Observational data

Observational data, mostly from the USA, have identified several characteristics associated with nutrition label use including that it increases with age, is higher in women and normal weight individuals and increases with education⁽ Reference Cowburn and Stockley ^{17 –} Reference Stran and Knol ^{22 )}. Extensive research into the impact of food labelling across the EU has shown that many consumers can effectively use a nutrition label to rank a food for healthiness⁽ Reference Grunert, Wills and Fernandez-Celemin ^{19 ,} Reference Bonsmann, Celemin and Grunert ^{23 )}. In addition, consistency and simplicity have modest but meaningful effects on facilitating consumer attention to the label, for example the addition of a FoP label or health logo⁽ Reference Grunert and Wills ^{24 –} Reference Mejean, Macouillard and Peneau ^{27 )}. Consumers rate their use of nutrition labels as high; across the EU for example, up to 60 % of consumers indicate regular use of nutrient information on a food label. However, information such as price and use by date is more commonly referred to (approximately 90 % in both cases)^{( 28 )}. A study based on in-store observations involving over 2000 consumers in three major UK retailers, found that 27 % of UK shoppers ‘always’ looked specifically at the nutrition information on the label, with RI and the back of pack nutrition label the main sources consulted⁽ Reference Grunert, Wills and Fernandez-Celemin ^{19 )}.

Evidence suggests that consumer use of nutrition labels is largely driven by consumer belief in the importance of a healthy diet⁽ Reference Neuhouser, Kristal and Patterson ^{20 ,} Reference Lin, Lee and Yen ^{29 ,} Reference Roberto, Bragg and Seamans ^{30 )} as well as motivation⁽ Reference Bonsmann and Wills ^{31 )}. Those consumers who frequently consult nutrition information are observed to have higher dietary intakes of fibre and iron compared with those who rarely or never use the information⁽ Reference Variyam ^{32 )}. In addition, estimated energy intake (EI) among those reporting use of serving size information is on average 627 kJ (150 kcal) less per day than non-users⁽ Reference Ollberding, Wolf and Contento ^{33 )}. Those who report trying to lose weight compared with those who do not are shown to be three times more likely to read nutrition labels⁽ Reference Satia, Galanko and Neuhouser ^{34 )}. Similarly, high nutrition label use is shown to be positively related to a lower fat intake, higher fruit and vegetable intake, higher nutritional knowledge and greater tendency to read serving size information⁽ Reference Graham and Laska ^{21 ,} Reference Satia, Galanko and Neuhouser ^{34 )}. Nutrition labels may be particularly effective for consumers who are actively seeking information on particular nutrients due to personal relevance⁽ Reference Post, Mainous and Diaz ^{18 ,} Reference Neuhouser, Kristal and Patterson ^{20 ,} Reference Kang, Shim and Lee ^{35 )}. For example, individuals with type 2 diabetes, hypertension, hyperlipidaemia or any combination of these three diseases were found to read food labels significantly more than individuals with no such concern (71 v. 60 %)⁽ Reference Post, Mainous and Diaz ^{18 )}. Furthermore, in the latter study, the odds of reading food labels when advised by a health professional to reduce EI or weight, was 50 % higher than in those without health professional intervention. Of note, the nutrients of most relevance to these individuals were those related to their disease; namely fat in the case of hyperlipidaemia and salt in the case of hypertension. A strong association between nutrition label use and perceptions of the benefits of a low-fat diet among 1450 US adults has been reported⁽ Reference Neuhouser, Kristal and Patterson ^{20 )}. Those who stated that consumption of a low-fat diet was very important to their health were almost ten times more likely to read the nutrition label compared with those who rated a low-fat diet as unimportant. Similarly, in a representative sample of US adults (n 4024), individuals with high intakes of fat, saturated fat or cholesterol were found to be less likely to search for such information on a food label than those with low intakes of these nutrients⁽ Reference Lin, Lee and Yen ^{29 )}. In the latter study, the lower the nutrition label use, the less importance was placed on the nutrition information provided and the poorer the individual's diet. The cumulative evidence from these studies provides a useful insight into the potential link between dietary behaviour and label use. Moreover, the data provide useful information on the key characteristics of nutrition label use.

Experimental evidence

In contrast to the observational data, the direct impact of nutrition labelling on food intake under controlled conditions has not been extensively examined. Studies have mainly focused on consumption of a single food product⁽ Reference Steenhuis, Kroeze and Vyth ^{36 ,} Reference Roberto, Shivaram and Martinez ^{37 )} or single meal occasion⁽ Reference McCann, Wallace and Robson ^{38 )} or have been conducted within a restaurant setting to test the effects of CL⁽ Reference Roberto, Larsen and Agnew ^{39 )}. The impact of nutrition labelling in combination with education on subsequent EI at lunch has been explored⁽ Reference Temple, Johnson and Recupero ^{40 )}. Participants (n 47; age 18–50 years) were exposed to one of two video groups (a nutrition labelling educational video or a non-nutrition labelling related video) after which they were offered a buffet style lunch under one of two labelling conditions (nutrition labels with standard US Department of Agriculture nutrition facts panel or no nutrition labels). Although no effect of the video information on food choice was found, participants who were exposed to nutrition information about the buffet food items consumed significantly less energy at the lunch than those for which food items were not labelled. Notably, exposure to nutrition labelling increased selection of less energy dense products by the women only. In contrast, a further study found no differences in EI according to exposure to energy density information among 40 normal-weight women (age 18–45 years)⁽ Reference Kral, Roe and Rolls ^{41 )}. The women consumed breakfast, lunch and dinner in a laboratory on three separate days during which the energy density of the entree dish at each meal was either high, medium or low (5·23, 6·28 or 7·32 kJ/g, respectively). Subjects were assigned to either the information group (in which they received information on the energy density of the entree dish with each meal), or to the no-information group (in which they did not receive any nutrition information). No differences between information groups in food intake across the three levels of energy density were observed. Nevertheless, consumption of the low-energy dense dishes in general resulted in a significantly reduced EI compared with consumption of the high-energy dense dishes. Similarly, fat and EI following consumption of fat-free ‘potato chips’ v. regular ‘potato chips’ with and without nutrition information has been examined⁽ Reference Miller, Castellanos and Shide ^{42 )}. For the no-information condition, both types of product were presented in bags simply labelled ‘potato chips’. In the information condition, the bags containing the healthier and regular products were labelled as ‘fat-free potato chips’ or ‘potato chips’, respectively, and both had a corresponding US Nutrition Facts Panel. No differences in energy or fat intake were observed according to information condition. However, overall fat intake was found to be significantly reduced following consumption of the low-fat product. In addition, this resulted in a significantly reduced fat intake over 24 h compared with consumption of the regular product. Such data provide evidence for the benefits of reduced fat/energy products in improving dietary quality.

Further experimental data suggest that serving size information may be overlooked by consumers. For example, in a sample of US adults (n 216; age 18–72 years), portion size self-served and consumed of a breakfast cereal was examined according to three conditions of a FoP logo (consisting of a green tick and including energy per serving and per pack)⁽ Reference Roberto, Shivaram and Martinez ^{37 )}. Consumer knowledge and perceptions relating to various aspects including energy and perceived healthiness were also examined. Participants were randomly assigned to one of three study conditions: no label (control condition); label with servings per pack and energy (Kcal) per serving (condition two) and label with servings per pack, suggested energy per serving and suggested serving size (condition three). The study found that almost twice as many participants were able to accurately identify the amount of kcal per serving in the two label conditions relative to the control condition, supporting evidence regarding the ability of consumers to competently interpret simple portion size information from a food label^{( 28 )}. Given the addition of serving size information on the third label condition, the authors predicted that participants in that group would consume the least amount of cereal, i.e. in line with the suggested serving size information provided. However, while NS, there was a trend towards greater portion size consumption in condition three compared with the control condition and condition two (232·6 g v. 219·2 g and 219·9 g, respectively) suggesting that participants in condition three ignored the serving size information provided. Moreover, cereal consumed overall equated to almost twice the recommended serving size.

Nutrition labelling: counterproductive effects?

Food portion sizes, particularly of high-energy dense foods, have increased significantly over time, particularly in the USA⁽ Reference Young and Nestle ^{43 )}, with similar trends observed across Europe⁽ Reference Steenhuis, Leeuwis and Vermeer ^{44 )}. A wider range of portion sizes are now available and this has added to confusion amongst consumers about what constitutes an appropriate food portion size⁽ Reference Faulkner, Pourshahidi and Wallace ^{45 )}. Furthermore, consumers tend to increase their food and EI when presented with large food portions⁽ Reference Levitsky and Youn ^{46 –} Reference Kelly, Wallace and Robson ^{48 )}. However, just 30 % of European consumers indicate that they seek out serving size recommendations provided on food packaging, in contrast to a much greater proportion who seek out ingredients and nutrient values as referred to earlier (60 % in both cases^{( 28 )}). These issues are further exacerbated by the fact that foods labelled as ‘healthy’ seem to elicit a significant underestimation of energy content, resulting in inappropriate portion size perceptions⁽ Reference Faulkner, Pourshahidi and Wallace ^{15 ,} Reference Carels, Konrad and Harper ^{49 )}, an effect shown to be driven by dietary restraint⁽ Reference Lemmens, Born and Rutters ^{50 )}. The proposed explanation is that perceived healthiness is associated with expectations about energy content, such that foods perceived as ‘healthy’ induce less consumption guilt, thus providing a license to overeat, commonly termed as the ‘health halo’ effect⁽ Reference Wansink and Chandon ^{16 )}. This hypothesis has been tested under both natural and laboratory conditions, using typically consumed foods equal in energy density but differing in perceived healthiness such as M&M v. granola⁽ Reference Wansink and Chandon ^{16 )} and low fat v. standard coleslaw⁽ Reference Faulkner, Pourshahidi and Wallace ^{15 )}. For example, it has been shown that when iso-energy dense foods were labelled as regular v. low fat, EI increased in the low fat condition during a single consumption occasion by up to 50 %⁽ Reference Wansink and Chandon ^{16 )}. Importantly, these patterns are strongly affected by weight status, with overweight individuals particularly receptive to such cognitive cues. In support of this a further study has shown that when participants (n 47; age 18–65 years) were exposed to three iso-energy dense meals (chicken curry) labelled as low fat/low energy, high fat/high energy or standard, the participants associated less guilt with consumption of the low-fat/low-energy meal, and consumed 28 g more of this meal compared with the standard labelled meal, resulting in an additional 3 % (162 kJ) EI. Notably this result was driven by the overweight subjects. In addition, participants reported a significantly higher rating of taste for both the standard and the high fat/high energy labelled meal compared with the low fat/low energy labelled meal⁽ Reference McCann, Wallace and Robson ^{38 )}. In contrast, further research showed that the use of a nutrition logo did not result in increased consumption and had no effect on the rating of taste of a sweet pastry product among a similar sample of females⁽ Reference Steenhuis, Kroeze and Vyth ^{36 )} suggesting that different food types may elicit different responses.

Under the EU Food Information for Consumers regulation (EU 1169/2011), back of pack nutrition information must be displayed per 100 g/ml, while per serving amounts may be provided as a voluntary addition^{( 5 )}. In an online survey involving over 13 000 adults across six EU countries (Germany, UK, Spain, France, Poland and Sweden), the addition of nutrient information per portion on a range of typically consumed food products greatly increased the accuracy with which participants were able to determine an appropriate portion size compared with when the information was only provided per 100 g^{( 28 )}. Furthermore, within the same sample, it was found that portion size information had a significant impact on participants perceptions of the healthiness of the product⁽ Reference Raats, Hieke and Jola ^{51 )}. For example, participants were asked to rate the healthiness of three pairs of typically consumed foods (biscuits, sandwiches and yoghurts) based on nutrient values presented for three different portion size information conditions (per 100 g, per typical portion or per half typical portion). The impact of Guideline Daily Amounts information was also tested. In line with previous research, participants were able to distinguish between more and less healthy products using the nutrition information provided. However, in the case of biscuits, when the nutrient composition was displayed per 100 g (providing 223 8kJ/535 kcal), consumers rated the product as significantly less healthy than when it was displayed per typical portion (18 g; providing 402 kJ/96 kcal) or half typical portion (9 g; providing 201 kJ/48 kcal). For sandwiches on the other hand, participants rated the product with the 100 g label only (providing 669 kJ/160 kcal) as significantly more healthy than when provided with information per typical portion (250 g; providing 1674 kJ/400 kcal) or half typical portion (125 g; providing 834 kJ/200 kcal). These data provide evidence for the importance of nutrition information to be displayed per portion as well as the mandated per 100 g/ml. Thus the standardisation of serving size information on a food label, presently not mandated in the EU, is an important consideration.

Menu labelling

Since the implementation of mandatory CL in the USA^{( 11 )} numerous observational and experimental studies have been conducted in various settings to assess its impact on food choice and EI, many of which have shown positive effects⁽ Reference Roberto, Larsen and Agnew ^{39 ,} Reference Gerend ^{52 –} Reference Ellison, Lusk and Davis ^{54 )}. However, a recent extensive review of the related literature reported no clear patterns due to a lack of comparability across studies. Thus the effectiveness of CL as a weight management strategy is inconclusive⁽ Reference Kiszko, Martinez and Abrams ^{55 )}. Nevertheless, the data to date provide a useful insight into the main predictors of CL use which, in common to food packaging nutrition label use, include sex (higher in females⁽ Reference Gerend ^{52 ,} Reference Krieger, Chan and Saelens ^{56 )}), being on a weight loss diet⁽ Reference Girz, Polivy and Herman ^{57 )} and having a high level of nutrition knowledge⁽ Reference Auchincloss, Young and Davis ^{58 )}. Moreover, the data have provided evidence that consumer awareness of CL has increased over time. For example self-reported awareness among US consumers of the presence of energy on menus increased significantly from 25 % 3 months pre-regulation in 2008, to 64 % 3 months post-regulation⁽ Reference Dumanovsky, Huang and Bassett ^{59 )}. Furthermore, cross-sectional surveys carried out 1 year before and 18 months after the 2008 regulation observed a small but significant reduction in energy intakes in some fast food establishments surveyed⁽ Reference Dumanovsky, Huang and Nonas ^{60 )}. While increased awareness of nutrition labelling does not necessarily result in greater application⁽ Reference Grunert and Wills ^{24 )}, the increase over time is encouraging, and with continued promotion and food business support, a consistent positive impact of CL may begin to emerge. Indeed, provision of nutrition information within a worksite cafeteria, in combination with education and increased exposure to low-energy dense food products, resulted in an improved dietary intake in overweight and obese adults over a 3-month period⁽ Reference Lowe, Tappe and Butryn ^{61 )}. Such data show promise for a positive impact of the sustained provision of nutrition information at the point of purchase. Moreover, the latter study shows that a combination of familiarity with surroundings, exposure to healthier food products and raised awareness via education may be key drivers in its impact.