The prevalence of overweight and obesity is increasing at an alarming rate(Reference Shields1, 2). Globally, approximately 1·6 billion adults are overweight and over 400 million are obese(2). Although obesity is more common in high-income countries, increases in obesity have occurred in many low- and middle-income countries, particularly among urban populations(2). The increasing prevalence of overweight and obesity places a considerable burden on public health, including increases in CVD, diabetes, arthritis, sleep and breathing disorders, depression, as well as functional limitations(Reference Luo, Morrison and de Groh3). Diet is also estimated to account for approximately 30 % of cancers in industrialized countries, making it the second largest modifiable risk factor after cigarette smoking(4). The economic burden of overweight and obesity is considerable, with direct health-care costs in the billions for most Western countries(5).
Nutrition labelling on food products has emerged as a prominent policy tool for promoting healthy eating(Reference Cowburn and Stockley6). As a health education intervention, mandatory nutrition labels have broad reach and are present at the point of purchase, as well as when food is prepared or consumed(Reference Feunekes, Gortemaker and Willems7). The display of nutritional information on pre-packaged foods is mandatory in most high-income countries. In the USA, the Nutrition Labelling and Education Act of 1990 mandates that pre-packaged foods carry a nutrition label, with exceptions for foods intended for immediate consumption(8). In Canada, mandatory nutrition labelling was first implemented on pre-packaged foods in December 2005 and became mandatory on virtually all pre-packaged foods in 2007(8–10). Nutrition labelling on pre-packaged foods remains voluntary in the European Union, except in the case of health claims, although mandatory regulations are under development(11) (see Fig. 1 for examples of nutrition labels in the USA, Canada, Australia and the UK).
Fig. 1 Examples of nutrition labels on pre-packaged foods
There is a large and growing evidence base on the impact of nutritional labels, including six literature reviews between 1991 and 2007(Reference Cowburn and Stockley6, Reference Geiger, Wyse and Parent12–Reference Mhurchu and Gorton16). The most recent reviews have focused on specific geographical areas, including European countries(Reference Cowburn and Stockley6, Reference Grunert15) and Australia and New Zealand(Reference Mhurchu and Gorton16), with the exception of Cowburn and Stockley, who reviewed literature up to 2002 across a broader geographical area(Reference Drichoutis14). The findings of these reviews are generally consistent: self-reported use of nutrition labels was found to be prevalent(Reference Cowburn and Stockley6, Reference Geiger, Wyse and Parent12–Reference Mhurchu and Gorton16); however, consumers often report difficulty in interpreting quantitative information contained in labels(Reference Cowburn and Stockley6, Reference Baltas13, Reference Drichoutis14, Reference Mhurchu and Gorton16). Some consumers found different nutrition label formats confusing(Reference Grunert15) and generally preferred graphical information to the traditional label(Reference Geiger, Wyse and Parent12). Label use was more prevalent among the female population(Reference Cowburn and Stockley6, Reference Baltas13–Reference Grunert15), and could be predicted by health awareness(Reference Baltas13), income(Reference Cowburn and Stockley6, Reference Drichoutis14–Reference Mhurchu and Gorton16) and education level(Reference Cowburn and Stockley6, Reference Baltas13–Reference Mhurchu and Gorton16). Finally, one review concluded a positive effect of nutrition labels on diet(Reference Drichoutis14).
The evidence base and regulatory practice have grown considerably in the 7 years since the last systematic review was conducted. Several countries have implemented mandatory nutrition labelling legislation within this time. Furthermore, many of the previous reviews did not include studies on the link between label use and diet(Reference Cowburn and Stockley6). In light of this, the current systematic review aims to examine the existing body of evidence regarding the prevalence of consumer use and understanding of nutrition labels, as well as the impact of nutrition labelling on consumer dietary habits.
Methods
Inclusion criteria
The present review was restricted to studies that examined consumer behaviour related to nutrition labels on pre-packaged foods, published in English in peer-reviewed journals or research reports completed on behalf of government agencies. Studies were included if they examined the prevalence or determinants of nutrition label use, or if they measured consumer knowledge, understanding, perceptions or format preferences related to nutrition labels. Articles that examined the relationship between nutrition label use or legislation and consumer diet were also included. Articles that only examined health claims, food safety labelling, brand naming, package design or shelf labelling were excluded, as were articles that focused on labelling at the point of purchase.
Search strategy
Electronic searches were conducted using the following databases: MEDLINE, CSA Illumina Social Sciences Subject Area (covering forty-six databases); Web of Science (including Science Citation Index Expanded (SCI-EXPANDED) – 1900–present); and the Cochrane Library. Additional searches using the reference lists of relevant articles were also conducted.
The initial search generated a total of 23 801 citations, of which 1450 titles appeared to meet the inclusion criteria and were reviewed. Of these abstracts, 247 were selected for article retrieval. Following review of the full-text articles, 109 were excluded on the basis that they did not meet the inclusion criteria. The remaining 138 articles were assessed using a data extraction form and were rated on eight methodological criteria (see Table 1). Studies were included in the review if they met all of the eight criteria. This led a total of 120 articles being included in the review.
Table 1 Methodological evaluation criteria for including articles in the review
Y, yes; N, no; N/A, not applicable.
Results
The 120 articles selected for review originated from seven jurisdictions: the USA (n 87), Europe (n 13), Canada (n 9), Australia and New Zealand (n 4), Norway (n 2), Germany (n 1), Thailand (n 1) and Trinidad (n 1), as well as one study jointly from the UK and the USA, and one including participants from the Netherlands, Germany, France and the UK. Cross-sectional surveys were the most common study design (n 96), followed by experimental designs (n 19), ‘natural experiments’ (n 7) and longitudinal population-based surveys (n 2). Thirteen of the surveys were based on nationally representative samples and thirty were conducted with individuals who reported being the primary food shopper for their household, or who were approached while shopping for food at the point of purchase.
Prevalence of label use
Of the 120 studies reviewed, sixty-five reported the frequency with which consumers attended or used nutrition labels on pre-packaged foods(Reference Geiger, Wyse and Parent12, Reference Grunert15–Reference Hager, Geiger and Hill79). Among studies targeted at the general population, the prevalence of self-reported label use was generally high (e.g. 82 % in New Zealand(Reference Gorton, Ni Mhurchu and Chen64), 52 % in Canada(80), 47 % in the EU(Reference Drichoutis, Lazaridis and Nayga69) and 75 % in the USA(18)) according to the most recent nationally representative data in each country. Definitions of label ‘use’ varied across studies, complicating comparisons. For example, several studies defined users as those who cite nutrition labels as a source of nutrition information, rather than other sources such as health-care practitioners(Reference Reid, Conrad and Hendricks32, Reference Briley, Owens and Gillham43, Reference Worsley44). Studies also used different time frames for label use, including ‘ever’ use v. use in the past 1 month(Reference Geiger, Wyse and Parent12) or 12 months(Reference Kreuter, Brennan and Scharff53). Overall, these studies indicate that use of nutrition labels among the general population is generally high and typically above 50 %.
Age
A majority of studies found that middle-aged or younger adults were more likely to use nutrition labels than were older individuals(Reference Bender and Derby25, Reference Reid, Conrad and Hendricks32, Reference Mannell, Brevard and Nayga37, Reference Wandel42, Reference Worsley44, Reference Loureiro, Gracia and Nayga47, Reference Drichoutis, Lazaridis and Nayga69–Reference Finke74, 80–Reference Burton and Andrews84), with several exceptions(Reference Coulson20, Reference Satia, Galanko and Neuhouser30, Reference Jensen, Adams and Hollis35, Reference Misra85, Reference Drichoutis, Lazaridis and Nayga86). For example, a large survey of the nutrition perceptions of Americans found that older participants tended to trust nutrition labels as a source of accurate nutrition information to a less extent than younger respondents(Reference Worsley44).
Six articles examined the impact of nutrition labels on adolescents(Reference Huang, Kaur and McCarter36, Reference Hawthorne, Moreland and Griffin54, Reference McCullum and Achterberg87–Reference James, Rienzo and Frazee90). Of these, two studies indicated that use of nutritional labels was low among adolescents(Reference Hawthorne, Moreland and Griffin54, Reference James, Rienzo and Frazee90). Only one 2004 study of youth at an urban primary care clinic in the USA reported a self-reported prevalence rate: 22 % reported ‘always’ reading nutrition labels, 57 % ‘sometimes’ and only 22 % reported ‘never’ reading them(Reference Huang, Kaur and McCarter36). Evaluations of a 2006 US web-based nutrition intervention in adolescents found no improvement in food label use as a result of the intervention(Reference Long, Armstrong and Amos88); however, a similar 2008 study found that web-based interventions increased adolescents' use of labels(Reference Abood, Black and Coster89). A single qualitative study examined the use of nutrition labels among children. The majority of US children in grades 3–6 had difficulty using nutrition labels and could not categorize healthy foods on the basis of label(Reference Lytle, Eldridge and Kotz91).
Gender
Women report using labels significantly more often than men in a majority of studies that include both genders(Reference Bender and Derby25, Reference Satia, Galanko and Neuhouser30, Reference Rasberry, Chaney and Housman33, Reference Jensen, Adams and Hollis35, Reference Wandel42–Reference Worsley44, Reference Peters-Texeira and Badrie46, Reference Kreuter, Brennan and Scharff53, Reference Marietta, Welshimer and Long Anderson59, Reference Gorton, Ni Mhurchu and Chen64, Reference Drichoutis, Lazaridis and Nayga69, Reference Neuhouser, Kristal and Patterson70, Reference Kim, Nayga and Capps72–Reference Finke74, Reference Abbott76, 80, Reference Kim, Nayga and Capps82, Reference Misra85, Reference Jay, Adams and Herring92, Reference Shine, O'Reilly and O'Sullivan93). Women were also more likely to report that nutrition labels had influenced their food choices(Reference Kreuter, Brennan and Scharff53) and to trust nutrition labels(Reference Worsley44). Similar rates of nutrition label use have been documented among women of different levels of income and socio-economic status(Reference Byrd-Bredbenner, Alfieri and Wong34, Reference Michel, Korslund and Finan38, Reference Haldeman, Pérez-Escamilla and Ferris51, Reference Alfieri and Byrd-Bredbenner55, Reference Byrd-Bredbenner, Wong and Cottee57). Only four studies reported no significant difference between male and female participants’ use of nutrition labels(Reference Klopp and MacDonald17, Reference Peters-Texeira and Badrie46, Reference Nayga94, Reference Burton, Garretson and Velliquette95).
Income/education
Most studies have concluded that individuals with lower income are less likely to use nutrition labels(Reference Blitstein and Evans27, Reference Wang, Fletcher and Carley62, Reference Kim, Nayga and Capps82, Reference Nayga, Lipinski and Savur96), with only two studies finding the opposite effect(Reference Drichoutis, Lazaridis and Nayga67, Reference Drichoutis, Lazaridis and Nayga86) and one reporting no significant effect of income(Reference Jensen, Adams and Hollis35). Individuals with lower income were also more likely to have lower levels of nutrition knowledge(Reference Drichoutis, Lazaridis and Nayga86, Reference Nayga94), which were associated with label non-use(Reference Elbon, Johnson and Fischer50, Reference Guthrie, Fox and Cleveland63, 80, Reference Hess, Yanes and Jourdan97). Similar effects have been observed for education levels: individuals with greater education have reported greater use of nutrition labels in most studies(Reference Klopp and MacDonald17, Reference Satia, Galanko and Neuhouser30, Reference McArthur, Chamberlain and Howard41, Reference Wandel42, Reference Worsley44, Reference Wang, Fletcher and Carley62, Reference Guthrie, Fox and Cleveland63, Reference Drichoutis, Lazaridis and Nayga67, Reference Neuhouser, Kristal and Patterson70, Reference Wandel and Bugge73, Reference Abbott76, 80, Reference Drichoutis, Lazaridis and Nayga86, Reference Jay, Adams and Herring92, Reference Shine, O'Reilly and O'Sullivan93, Reference Nayga, Lipinski and Savur96, Reference Hess, Yanes and Jourdan97), with only two exceptions(Reference Jensen, Adams and Hollis35, Reference Nayga94). Seven studies targeting socio-economically disadvantaged populations reported variable rates of nutrition label use, ranging from 20 % to 74 %(Reference Perez-Escamilla, Himmelgreen and Bonello23, Reference Michel, Korslund and Finan38, Reference Hyman, Simons-Morton and Ho39, Reference McArthur, Chamberlain and Howard41, Reference Haldeman, Pérez-Escamilla and Ferris51, Reference Signal, Lanumata and Robinson61, Reference Jay, Adams and Herring92), although these rates were typically lower than those reported for the general population.
Mixed findings were observed with respect to the effect of employment(Reference Drichoutis, Lazaridis and Nayga67, Reference Abbott76, Reference Nayga94, Reference Nayga, Lipinski and Savur96), job satisfaction(Reference Drichoutis, Lazaridis and Nayga69) and rural v. urban habitation on label use(Reference Jensen, Adams and Hollis35, Reference Mannell, Brevard and Nayga37, Reference Wang, Fletcher and Carley62, Reference Kim, Nayga and Capps82, Reference Lin, Lee and Yen98, Reference Ward and Jauregui99). Only one study of older Americans in 1990 directly compared rural with urban groups, with no significant difference in label use(Reference Briley, Owens and Gillham43). Larger households and those with children were found to more likely use labels(Reference McArthur, Chamberlain and Howard41) and support their mandatory implementation(Reference Jensen, Adams and Hollis35, Reference Mannell, Brevard and Nayga37), as were married couples(Reference Blitstein and Evans27).
Race/ethnicity
A majority of studies have found that Caucasian participants are significantly more likely to use nutrition labels than are other ethnic groups(Reference Dooley, Novotny and Britten21, Reference Hyman, Simons-Morton and Ho39, Reference Kim and Douthitt71, Reference Gans, Burkholder and Risica100), with one study of African-American adults in North Carolina reporting high levels of use(Reference Satia, Galanko and Neuhouser30). Studies with Latino adults in the USA also reported lower rates of label use(23,51). For example, a study comparing ethnic groups in the USA found that only half of the proportion of Latinos, compared with their African-American and ‘white’ respondents, had ‘ever used’ labels(Reference Hyman, Simons-Morton and Ho39). Low rates have also been observed among ethnic minorities in New Zealand(Reference Signal, Lanumata and Robinson61). Racial/ethnic differences have also been observed with respect to the type of information sought from nutrition labels(Reference Bender and Derby25). For example, Latinos were found to be more likely to check dietary fibre and Na information(Reference Kim, Nayga and Capps72).
Health behaviours
A wide range of studies have examined the association between label use and health practices. Individuals with healthier eating habits report greater use of nutrition labels, either as a result of personal preference(Reference Bender and Derby25, Reference Satia, Galanko and Neuhouser30, Reference Elbon, Johnson and Fischer50, Reference Kreuter, Brennan and Scharff53, 80) or because of the requirements of a health-related diet(Reference Bender and Derby25, Reference Mannell, Brevard and Nayga37, Reference McArthur, Chamberlain and Howard41, Reference Briley, Owens and Gillham43, Reference Gorton, Ni Mhurchu and Chen64, Reference Drichoutis, Lazaridis and Nayga67, Reference Lin and Yen68, Reference Abbott76, Reference Drichoutis, Lazaridis and Nayga86, Reference Nayga, Lipinski and Savur96, Reference Krystallis and Ness101). Greater use has also been reported by individuals more concerned with dietary guidelines(Reference Reid, Conrad and Hendricks32, Reference Rasberry, Chaney and Housman33, Reference Jensen, Adams and Hollis35, Reference McArthur, Chamberlain and Howard41, Reference Kreuter, Brennan and Scharff53, Reference Guthrie, Fox and Cleveland63, Reference Drichoutis, Lazaridis and Nayga86, Reference Nayga94, Reference Nayga, Lipinski and Savur96) and by those who place greater emphasis on the nutritional quality of food while shopping(Reference Jensen, Adams and Hollis35, Reference Lin and Yen68, Reference Kim, Nayga and Capps72, Reference Nayga94, Reference Nayga, Lipinski and Savur96, Reference Lin, Lee and Yen98, Reference Higginson, Kirk and Rayner102). Nutrition and label knowledge(Reference Klopp and MacDonald17, Reference Shine, O'Reilly and O'Sullivan31, Reference Jensen, Adams and Hollis35, Reference Marietta, Welshimer and Long Anderson59, 80, Reference Drichoutis, Lazaridis and Nayga86, Reference Lin, Lee and Yen98, Reference Fitzgerald, Damio and Segura-Pérez103), nutrition education(Reference Kessler and Wunderlich19, Reference Kessler and Wunderlich40, Reference McArthur, Chamberlain and Howard41) and knowledge of diet–disease relationships(Reference Klopp and MacDonald17, Reference Szykman, Bloom and Levy29, Reference Shine, O'Reilly and O'Sullivan31, Reference Rasberry, Chaney and Housman33, Reference Smith, Taylor and Stephen60, Reference Kim and Douthitt71, Reference Kim, Nayga and Capps72, Reference Kim, Nayga and Capps82) or of specific diseases(Reference Szykman, Bloom and Levy29) have also been associated with label use, with few exceptions(Reference Drichoutis, Lazaridis and Nayga69, Reference Misra85, Reference Nayga94). Weight control(Reference Satia, Galanko and Neuhouser30, Reference Rasberry, Chaney and Housman33) and diagnosis of a disease(Reference Satia, Galanko and Neuhouser30, Reference McArthur, Chamberlain and Howard41, Reference Kreuter, Brennan and Scharff53, Reference Drichoutis, Lazaridis and Nayga69, Reference Hager, Geiger and Hill79, Reference Kim, Nayga and Capps82, Reference Variyam and Cawley104–Reference Miller, Probart and Achterberg106) have also been associated with greater label use.
Grocery shopping habits have been identified as being a strong predictor of nutritional label use. Consumers who spend more time, or report having more time to shop for groceries, were more likely to be label users(Reference Drichoutis, Lazaridis and Nayga86, Reference Nayga, Lipinski and Savur96, Reference Lin, Lee and Yen98), and lack of time was consistently reported as a reason for non-use(Reference Klopp and MacDonald17, Reference Rasberry, Chaney and Housman33, Reference Mannell, Brevard and Nayga37). Shoppers who placed less emphasis on price(Reference Drichoutis, Lazaridis and Nayga67, Reference Drichoutis, Lazaridis and Nayga86, Reference Nayga, Lipinski and Savur96) were more likely to use nutrition labels, although one study reported no association(Reference Jensen, Adams and Hollis35). The importance of taste was positively related to label use in three studies(Reference Jensen, Adams and Hollis35, Reference Nayga94, Reference Nayga, Lipinski and Savur96) and was negatively related by two others(Reference Guthrie, Fox and Cleveland63, Reference Drichoutis, Lazaridis and Nayga86). Meal planning(Reference Lin and Yen68, Reference Kim, Nayga and Capps72, Reference Drichoutis, Lazaridis and Nayga86) and grocery spending were other variables related to use(Reference Wang, Fletcher and Carley62).
Among health behaviours not directly related to nutrition, using supplements, exercising regularly and not smoking were associated with the use of nutrition labels(Reference Kessler and Wunderlich19, Reference Satia, Galanko and Neuhouser30, Reference Kessler and Wunderlich40, Reference Lin and Yen68, Reference Misra85).
Attitudes/perceptions towards nutrition labels
Many consumers have reported that nutrition labels are an important source of information(Reference Lindhorst, Corby and Roberts22, Reference Smith, Taylor and Stephen60, Reference Krystallis and Ness101, Reference Silayoi and Speece107, Reference Reid and Hendricks108), although ingredients and health claims may be perceived as more important(Reference Reid and Hendricks108). Most consumers were willing to use information if it was provided on the label(Reference Hess, Yanes and Jourdan97), although consumers’ beliefs about the healthiness of foods did not necessarily depend on information on the label(Reference Aikman, Min and Graham109). There was, however, popular support for mandatory labelling in studies, although conflicting findings have been found for consumers’ willingness to pay extra for nutrition information(Reference Daly81). Positive attitudes were higher among individuals reporting greater use of labels(Reference Byrd-Bredbenner56, Reference Marietta, Welshimer and Long Anderson59, Reference Misra85, Reference Reid and Hendricks108, Reference Byrd-Bredbenner and Kiefer110); however, negative attitudes were also prevalent in the literature(Reference McArthur, Chamberlain and Howard41, Reference James48, Reference Miller and Brown111, Reference Tessaro, Rye and Parker112). Many consumers believed that serving sizes and health claims were misleading and were sceptical of the compliance of labels to regulatory law(Reference Misra85, Reference Sullivan113). The credibility of manufacturers’ health claims was rated poorly, especially when these claims contradicted nutrition information on the label(Reference Marietta, Welshimer and Long Anderson59, Reference Misra85, Reference Keller, Landry and Olson114–Reference Balasubramanian and Cole116); however, in one case, health claims helped consumers to choose more nutritious products(Reference Kozup, Creyer and Burton117). Trust in labels also predicted use(Reference Worsley44, Reference Smith, Taylor and Stephen60, Reference Drichoutis, Lazaridis and Nayga69), and was greater among younger respondents and among those with higher levels of education(Reference Worsley44).
Comprehension and understanding of nutrition labels
Studies suggest that consumers generally find nutrition labels to be useful(Reference Klopp and MacDonald17, Reference Hawthorne, Moreland and Griffin54, Reference Marietta, Welshimer and Long Anderson59, Reference Misra85), although consumers in the USA(Reference Heimbach and Stokes118) and Australia(Reference Crawford and Baghurst26) report a desire for simpler presentation of information(Reference Crawford and Baghurst26, Reference Heimbach and Orwin49, Reference Heimbach and Stokes118). In one case, Australian participants requested more detailed information(Reference Crawford and Baghurst26). Following the Nutrition Labeling and Education Act in the USA, which implemented a consistent label format in 1993, 80 % of consumers thought that the label was more helpful and the proportions of those seeking more information declined, except with respect to cholesterol information(Reference Kristal, Levy and Patterson28).
There is mixed evidence with respect to the ease(Reference Satia, Galanko and Neuhouser30, 80, Reference Shine, O'Reilly and O'Sullivan93) or difficulty of using nutrition labels(Reference Crawford and Baghurst26, Reference Silayoi and Speece107, Reference Heimbach and Stokes118). Frequent label use was associated with better understanding in general(Reference Drichoutis, Lazaridis and Nayga69, Reference Lin, Lee and Yen98, Reference Goldberg, Probart and Zak119), with other studies providing mixed results(Reference Byrd-Bredbenner56, Reference Levy and Fein120). Younger participants(Reference Kreuter, Brennan and Scharff53), as well as those with higher education(Reference Rothman, Housam and Weiss52, Reference Kreuter, Brennan and Scharff53), income, literacy and numeracy(Reference Rothman, Housam and Weiss52), were more likely to report understanding nutrition labels. One longitudinal study found that self-reported awareness of nutrition terms, but not understanding, improved between 1984 and 1994 in Canada(Reference Reid, Conrad and Hendricks32).
Several studies reported a good understanding of nutrition labels based on consumers’ performance on tasks requiring them to retrieve or manipulate information(Reference Byrd-Bredbenner56, Reference Burton, Garretson and Velliquette95, Reference Levy, Fein and Schucker121). Understanding was greater in younger(Reference Byrd-Bredbenner and Kiefer110), female, educated and white participants(Reference Levy, Fein and Schucker121), and was also related to knowledge(Reference Burton, Garretson and Velliquette95, Reference Burton, Biswas and Netemeyer122, Reference Howlett, Burton and Kozup123), perceived understanding(Reference Burton, Garretson and Velliquette95), attitude towards and motivation to use the nutrition label(Reference Burton, Garretson and Velliquette95, Reference Howlett, Burton and Kozup123), as well as frequent label use(Reference Drichoutis, Lazaridis and Nayga69, Reference Lin, Lee and Yen98, Reference Goldberg, Probart and Zak119). Self-reported understanding is generally high among lower-income groups(Reference Michel, Korslund and Finan38, Reference McArthur, Chamberlain and Howard41, Reference Haldeman, Pérez-Escamilla and Ferris51); however, with the exception of one sample of individuals eligible for a US food supplementation programme(Reference McArthur, Chamberlain and Howard41), most showed poor performance on items measuring their ability to use the label, especially when calculations were required(Reference Michel, Korslund and Finan38, Reference Haldeman, Pérez-Escamilla and Ferris51). Low perceived self-efficacy in using the label was also reported by women on social assistance(Reference Sullivan113).
A variety of studies indicate that many consumers have difficulty with the quantitative information presented on labels, especially with respect to recommended daily amounts, per cent daily values, serving sizes or other forms of reference information on the label(Reference Klopp and MacDonald17, Reference Dooley, Novotny and Britten21, Reference Shine, O'Reilly and O'Sullivan31, Reference Rothman, Housam and Weiss52, Reference Hawthorne, Moreland and Griffin54, Reference Daly81, Reference Misra85, Reference Levy and Fein120, 124). This difficulty was common among diabetics(Reference Miller, Probart and Achterberg106), chronic kidney disease patients(Reference Hager, Geiger and Hill79), older adults(Reference Daly81, Reference Byrd-Bredbenner and Kiefer110, Reference Block and Peracchio125), adolescents(Reference Hawthorne, Moreland and Griffin54), infrequent label users(Reference Klopp and MacDonald17) and those with less education(Reference Daly81).
Other tasks that were reportedly confusing for consumers included comparisons between products(Reference Rothman, Housam and Weiss52), determination of energy per serving and per package(Reference Pelletier, Chang and Delzell126) and comprehension of E-numbers representing additives(Reference Wandel42). For example, 24 % of consumers in Trinidad read nutrition labels without understanding them(Reference Peters-Texeira and Badrie46), and this was listed as a reason for non-use among many groups of consumers(Reference Mannell, Brevard and Nayga37, Reference Peters-Texeira and Badrie46, Reference Shine, O'Reilly and O'Sullivan93, Reference Block and Peracchio125). Several studies conducted among females in the USA and UK have also provided mixed evidence, showing that most participants could locate nutrition information, but had difficulties with per cent daily value and information on food claims(Reference Byrd-Bredbenner, Alfieri and Wong34, Reference Alfieri and Byrd-Bredbenner55, Reference Byrd-Bredbenner, Wong and Cottee57). Frequent label reading, better education, better self-assessments of diet quality, health status and nutrition knowledge were related to these skills(Reference Alfieri and Byrd-Bredbenner55). Educational interventions targeting label knowledge and understanding have generally shown positive results in a range of sub-populations(Reference Dooley, Novotny and Britten21, Reference Lindhorst, Corby and Roberts22, Reference Hawthorne, Moreland and Griffin54, Reference Block and Peracchio125, Reference Miller, Edwards and Kissling127, Reference Li, Miniard and Barone128), including among low-income and literacy groups(Reference Jay, Adams and Herring92).
Label format and content
Compared with ‘traditional’ nutrition labels with quantitative information on nutrient content, several studies have reported greater effectiveness for labels using graphics and symbols(Reference Marino and Mahan129–Reference Geiger, Wyse and Parent131), adjective labels(Reference Levy, Fein and Schucker132) and labels with minimal numerical content(Reference Abbott76). For example, information accompanied by graphics helped consumers to better apply reference information, especially consumers who had not seen labels before(Reference Signal, Lanumata and Robinson61). The use of well-recognized health symbols(Reference Feunekes, Gortemaker and Willems7, Reference Miller, Probart and Achterberg106) and ‘traffic lights’ may be particularly effective(Reference Gorton, Ni Mhurchu and Chen64, Reference Jones and Richardson133, Reference Kelly, Hughes and Chapman134). For example, traffic light symbols – which typically display green, amber or red labels to indicate whether foods contain low, medium or high amounts of contents such as fat, saturated fat, sugars – have been found to increase consumer ability to identify healthier food options and consumer attention in general(Reference Jones and Richardson133, Reference van Kleef, van Trijp and Paeps135, Reference Borgmeier and Westenhoefer136).
Research also suggests that placing nutrition information on the front of packages is more effective than information positioned on the side or back of packages(Reference Grunert15, Reference McCullum and Achterberg87, Reference Wansink137). Front-of-pack (FOP) labels may disproportionately benefit those with low-nutrition education and knowledge of nutrition labels(Reference Grunert15). For example, in a 2009 study conducted in Australia, consumers supported the idea of FOP labelling, especially when it is consistent across products and manufacturers(Reference Kelly, Hughes and Chapman134). Simple energy information on the FOP was also well received in a 2007 study conducted in Germany, The Netherlands, France and the UK(Reference van Kleef, van Trijp and Paeps135). Preliminary evidence suggests that FOP labels may also promote healthier food purchasing behaviours, although additional research is needed(Reference Feunekes, Gortemaker and Willems7, Reference Grunert15).
Evidence is mixed with respect to the level of detail or complexity of information favoured by consumers. More detailed information was favoured by some consumers(Reference Abbott76, Reference Hager, Geiger and Hill79, Reference Geiger, Wyse and Parent131), especially non-label users(Reference Klopp and MacDonald17), whereas frequent users preferred less detail(Reference Klopp and MacDonald17). Simplified labels have been shown to promote more accurate nutrition judgements of unhealthy products(Reference Finke74, Reference Burton and Andrews84, Reference Goldberg, Probart and Zak119) and improved performance on diet-related tasks(Reference Levy, Fein and Schucker121, Reference Geiger, Wyse and Parent131), even when daily reference values were added(Reference Levy, Fein and Schucker121). The use of reference information, such as per cent daily value, is often welcomed by consumers(Reference Burton, Biswas and Netemeyer122, Reference Jones and Richardson133, Reference Kelly, Hughes and Chapman134); however, many struggle to apply the quantitative values(Reference Feunekes, Gortemaker and Willems7, 124, Reference Barone, Rose and Manning138). Labels presenting information in two columns side-by-side have also been shown to reduce food consumption by non-dieters when compared with a version presenting information as a single, longer column(Reference Antonuk and Block139). Consumers have also expressed a desire for nutrient information listed in the context of a healthy diet(Reference Shine, O'Reilly and O'Sullivan93, Reference Geiger, Wyse and Parent131); larger, more legible print(Reference Mannell, Brevard and Nayga37, Reference Lewis and Yetley130); simpler terms(Reference Abbott76); explanations of terms or nutrients(Reference Mannell, Brevard and Nayga37, Reference Shine, O'Reilly and O'Sullivan93); the use of colour and a consistent appearance across nutrition labels(Reference Shine, O'Reilly and O'Sullivan93).
Types of nutrition information sought by consumers
Consumers tend to look more closely at nutrients they wish to avoid(Reference Shine, O'Reilly and O'Sullivan93). To this end, the nutrients most commonly sought were fat(Reference Crawford and Baghurst26, Reference Kristal, Levy and Patterson28, Reference Mannell, Brevard and Nayga37, Reference Wandel42, Reference Heimbach and Orwin49, Reference Kreuter, Brennan and Scharff53, Reference Wandel and Bugge73, Reference Hager, Geiger and Hill79, Reference Misra85, Reference Hess, Yanes and Jourdan97, Reference Reid and Hendricks108, Reference Garretson and Burton115, Reference Heimbach and Stokes118), energy content(Reference Kristal, Levy and Patterson28, Reference Mannell, Brevard and Nayga37, Reference Kreuter, Brennan and Scharff53, Reference Hager, Geiger and Hill79, Reference Hess, Yanes and Jourdan97, Reference Heimbach and Stokes118), protein(Reference Heimbach and Orwin49, Reference Hager, Geiger and Hill79, Reference Hess, Yanes and Jourdan97, Reference Heimbach and Stokes118), cholesterol(Reference Kristal, Levy and Patterson28, Reference Hess, Yanes and Jourdan97), carbohydrates(Reference Wandel42, Reference Heimbach and Stokes118), vitamins and minerals(Reference Hess, Yanes and Jourdan97, Reference Heimbach and Stokes118), types of fat(Reference Wandel42, Reference Hess, Yanes and Jourdan97), serving size(Reference Misra85), additives(Reference Wandel42, Reference Wandel and Bugge73) and Na information(Reference Wandel42, Reference Hess, Yanes and Jourdan97, Reference Heimbach and Stokes118). Low-fat dieters were more likely to look at fat information(Reference Wandel42, Reference Kreuter, Brennan and Scharff53, Reference Lin, Lee and Yen98), and younger(Reference Drichoutis, Lazaridis and Nayga67, Reference Wandel and Bugge73) female participants were more likely to look at energy than men(Reference Marietta, Welshimer and Long Anderson59, Reference Drichoutis, Lazaridis and Nayga67). Cholesterol was most often looked at by older(Reference Elbon, Johnson and Fischer50, Reference Drichoutis, Lazaridis and Nayga67, Reference Lin, Lee and Yen98), suburban participants who believed in a diet–disease relationship(Reference Lin, Lee and Yen98) and had high cholesterol(Reference Kreuter, Brennan and Scharff53), and less often looked at by white, well-educated individuals with low cholesterol intake(Reference Lin, Lee and Yen98). Those with experience reading labels were more likely to use carbohydrates and fibre information(Reference Marietta, Welshimer and Long Anderson59), and younger individuals were more likely to use vitamin and mineral information(Reference Drichoutis, Lazaridis and Nayga67).
Similar to the general population, adolescents were most likely to seek fat and energy information(Reference McCullum and Achterberg87). A range of studies have also examined information sought by a range of other sub-populations, including low-income women(Reference Michel, Korslund and Finan38, Reference McArthur, Chamberlain and Howard41), Latino populations(Reference Kim, Nayga and Capps72) and diabetes patients(Reference Miller, Probart and Achterberg106).
The impact of nutrition labels on diet
Observational studies have consistently found an association between use of nutrition labels and healthier diets(Reference Neuhouser, Kristal and Patterson70–Reference Kim, Nayga and Capps72, Reference Kim, Nayga and Capps82, Reference Fitzgerald, Damio and Segura-Pérez103, Reference Nayga140, Reference Kristal, Hedderson and Patterson141). Several studies have reported an association between label use and lower fat consumption(Reference Neuhouser, Kristal and Patterson70–Reference Kim, Nayga and Capps72, Reference Kim, Nayga and Capps82, Reference Nayga140, Reference Kristal, Hedderson and Patterson141). Label users are also more likely to eat healthier varieties of foods(Reference Nayga140), and to have reduced Na(Reference Kim, Nayga and Capps72, Reference Fitzgerald, Damio and Segura-Pérez103), cholesterol(Reference Guthrie, Fox and Cleveland63, Reference Kim, Nayga and Capps72) and energy intakes, coupled with increased fibre(Reference Kim, Nayga and Capps72, Reference Variyam142), Fe(Reference Variyam143) and vitamin C intakes(Reference Neuhouser, Kristal and Patterson70). Cross-sectional associations between label use and healthier diets are also related to socio-economic status(Reference Variyam142), education(Reference Kim and Douthitt71, Reference Kim, Nayga and Capps82), age(Reference Kim and Douthitt71, Reference Kim, Nayga and Capps72) gender(Reference Huang, Kaur and McCarter36, Reference Kim, Nayga and Capps72) and ethnicity/race(Reference Kim and Douthitt71, Reference Kim, Nayga and Capps72).
Three longitudinal studies in the USA have evaluated the implementation of new nutrition labels on dietary patterns. In the USA, the 1990 Nutrition Labeling and Education Act came into effect in 1994 and required nutrition labels on all pre-packaged foods. A study comparing nationally representative surveys of consumers in 1989 and 1995 found that frequent label users in 1995 had a significantly greater probability of consuming a low-fat diet than both non-label users in 1995 and frequent label users in 1989(Reference Finke74). In addition, fat intake among less-educated respondents decreased significantly during the ‘pre–post’ study period(Reference Finke74). A second study found that BMI of nutrition label users fell significantly following implementation of the Act, with the greatest change among those with the highest BMI score(Reference Variyam and Cawley104). In addition, low-fat and low-Na food purchases increased significantly following the impact of new labels, although the same effect was not observed for low-energy choices, or healthy nutrients such as vitamins and minerals(Reference Balasubramanian and Cole116).
Evidence from five experimental studies is generally consistent with cross-sectional and longitudinal findings. Two experiments compared consumption of low-fat with energy-dense foods by randomizing participants to either a blind or information condition(Reference Miller, Castellanos and Shide144, Reference Kral, Roe and Rolls145). Both studies found that, although participants tended to consume greater amounts of reduced-energy food in terms of food weight, total energy intake was significantly lower among those who consumed reduced-energy food(Reference Miller, Castellanos and Shide144, Reference Kral, Roe and Rolls145). Only one study showed this effect on daily energy consumption, as opposed to short-term intake during the study(Reference Kral, Roe and Rolls145). Participants who received nutrition information consumed more of the low-energy version of the food(Reference Miller, Castellanos and Shide144). A third experiment found no differences in participants’ satiety after consuming fat-free compared with regular potato chips, irrespective of the provision of information(Reference Miller, Bell and Pelkman146). Finally, providing nutrition information also increased healthier purchase intentions and accurate perceptions of nutrient content(Reference Baixauli, Salvador and Hough147).
Discussion
Research conducted to date indicates that nutrition labels on pre-packaged foods are among the most prominent sources of nutrition information. Evidence also suggests that consumers perceive nutrition labels to be a highly credible source of information, and many consumers report using nutrition labels to guide their selection of food products.
The use of nutrition labels varies considerably across population subgroups. Use is particularly high among individuals with health conditions and special dietary requirements – those with the greatest need for nutritional information. However, label use is notably lower among children, adolescents and older adults. More research targeting these populations is needed, given their increased prevalence of obesity(Reference Shields1, 148), nutrient deficiencies(Reference Ferrini and Ferrini149) and chronic disease(Reference Ferrini and Ferrini149–Reference Ferrini and Ferrini153). Individuals with lower socio-economic status are also less likely to use nutrition labels, which is particularly problematic given that low socio-economic status is associated with an increased risk of being overweight and obese(Reference Mackenbach154).
The evidence in this review shows a consistent link between the use of nutrition labels and healthier diets. The causal nature of this association is likely bidirectional: nutrition labels may promote healthier eating, whereas individuals with healthier diets are more likely to seek out nutritional labels in the first place. However, there is sufficient evidence from a range of study designs to conclude that providing nutrition information on packages has a positive impact on diet. In countries such as the USA, government agencies and non-government organizations have estimated the impact of mandatory nutrition labelling to be in the range of billions of dollars(Reference Variyam and Cawley104), although the magnitude of benefit and the extent to which it varies across different types of nutrition labels and population subgroups cannot be estimated with any precision from the existing evidence base.
Research to date also highlights the need to balance the complexity of information presented on labels with consumers’ ability to process this information in a quick and meaningful manner. Nutrition labels that require calculations with respect to nutrient amounts and serving sizes are confusing to many consumers, particularly those with lower education and literacy skills(Reference Nayga, Lipinski and Savur96). Educational interventions aimed at improving the understanding of nutrition labels have shown promise and a broader application of these interventions may provide one potential solution(Reference Jay, Adams and Herring92, Reference Carson and Hedl155); however, the evidence highlights the need to improve the ways in which nutrition information is presented to consumers on food packages.
Future research should examine the effectiveness of using symbols, images and different graphical layouts to a greater extent. Indeed, there is growing evidence regarding the consumer-friendly nature symbols used by the industry, as well as the greater impact of FOP labels, compared with labels on the side or back of packages. These formats may be more consumer-friendly in part because nutrition information is more accessible and in part because of a widespread desire for more ‘prescriptive’ information that identifies ‘healthier’ food from less-healthy options. Indeed, an expert panel commissioned by the UK Food Standards Agency recently concluded that FOP formats are effective and the strongest FOP label is one that combines the use of words ‘high, medium and low’, traffic light colours and percentage of Guideline Daily Amount, in addition to levels of nutrients in a portion of the product(Reference Malam, Cleeg and Kirwan156).
More generally, there is increasing evidence that labelling regulations need to take the entire package into consideration to maximize their effectiveness. Industry ‘health claims’ are regulated to different extents across jurisdictions and the use of FOP symbols, which imply healthier alternatives, is largely unregulated. Ideally, consumers would use nutrition labels to help interpret health claims; however, in practice, many consumers rely solely on health claims(Reference Reid and Hendricks108). Indeed, there have been mixed reports as to whether consumers can determine whether claims are truthful(Reference Byrd-Bredbenner56, Reference Keller, Landry and Olson114, Reference Levy and Fein120, Reference Reid and Hendricks157).
Limitations
This review is subject to several limitations. First, it is possible that relevant articles were not included in the review, given the rapidly evolving evidence base. Attempts were made to minimize this limitation by using a comprehensive searching strategy and a systematic selection process using two independent reviewers and inclusion criteria. Second, the articles included in this review were disproportionately from high-income Western countries, and from North America in particular. Therefore, it is unclear as to what extent the findings in this review apply to jurisdictions with different labelling regulations and in much different cultural and geographical contexts. Additional research on the impact of nutrition labels in low- and middle-income countries should be considered a priority. Even among the Western countries included in this review, there are important differences in labelling regulations that were not fully examined. The diversity in study protocols, measures and samples also presents challenges in terms of comparing studies. We have tried to note major differences wherever possible; however, it is likely that methodological differences between studies account for at least some of the variability in the findings. Finally, much of the evidence on the impact of nutrition labels is based on self-report data, which may over-report the use of nutrition labels, meaning that other factors, such as greater awareness of the link between nutrition and chronic disease, may be responsible for population-level changes over time that have been attributed to nutrition labels.
Conclusions
Population-level interventions and changes to the food environment are necessary to halt the rising health and economic burden from obesity. The evidence to date indicates that nutrition labels on pre-packaged foods are a cost-effective population-level intervention with considerable reach. In order to capitalize upon the potential of nutrition labels, governments will need to explore new formats and different types of information content to ensure that nutrition information is accessible and understandable. A number of jurisdictions are in the process of developing new formats and revising labelling standards, such as the European Union(158). There is an immediate need for evidence to inform these regulatory developments. Regulators should also consider expanding the scope of mandatory nutrition labelling. In the vast majority of jurisdictions, nutrition labelling regulations are limited to pre-packaged food products and do not apply to foods served in restaurants or fast-food outlets, which account for a significant proportion of dietary intake in many high-income countries(Reference Guthrie, Lin and Frazao159, 160). Mandatory display of nutrition information on menus and menu boards of food outlets may be a promising means of increasing the impact of nutrition labelling regulations(Reference Berman and Lavizzo-Mourey161) and harmonizing nutrient information across information channels.
Acknowledgements
Funding support for this manuscript was provided by the Propel Centre for Population Health Impact, with funds from the Canadian Cancer Society. The authors have no conflict of interest to declare. Each of the authors made a direct contribution to this manuscript. D.H. conceived of the study; J.D. and S.C. conducted the article searching; and J.D., S.C. and D.H. co-authored the manuscript. The authors acknowledge the assistance of Priya Kekree and Samantha Daniel in preparing the manuscript for publication.
