparative claim “when it compares the composition of the food in question with the composition of other foods” (e.g., “higher in fibre” and “reduced sugar”) [1
A health-related ingredient claim was defined as a claim communicating the presence of an ingredient(s) which is not a nutrient or other substance as defined in the EU Regulation [1
] but which implies health benefits. In most cases, these claims related to the content of ingredients that are considered as a healthy (e.g., “Contains one of your five a day”) or at least a healthier alternative (e.g., “Sweetened only with brown sugar”).
A health claim was defined as “any claim that states, suggests or implies that a relationship exists between a food category, a food or one of its constituents and health” [1
A general health claim (covered by Article 10(3) of the EU Regulation [1
]) was defined as a claim referring to benefits for general health or well-being. A typical example of such a claim is “Good for your health” or “Healthier choice (within this product group)”.
A nutrient and other function claim (as covered by Article 13 of the EU Regulation [1
]) was defined as a health claim that describes or refers to one of the following: (a) the role of a nutrient or other substance in growth, development and the functions of the body; (b) psychological and behavioral functions; or (c) slimming or weight-control or a reduction in the sense of hunger or an increase in the sense of satiety or a reduction of the available energy from the diet. Typical examples of such a claim are “Calcium builds strong teeth” and “Fibre helps maintain a healthy digestive system”.
A reduction of disease risk claim (covered by Article 14.1(a) of the EU Regulation [1
]) was defined as a claim communicating that the consumption of a food category, a food or one of its constituents significantly reduces a risk factor in the development of a human disease. Typical examples of such a claim are “Plant sterols reduce blood cholesterol. High cholesterol is a risk factor in the development of coronary heart disease” and “Reduces the risk factor for development of dental caries”.
A children’s development and health claim (covered by Article 14.1(b) of the EU Regulation [1
]) was defined as a health claim where children’s development and/or health was specifically mentioned. Typical examples of such a claim are “Calcium is needed for normal growth and development of bones in children” and “For your baby’s safe and balanced diet”. Claims where children were not specifically mentioned, even if these claims were found on foods intended solely for use by children, were not considered children’s development and health claims.
Claims such as “natural”, “organic” and “halal” were not considered NHCs on the basis that they refer to the method of the production rather than the content. Similarly, information on the presence of additives, preservatives, flavorings, etc. was not considered a health-related claim. Allergy advice (e.g., “contains nuts”) was not considered to be an NHC nor were references to the presence of a food or food group in the product where there was no clear relationship to its health benefits, e.g., references to milk content. Claims relating to the endorsement of a health-related organization were also not included, e.g., “[Product] supported by the [health association]”.
The present study assesses the prevalence of NHC on packaged foods currently sold in Germany, the Netherlands, Slovenia, Spain and the UK. Overall, approximately one quarter (26%) of foods sampled in this study carried an NHC. Twice as many foods carried nutrition claims (21%) compared to health claims (11%), followed by a much smaller percentage of health-related ingredient claims (4%).
On a country-level, the overall percentage of foods with at least one claim varied from 35% in the UK to 21% in Germany. The UK also had the highest number of foods carrying a health-related ingredient claim (10%), compared to only two foods (0.5%) surveyed in Slovenia. UK health-related ingredient claims were mostly related to fruit and vegetable consumption (e.g., “One of your five a day”) or wholegrain (e.g., “With wholegrain”). Regarding health claims, country-by-country variation was lower and did not follow the same distribution order. The highest frequency of foods carrying at least one health claim (including symbolic representations) was found in the Netherlands and the lowest frequency in Spain. A closer look at the Dutch data show that the majority of health claims identified here comprised of the Choices logo, which was classified as a symbolic general health claim. All five surveyed countries have been governed by the same European food regulation, since 2007. Our results demonstrate that reasons for the reported country differences regarding the prevalence of health claims and symbols appear to be more complex than just the implementation of a supranational regulation. EU Member States differ in their history of use of health claims and symbols prior to the EU regulation, food operators in different countries may employ different marketing strategies for their products, and the use of health symbols is often linked to national organizations issuing those symbols. Furthermore, it has been reported that EU Member States take different approaches to implementing the EU regulation on a national level. This includes national legislation on the responsibilities of food authorities (e.g., inspections), as well as the general control of the food supply [30
]. These differences are currently being investigated in another on-going EU funded project called REDICLAIM (www.rediclaim.eu
It appears that when foods carry claims, they tend to carry more than one claim, which can either mean that they repeat the same claim on several parts of the package or include more than one claim. The latter is often the case for nutrition claims, i.e., several nutrients are mentioned (e.g., “No sugar, low calories” and “Contains vitamin A, source of iron”). Furthermore, because health claims are commonly communicating functions of specific nutrients or other substances, such claims are commonly accompanied by related nutrition claims (e.g., “High in calcium” and “Calcium is needed for the maintenance of normal bones”).
Due to differences in methods, data collection timeframe, countries and food categories surveyed, it is difficult to compare the present findings with those of past surveys and any comparison should be undertaken with caution. Nevertheless, it is useful to note the prevalence of claims that previous studies have reported.
Of the products sampled in this study, 21% carried nutrition claims, compared to 29% that were reported in a study undertaken in the UK in 2011 [17
], 37% in Slovenia in 2011 [19
] and 48% in Ireland in 2007 [18
]. In Serbia, a candidate country for EU membership, 7% were reported in a study undertaken in 2012 [22
]. Outside Europe, similar studies reported 46% of foods carrying nutrition claims in Canada in 2013 [9
] and 49% in the USA in 2010 [7
]. Compared to 11% of products carrying health claims in this study, a prevalence of 15% health claims on food products was reported in the British study [17
], 13% in the Slovenian study [19
] and 18% in the Irish study [18
]. The Serbian study only reported 6% health claims on the food products they sampled [22
]. It should be noted that health claims were not regulated in Serbia at the time that the study was done. Outside the EU, 14% health claims were found on food products in a study undertaken in Australia and New Zealand [11
] and 9% in the US-based study [7
]. While the highest proportion of health claims in the present study comprised of general health claims (7%), only few reduction of disease risk claims (0.6%) were identified. This is in line with findings from other prevalence studies, e.g., <1% of reduction of disease risk claims in the British [17
], the Irish [18
], the Slovenian [19
] and the Canadian study [9
] and only 1% in the Australian and New Zealand study [11
]. Regarding the prevalence of symbolic claims, the present study reported such claims on 4% of all food packages. The highest proportion of symbolic claims was observed in the Netherlands (12%) and the lowest in Germany (0.3%), which is in line with the results of the FLABEL study mentioned above [16
]. The US study reported that 6% of food packages had symbolic claims [7
]. A comparison with the Canadian study [9
] is not possible due to differences in the definition of symbolic claims between that study and this.
The present study is the first, to-date, to survey several countries across Europe using the same methods, with data collection having taken place after the Regulation went into effect. Aside from providing an updated overview of the prevalence of nutrition, health-related ingredient and health claims on the European market, the present study offers a novel method of data collection in which the sampling frame was rigorously defined. It is hoped that the detailed description of the methods will provide guidance for other researchers on replicating and advancing future claim prevalence studies.
Furthermore, the inclusion of five different countries and the geographical spread that this entails has allowed for European conclusions to be drawn. This study provides an example of how to monitor the prevalence of claims currently on the market, one of the tasks crucial in implementing and monitoring legislation. It is also a necessary step towards making research a useful source of information for future regulation. At the same time, differences across European countries were taken into account by using local researchers for the data extraction. Nuances in the language were more likely to be appreciated by native speakers in each country. This approach required more coordination efforts but the authors believe that this was appropriate and would be the most suitable method for future similar research.
The study was powered to detect a 10% difference in the prevalence of claims by country, which had an effect on comparisons in other respects. For example, the sample size did not allow for country-by-country analysis for each food category. It also did not allow for a store-by-store analysis for each claim type. Having limited resources, the decision was either to focus on a few food categories and the number of investigated jurisdictions or to broaden the number of food categories and countries while limiting the number of foods per country. The latter option was seen as a better way to produce a snapshot of the current situation in the EU market, i.e., what consumers are exposed to, across a variety of stores and food categories.
The selection of an appropriate food categorization system has posed a further challenge in this study. A variety of categorization schemes have been used in similar studies, which significantly limits the comparability of results. The selected scheme was chosen due to the fact that it has been adopted by the Global Food Monitoring Group for all future monitoring of the food supply [23
]. Although this study is the first to use this categorization scheme for this type of labeling survey, it is the authors’ hope that this decision will contribute to the harmonization of food information reporting in future studies.
Last but not least, the cross-sectional design can be considered an additional limitation of this study. In order to examine changes in the food supply, it is recommended to repeat this study at a later point, e.g., after all transition periods of the EU Regulation have ended. A further avenue of future research could be to analyze differences in the prevalence of claims between retail (private label) products and branded products.