Cecchini2015.pdf

obesity reviews

doi: 10.1111/obr.12364

Public Health/Behaviour

Impact of food labelling systems on food choices and eating behaviours: a systematic review and metaanalysis of randomized studies M. Cecchini and L. Warin

Health Division, OECD, Paris, France

Received 13 July 2015; revised 3 November 2015; accepted 4 November 2015 Address for correspondence: Dr M. Cecchini, OECD, Health Division, 2, Rue André-Pascal, Paris 75016, France. E-mail: [email protected]

Summary Food labels are considered a crucial component of strategies tackling unhealthy diets and obesity. This study aims at assessing the effectiveness of food labelling in increasing the selection of healthier products and in reducing calorie intake. In addition, this study compares the relative effectiveness of traffic light schemes, Guideline Daily Amount and other food labelling schemes. A comprehensive set of databases were searched to identify randomized studies. Studies reporting homogeneous outcomes were pooled together and analysed through meta-analyses. Publication bias was evaluated with a funnel plot. Food labelling would increase the amount of people selecting a healthier food product by about 17.95% (confidence interval: +11.24% to +24.66%). Food labelling would also decrease calorie intake/choice by about 3.59% (confidence interval: 8.90% to +1.72%), but results are not statistically significant. Traffic light schemes are marginally more effective in increasing the selection of healthier options. Other food labels and Guideline Daily Amount follow. The available evidence did not allow studying the effects of single labelling schemes on calorie intake/choice. Findings of this study suggest that nutrition labelling may be an effective approach to empowering consumers in choosing healthier products. Interpretive labels, as traffic light labels, may be more effective. Keywords: Calorie intake, food choice, food labelling, meta-analysis, obesity. obesity reviews (2016) 17, 201–210

Introduction Dietary patterns have gone through substantial changes during the last 30–40 years. Calorie availability has been increasing worldwide by approximately 580 kcal per capita per day (1). At the same time, calorie consumption has significantly risen in the USA (2) and in a number of European countries (3). A substantial part of this upsurge can be attributed to increased consumption of ultra-processed products (i.e. ready-to-consume foodstuff mostly made from industrial ingredients) that now account for up to more than 50% of total calorie intake in high-income countries (4). The recent economic crisis may have further

exacerbated a switch to lower-priced (per calorie) and less healthy food (5). Underpinned by these trends, unhealthy diet and one of its consequences, obesity, have become a major public health concern (6). The latest data show that the majority of the adult population in the OECD area and one in five children are overweight or obese (7). If recent trends continue, these figures will inevitably worsen (8). In 2010 alone, more than 11 million deaths worldwide were amenable to diseases related to unhealthy diet and obesity, including diabetes, cancers and cardiovascular diseases (9). Food labelling is increasingly considered a crucial component of comprehensive strategies to tackle unhealthy diet 201

© 2015 World Obesity

17, 201–210, March 2016

202 Impact of food labelling M. Cecchini & L. Warin

and associated chronic diseases (10). For example, EU, the USA, the UK and Chile are all discussing or implementing new legislation on food labelling (7). In particular, food labels are regarded as a possible tool to empower consumers and to facilitate healthier food choices (11). Previous studies in the field suggest that food labelling would be associated with healthier eating habits (12–20) and would be a costeffective intervention (21). However, the majority of studies are based on a relatively small sample, and in some cases, results appear mixed. Previous systematic reviews have mainly focused on consumers understanding of food labelling schemes (22,23) and did not attempt to quantify the effects of food labelling in changing food choices. This systematic review and meta-analysis enrich the literature by providing a quantitative assessment of the impact of food labelling. In particular, this analysis aims at assessing the effectiveness of food labelling schemes in increasing the selection of healthier products and in reducing calorie intake/choice. The secondary objective is to determine whether food labels’ format influences choices and consumption. This analysis considers three types of food labelling schemes: traffic light, Guideline Daily Amount (GDA) and other types of food labelling (e.g. front-of-pack logos). Traffic light and GDA are among the most widely adopted nutrition labelling schemes (7) and are central to the ongoing policy (24) and academic debate (25).

Methodology This review was restricted to peer-reviewed studies that were designed to evaluate the impact of food labelling in terms of likelihood of selection of either a healthier option or calorie choice/intake. Studies had to be randomized and should include a control population with either a ‘before/after’ design or a ‘case/control’ design. The inclusion of randomized studies minimizes the influence of confounding factors and increases the strength and reliability of the pooled results. Studies that did not focus on consumers or that took marketing or psychological perspectives were excluded. For example, we excluded studies that investigated the type of labelling format that looks more appealing to consumers. In addition, the review protocol excluded studies that evaluated menu labelling as well as studies that did not provide the requested quantitative data to feed the meta-analysis. Finally, studies had to be in English or French and had to be published between January 2008 and April 2015 when the final search took place. The search was conducted using the following databases: Pubmed, Biomed, Science Direct, Sage Database, Google Scholar, EBSCO Host Database Academic. Searched keywords included were the following: ‘Food labelling’, ‘trafficlight labelling’, ‘traffic-light nutrition labelling’, ‘Nutrition labelling’, ‘Nutrition claims labelling’, ‘Nutrition claims regulation’, ‘EU regulation food labelling’, ‘EU nutrition labelling 17, 201–210, March 2016

obesity reviews

program’, ‘Traffic light label’, ‘Public health nutrition’, ‘Health logo’ ‘pick the tick’, ‘Guideline Daily Amounts’, ‘the Heart Symbol’, ‘the Choices logo’, ‘The Choices program in the Netherlands’, ‘The Keyhole program in Sweden’, ‘Program less salt is healthier’, ‘Green check marks’, ‘EU regulation food labelling’, ‘EU nutrition labelling program’ and ‘EU nutritional labelling program’. The search strategy was developed on a previous systematic review of interventions to promote healthy diet and physical activity (26). We identified all the papers on food labelling included in the review, and then, we extracted all the keywords from the identified papers. The retrieved keywords were searched individually or in combination using the standard Boolean operators. A full version of the articles meeting the inclusion criteria was obtained and analysed. Papers were divided into three categories: GDA, traffic light and other food labelling, according to the type of food labelling investigated. If a study evaluated multiple types of labelling, results for each type were considered as an independent study. A researcher extracted all the information on the characteristics of the study including the geographical area where the study was conducted, the food studied and the target population as well as the quantitative data needed to carry out the metaanalysis (i.e. sample sizes, means and standard deviations of control and intervention groups). In some cases, missing dimensions, usually standard deviations, had to be calculated by using standard approaches (27). This analysis focuses on two outcomes. The first outcome is the number of people that, following the implementation of food labelling, switches to a healthier product. The outcome was analysed as a percentage by dividing the number of people choosing the healthier option by the total number of people in the group. The second outcome is change in calorie intake/choice following the introduction of a food labelling scheme. For some studies, food quantities (e.g. grammes of pizza or cereals) or food nutrients (e.g. grammes of fat) had to be converted into calories using a nutrient database (28). Included studies investigated a broad range of products with different calorie content. To adjust for this heterogeneity, the change in calorie intake in the intervention group was scaled to a percentage by reporting it to the calorie intake of the control group. A second set of analyses investigated the two same outcomes in terms of, respectively, standardized mean difference (SMD) and absolute number of calories. Details of the methodology and results can be found in Appendix 2. Data were analysed with STATA 13 (StataCorp, College Station, TX, USA). Forest plots were generated, and overall estimates of the pooled relation and 95% confidence intervals (CI) were calculated with the use of fixed-effect and randomeffect models. Heterogeneity across studies was tested with the I2 statistics (29). In case of low heterogeneity (30), we carried out a graphical assessment of the potential publication bias through a funnel plot (31). © 2015 World Obesity

obesity reviews

Results A total of 137 articles were originally retrieved and selected for the review (Fig. 1); 76 studies were discarded because they were not peer-reviewed original investigations. Most of the documents consisted in reports from international organizations, opinion pieces and other types of reports. Another 52 studies were pulled out for review but excluded in a second step because they did not meet our inclusion criteria. Only nine studies met our inclusion criteria and reported all the data needed to carry out a meta-analysis. Three out of these nine studies focused on food intake, purchase or choice (13,14,18); other five reported data on food choice healthiness (15–17,19,20), while one study reported results for the two dimensions (12). The majority of the studies (12,14–17,19) assessed front-of-pack labelling; three studies (13,18,20) did not explicitly specify the position of the label. Two studies (19,20) reported results in a format that was not directly comparable with the other studies and were excluded from the main analyses but were included in the additional set of analyses (Appendix 2) after transformation into an SMD. The key characteristics of all the studies included in the analyses are reported in Table 1. The full list of studies identified in the systematic review is reported in Annex 1. Six studies (12,13,15,16,19,20) assessed the effectiveness of at least two different food labelling schemes with the paper by Watson et al. (15) assessing seven different variants. All the studies were carried out in high-income countries with the USA, Australia and the UK being the three most represented countries (two studies each). Other three studies were carried out in France, Germany and Canada. Five

Figure 1

Impact of food labelling M. Cecchini & L. Warin 203

studies were designed as experimental analyses: three of them were carried out in a controlled setting (12,14,18) and two in a real-world setting (13,17). The remaining four studies were carried out as online randomized trials (15,16,19,20). The majority of studies were carried out on a relatively small sample of participants (median number of participants 703) with only three studies enrolling more than 1000 participants (15,17,19). Some studies concluded that food labelling may have some effect on consumers’ choice (15,18,19). However, often, individual studies did not find any statistically significant effect on consumption (13,14). Various studies (17,32–34) suggest that other factors (e.g. socio-demographic characteristics) may have an effect in determining how food labelling influences food selection. Figure 2 presents the results of the meta-analysis for selecting/purchasing a healthier option. Results are presented as the share (expressed in percentage) of purchasers that, following the introduction of food labelling, switches to a healthier product. Single studies present very heterogeneous estimates. The effectiveness of food labelling ranges between 3.50% (CI: 7.32% to +0.32%) and +52.20% (CI: 47.61% to 56.79%). Studies in the GDA category present more homogeneous estimates, while the category other food labels shows the highest heterogeneity. Under the assumptions of a random-effect model, food labelling can be expected to increase the number of people selecting a healthier option by about 17.95% (CI: 11.24% to 24.66%). All the three categories of food labelling included in the analysis present a positive, statistically significant effect. Traffic light is the most effective labelling scheme and increases the number of people selecting a healthier option by 29.36% (CI: 19.73% to 39.00%). Other food labels and GDA follow with, respectively, an increase of 14.69%

Flow diagram of the literature search and filtering results of the systematic review.

© 2015 World Obesity

17, 201–210, March 2016

Country/ geographical area

Hamburg, Germany

London, UK

USA

Australia

USA

Marseille, France

Australia

Borgmeier and Westenhoefer, 2009

Crockett et al., 2014

17, 201–210, March 2016

Roberto et al., 2012 (appetite)

Watson et al., 2014

Roberto et al., 2012 (AJPM)

Gaigi et al., 2015

Spanos et al., 2015

Experiment in controlled setting

Experiment in real-world setting

Randomized online trial

Randomized online trial

Experimental in controlled setting

Experiment in real-world setting

Experiment in controlled setting

Type of study and setting

Serving size labels

Seven different front-of-pack labelling schemes comprising variants of the traffic light labelling scheme and the percentage daily intake scheme, and a star rating scheme (i) no label; (ii) traffic light; (iii) traffic light and protein/fiber traffic light; (iv) facts up front; (v) facts up front and info on ‘nutrients to encourage’ Logo ‘Le Choix Vita+’

Pizza

104 female undergraduate students at an Australian university

2083 French consumers

Food intake

Purchase

Food choice 703 US consumers recruited through an online database

US brand name products selected from eight categories

Dairy products, cooked meals and fresh snacks

Label use and understanding

Food choice

4357 grocery shoppers in Australia

216 US consumers

Cereals

The Smart Choices front-of-package nutrition label in the USA

Consumption

Food choice, envisaged consumption

Outcome of interest

Nine pairs of commonly purchased food products

287 participants attending a London cinema

Popcorn

(i) green low-fat label; (ii) red high-fat label; and (iii) no label

420 adult subjects from a random sample from Hamburg population

78 foods from different food categories

(i) simple ‘healthy choice’ tick; (ii) multiple traffic light label; (iii) monochrome Guideline Daily Amount label; (iv) coloured Guideline Daily Amount label; and (v) ‘no label’ condition.

Population studied

Type of food

Type of food labelling system studied

Main characteristics of the studies included in the meta-analysis

Author, year

Table 1

Different food label formats differ in the understanding of consumers. Traffic lights have the best effects and influence most the perceived healthiness of food but, despite this, are unlikely to influence food choice and consumption. Contrary to predictions, no main effects of nutritional labels on consumption were identified. Calories per serving information on front-of-pack labels can increase knowledge, but the Smart Choices symbol has little impact on behaviour. The number of correct responses in the control group was significantly less than in other schemes (P < 0.05). Even without an accompanying education campaign, front-of-pack labels can assist shoppers to make healthy choices. Overall, those in the traffic light condition performed better than those in the facts up front conditions on measures of nutrition knowledge and label perceptions. There was no main impact of the logo on food purchase, probably because of several factors such as lack of visibility of the logo on supermarket shelves and clients’ low socio-economic group. The nature of the information provided influences how much participants eat. Participants had lower consumption

Authors’ conclusions

204 Impact of food labelling M. Cecchini & L. Warin

obesity reviews

© 2015 World Obesity

Purchasing intentions 1017/1002 consumers for, respectively, plant sterols/oat fibre

955 people representative of the UK

Foods carrying cholesterollowering claims

Chocolate and cereal bars

Food choice, desired consumption

of pizza if the same portion was labelled ‘contains 4 servings’ compared to ‘contains 2 servings’. In both studies, all nutrition-related claims elicited more positive responses than the taste control claim. The taste control claim was rated the least influential or useful. In line with recent systematic review evidence, nutritional labels had no effect on choice of snacks.

Outcome of interest

Cholesterol-lowering claims on food

Emoticons and colours on food labels

Randomized online trial

Randomized online trial

Canada

UK

Wong et al., 2014

Vasiljevic et al., 2015

Type of food Type of food labelling system studied Type of study and setting Country/ geographical area Author, year

Table 1. (Continued)

Impact of food labelling M. Cecchini & L. Warin 205

Population studied

Authors’ conclusions

obesity reviews

© 2015 World Obesity

(CI: 3.56% to 25.82%) and 11.85% (CI: 5.43% to 18.28%). Results obtained by standardizing the input data with the SMD approach (Annex 2) and by including two additional studies (19,20) confirm that traffic lights are the most effective labelling scheme. GDA becomes the second most effective intervention. Findings of the analyses on the effectiveness of food labelling schemes in modifying calorie intake/choice are presented in Fig. 3. Results are presented as change in calorie intake/choice (percentage of the control group). A second set of analyses, reporting the effectiveness in terms of absolute number of calories, are presented in Annex 2. Studies generally report homogeneous, largely overlapping effects. Homogeneity across studies is also confirmed by low levels of the I2 statistics. Seven out of the 12 included interventions show that calorie intake/choice decreases once those food labels are introduced. However, no single study presents a statistically significant effect, and often, CIs are several times the size of the expected effect. Pooling together all the interventions returns an average effect of 3.59% (CI: 8.90% to +1.72%). This means that, on average, food labelling schemes can be expected to decrease calorie intake/choice by about 3.6%. However, this figure is not statistically significant. Given the low number of studies included in the GDA and traffic light categories, it is recommended not to compare results across subcategories as this would provide no meaningful conclusion. Other food labels have an effect of 2.88% (CI: 10.08% to +4.32%), which is slightly lower, but not statistically significantly different from the overall results. Analyses carried out in terms of absolute number of calories (Annex 2) confirm the findings of the main analysis. Studies included in Fig. 3 were also graphically assessed for any potential publication bias through a funnel plot (Fig. 4). The studies were plotted with the estimated effect on the horizontal axis and the standard error of the estimated effect on the vertical axis. Studies with a smaller sample scatter more widely at the bottom of the graph, while larger studies are closer to the true effect of the intervention and are positioned in the upper part of the diagram. Studies that are more precise fall within the 95% CI. In the absence of a publication bias, the plot should look symmetric.

Discussion This systematic review and meta-analysis summarize the current evidence on how different food labelling schemes may modify the selection of healthier products and calorie intake/choice. Results of the meta-analysis show that food labelling may play a significant role in facilitating consumers to select healthier food products. Traffic light schemes, in particular, would be the most effective food 17, 201–210, March 2016

206 Impact of food labelling M. Cecchini & L. Warin

obesity reviews

Figure 2 Effect of food labels in selecting/purchasing a healthier option. CI, confidence interval; D + L, random-effects estimate–DerSimonian and Laird method; GDA, Guideline Daily Amount; I V, fixed-effects estimate–inverse variance method; WMD, weighted mean difference.

labelling scheme in steering consumers’ choices. Conversely, our findings show a less clear picture in terms of whether food labelling schemes affect calorie choice or consumption. Overall, food labels would have the potential to decrease calorie choice/intake. However, single studies report large CIs, suggesting that different individuals respond to the introduction of food labels with a wide range of behaviours, indifference included. Previous systematic reviews in the field focused on consumers understanding of food labelling schemes (22,23) and on whether menu labels would affect selection and consumption of calories in restaurants (35,36). This study enriches the literature by investigating the effectiveness of food labels in increasing the selection of healthier products and in modifying calorie choice/intake. Our results suggest that menu labels and food labels may have similar effects on consumers’ eating behaviours. Previous literature showed that this may happen because 17, 201–210, March 2016

consumers would have only a limited understanding of the information provided by food labels (37). Findings of our study support this hypothesis by showing that interpretive nutrition labels, as traffic light systems, are more effective in helping consumers in choosing healthier products. Another reason may also explain why significant positive effects on food choice are not followed by a decrease in calorie choice/intake of a similar magnitude. Calorie content is only one of the multiple dimensions affecting the healthiness of products. For example, consuming a product with reduced salt content is healthier but does not reduce calorie intake. Again, substituting unhealthy nutrients (e.g. trans fats) with healthier options (e.g. polyunsaturated fats) may not produce any significant effect on calorie intake. The so-called ‘halo effect’ may also play a key role. Consumers tend to generalize to a whole food product a piece of information found on the label (38). This especially applies in the © 2015 World Obesity

obesity reviews

Impact of food labelling M. Cecchini & L. Warin 207

Figure 3 Effect of food labels on calorie intake/choice (percentage of control group). CI, confidence interval; D + L, random-effects estimate–DerSimonian and Laird method; ES, effect size; GDA, Guideline Daily Amount; I V, fixed-effects estimate–inverse variance method; WMD, weighted mean difference.

Figure 4 Funnel plot of studies analysing the effect of food labels on calorie intake/choice.

case of nutrient and health claims. For example, if a logo states that a product is a source of whole grain, consumers tend to think that the product is also low in calories, © 2015 World Obesity

saturated fats and sugar (38). As a result, consumers may even end up being tempted to eat more than they would otherwise. Consumers may also compensate a healthy dietary choice with consumption of other unhealthy food. For example, diet-beverage consumption is associated with higher calorie intake from discretionary food (39). The majority of the studies meeting the inclusion criteria were carried out in controlled settings or as online trials. Little is known on the potential transferability of the obtained effects to a real-world shopping situation (e.g. supermarkets) where most food purchase takes place. However, previous studies (21,23,40) concluded that about two-thirds of consumers read nutrition labels before purchasing a food product. In addition, users of food labels would benefit from a small decrease in body mass index, which would eventually decrease the likelihood of developing chronic diseases as diabetes, cancers and cardiovascular diseases. This would produce a significant increase in population health and savings in health expenditure. 17, 201–210, March 2016

obesity reviews

208 Impact of food labelling M. Cecchini & L. Warin

Findings of this study have strong policy implications. Many countries are in the process of debating, developing or implementing new food labelling schemes (7,41). Results of our work support the implementation of food labelling as a key tool to tackle unhealthy diet and obesity. Food labels empower consumers by providing nutrition information (23). Food labels may also prompt the industry to produce healthier food through nutrient reformulation (38,42). In fact, product reformulation may actually be one of the main mechanisms through which food labels could impact consumers’ diets (42) in the short term as changing people’s behaviour is more complex and may require longer than changing the environment. It is up to policymakers to decide the trade-off between completeness of information and facility of understanding. This analysis suggests that interpretive nutrition labels may be marginally more effective in steering consumers’ choices. This investigation has three main strengths. First, previous literature focused on analysing whether consumers understand labelling. This study, instead, investigates the effects of labelling on selecting foodstuff. Thus, it quantifies how much food labels can modify diets and, eventually, obesity prevalence. Second, this study goes beyond a qualitative analysis of the literature by pooling together quantitative evidence in a meta-analysis. Therefore, the results of this work can be used to compare the effectiveness of food labels with other labelling policies (e.g. menu labelling) and, more broadly, with other policies to tackle overweight and obesity (e.g. counselling and mass media campaigns). Third, this study compares the marginal effects of different labelling schemes producing new evidence about the expected effectiveness of GDA labels, traffic light labels and other types of food labels. The main potential limitation of this study lies with the quality of the studies included in the meta-analysis. The majority of studies are based on a relatively small sample, particularly by considering that the total number of participants is often divided into groups to test different food labelling schemes. Too small samples may not have enough statistical power to fully account for possible confounders (e.g. socio-demographic conditions and price). Small samples may also explain most of the cross-study (analysis on the selection of healthy options) and withinstudy (analysis on calorie intake/choice) heterogeneity that we found in the retrieved studies. The inevitable assumptions we had to make cluster together slightly different labelling formats could be, instead, a limitation for the analyses by type of labelling scheme. For example, this study groups together standard GDA labels with their ‘coloured’ variant in which levels of nutrients are emphasized by a traffic light approach. It cannot be excluded that variants of the same scheme have different effects. Future research on food labelling should focus on addressing three key deficiencies. The first concerns the small samples of the studies that do not allow for full consideration of 17, 201–210, March 2016

confounders and preclude analyses by population subgroups. Second, there is a need of a higher number of studies carried out in the ‘real world’ as opposed to laboratory settings. These two issues could be addressed with a closer collaboration between researchers and retailers, e.g. by carrying out large-scale randomized trials entailing the introduction of innovative labels in a random group of stores. The linkage of sales data with ‘loyalty programmes’ data would provide a solution to the two aforementioned issues. Finally, researchers should standardize study protocols and, in particular, outcome definitions. A number of studies had to be discarded because results were reported in non-standard units of measure (e.g. 43–45) or studied unusual outcomes (e.g. 46,47). In conclusion, from the evidence produced by this systematic review and meta-analysis, it appears that food labelling schemes would have a statistically significant effect in steering consumers’ choice towards healthier products. Interpretive nutrition labels, as traffic light schemes, may be more effective than other approaches. Food labels could also help consumers in choosing/consuming foodstuff with lower calorie content, but the available evidence is currently too limited to produce statistically significant results.

Conflict of interest statement No conflict of interest statement.

Acknowledgements The opinions expressed and arguments employed herein are solely those of the authors and do not necessarily reflect the official views of the OECD or of its member countries.

Supporting information Additional Supporting Information may be found in the online version of this article, http://dx.doi.org/10.1111/ obr.12364 Appendix 1. List of studies retrieved in the systematic review but discarded. Appendix 2. Additional results. Figure A1. Effect of food labels in selecting/purchasing a healthier option. Figure A2. Effect of food labels on calorie intake/choice (absolute calories).

References 1. World Health Organization. Global and Regional Food Consumption Patterns and Trends, 2015. [WWW document]. URL http://www.who.int/nutrition/topics/3_foodconsumption/en/ (accessed 21 October 2015).

© 2015 World Obesity

obesity reviews

2. Brunello G, Michaud PC, Sanz-de-Galdeano A. The Rise in Obesity Across the Atlantic: An Economic Perspective. IZA Discussion Papers 3529. Institute for the Study of Labor (IZA): Bonn, 2008. 3. Snowdon C. Institute of Economic Affairs. The fat lie, 2014. [WWW document]. URL http://www.iea.org.uk/sites/default/files/in-the-media/ files/Briefing_The%20Fat%20Lie.pdf (accessed 21 October 2015). 4. Monteiro CA, Moubarac JC, Cannon G, Ng SW, Popkin B. Ultra-processed products are becoming dominant in the global food system. Obes Rev 2013; 14(Suppl. 2): 21–28. 5. Cecchini M. Use of seemingly unrelated equations to assess changes in dietary behaviours during the UK’s economic crisis. Lancet 2014; 384(Suppl. 2): S8. 6. Ezzati M, Riboli E. Behavioral and dietary risk factors for noncommunicable diseases. N Engl J Med 2013; 369: 954–964. 7. Organization for Economic Co-operation and Development. Obesity Update. OECD Publishing: Paris, 2014. 8. Organization for Economic Co-operation and Development. Obesity and the Economics of Prevention: Fit not Fat. OECD Publishing: Paris, 2010. 9. Lim SS, Vos T, Flaxman AD et al. A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 2012; 380: 2224–2260. 10. WHO. Discussion Paper, Prevention and Control of NCDs: Priorities for Investment, 2011. [WWW document]. URL http:// www.who.int/nmh/publications/who_bestbuys_to_prevent_ncds.pdf (accessed 21 October 2015). 11. Méjean C, Macouillard P, Péneau S, Hercberg S, Castetbon K. Consumer acceptability and understanding of front-of-pack nutrition labels. J Hum Nutr Diet 2013; 26: 494–503. 12. Borgmeier I, Westenhoefer J. Impact of different food label formats on healthiness evaluation and food choice of consumers: a randomized-controlled study. BMC Public Health 2009; 9: 184. 13. Crockett RA, Jebb SA, Hankins M, Marteau TM. The impact of nutritional labels and socioeconomic status on energy intake. An experimental field study. Appetite 2014; 81: 12–19. 14. Roberto CA, Shivaram M, Martinez O, Boles C, Harris JL, Brownell KD. The Smart Choices front-of-package nutrition label. Influence on perceptions and intake of cereal. Appetite 2012; 58: 651–657. 15. Watson WL, Kelly B, Hector D et al. Can front-of-pack labelling schemes guide healthier food choices? Australian shoppers’ responses to seven labelling formats. Appetite 2014; 72: 90–97. 16. Roberto CA, Bragg MA, Schwartz MB et al. Facts up front versus traffic light food labels a randomized controlled trial. Am J Prev Med 2012; 43: 134–141. 17. Gaigi H, Raffin S, Maillot M, Adrover L, Ruffieux B, Darmon N. Expérimentation d’un fléchage nutritionnel dans deux supermarchés à Marseille “Le Choix Vita+”. Cah Nutr Diet 2015; 50: 1–9. 18. Spanos S, Kenda AS, Vartanian LR. Can serving-size labels reduce the portion-size effect? A pilot study. Eat Behav 2015; 16: 40–42. 19. Wong CL, Mendoza J, Henson SJ, Qi Y, Lou W, L’abbé MR. Consumer attitudes and understanding of cholesterol-lowering claims on food: randomized mock-package experiments with plant sterol and oat fibre claims. Eur J Clin Nutr 2014; 68: 946–952. 20. Vasiljevic M, Pechey R, Marteau TM. Making food labels social: the impact of colour of nutritional labels and injunctive norms on perceptions and choice of snack foods. Appetite 2015; 91: 56–63. 21. Cecchini M, Sassi F, Lauer JA, Lee YY, Guajardo-Barron V, Chisholm D. Tackling of unhealthy diets, physical inactivity, and obesity: health effects and cost-effectiveness. Lancet 2011; 376: 1775–1784.

© 2015 World Obesity

Impact of food labelling M. Cecchini & L. Warin 209

22. Cowburn G, Stockley L. Consumer understanding and use of nutrition labelling: a systematic review. Public Health Nutr 2014; 8: 21–28. 23. Campos S, Doxey J, Hammond D. Nutrition labels on prepackaged foods: a systematic review. Public Health Nutr 2011; 14: 1496–1506. 24. European Commission. Monitoring the Activities of the EU Platform on Diet, Physical Activity and Health – Draft Annual Report 2015. EC: Brussels, 2015. 25. Crosetto P, Muller L, Ruffieux B. Helping consumers with a front-of-pack label: number or colours? Experimental comparison between guideline daily amount and traffic light in a diet-building exercise. GAEL working paper 2015–10. Laboratoire d’Economie Appliquée de Grenoble: Grenoble, 2015. 26. World Health Organization. Interventions on Diet and Physical Activity: What Works: Evidence Tables, 2009. [WWW document]. URL http://www.who.int/dietphysicalactivity/evidence-tables-WW.pdf (accessed 21 October 2015). 27. Cochrane Community. Cochrane Handbook for Systematic Reviews of Interventions, 2011. [WWW document]. URL http:// community.cochrane.org/handbook (accessed 21 October 2015). 28. Nutrient Data Laboratory, USDA National Nutrient Database for Standard Reference. 2015. [WWW document]. URL http://ndb. nal.usda.gov/ (accessed 21 October 2015). 29. Higgins JPT, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analysis. BMJ 2003; 327: 557–560. 30. Sterne JA, Sutton AJ, Ioannidis JP et al. Recommendations for examining and interpreting funnel plot asymmetry in meta-analyses of randomised controlled trials. BMJ 2011; 343: 1–8. 31. Egger M, Davey Smith G, Schneider M, Minder C. Bias in metaanalysis detected by a simple, graphical test. BMJ 1997; 315: 629–634. 32. Méjean C, Macouillard P, Péneau S, Lassale C, Hercberg S, Castetbon K. Association of perception of front-of-pack labels with dietary, lifestyle and health characteristics. PloS ONE 2014; 9: 1–11. 33. Menard C, Dumas C, Gillot N et al. The French OQALI survey on dairy products: comparison of nutrient contents and other nutrition information on labels among types of brands. J Hum Nutr Diet 2012; 25: 323–333. 34. Van Wezemael L, Caputo V, Nayga RM, Chryssochoidis G, Verbeke W. European consumer preferences for beef with nutrition and health claims: a multi-country investigation using discrete choice experiments. Food Policy 2014; 44: 167–176. 35. Sinclair SE, Cooper M, Mansfield ED. The influence of menu labeling on calories selected or consumed: a systematic review and meta-analysis. J Acad Nutr Diet 2014; 114: 1375–1388. 36. Swartz JJ, Braxton D, Viera AJ. Calorie menu labeling on quick-service restaurant menus: an updated systematic review of the literature. Int J Behav Nutr Phys Act 2011; 8: 135. 37. Temple NJ, Fraser J. Food labels: a critical assessment. Nutrition 2014; 30: 257–260. 38. Emrich TE, Qi Y, Cohen JE, Lou WY, L’Abbe ML. Front-ofpack symbols are not a reliable indicator of products with healthier nutrient profiles. Appetite 2015; 84: 148–153. 39. Ruopeng A. Beverage consumption in relation to discretionary food intake and diet quality among US adults, 2003 to 2012. J Acad Nutr Diet 2015, http://dx.doi.org/10.1016/j.jand.2015.08.009 (accessed 03 December 2015). 40. Cecchini M, Sassi F. Preventing obesity in the USA: impact on health service utilization and costs. Pharmacoeconomics 2015; 33: 765–776. 41. Córvelan C, Reyes M, Garmendia ML, Uauy R. Structural responses to the obesity of non-communicable diseases epidemic: the Chilean law of food labelling and advertising. Obes Rev 2013; 14(Suppl. 2): 79–87. 42. Hawkes C, Smith TG, Jewell J. Smart food policies for obesity prevention. Lancet 2015; 385: 2410–2421.

17, 201–210, March 2016

210 Impact of food labelling M. Cecchini & L. Warin

43. Aschemann-Witzel J, Grunert KG, Van Trijp HCM et al. Effects of nutrition label format and product assortment on the healthfulness of food choice. Appetite 2013; 71: 63–74. 44. Goodman S, Hammond D, Hanning R, Sheeshka J. The impact of adding front-of-package sodium content labels to grocery products: an experimental study. Public Health Nutr 2013; 16: 383–391. 45. Cavanagh KV, Kruja B, Forestell CA. The effect of brand and caloric information on flavor perception and food

17, 201–210, March 2016

obesity reviews

consumption in restrained and unrestrained eaters. Appetite 2014; 82: 1–7. 46. Lahti-Koski M, Helakorpi S, Olli M, Vartiainen E, Puska P. Awareness and use of the heart symbol by Finnish consumers. Public Health Nutr 2012; 15: 476–482. 47. Graham DJ, Laska MN. Nutrition label use partially mediates the relationship between attitude toward healthy eating and overall dietary quality among college students. J Acad Nutr Diet 2012; 112: 414–418.

© 2015 World Obesity