Effect of nutrition labels on dietary quality among college students: a systematic review and meta-analysis

Effect of nutrition labels on dietary quality among college students: a systematic review and... Abstract Context College students are at an elevated risk of poor nutrition and eating habits. Objective The aim of this systematic review was to examine and quantify the effect of nutrition labels on diet quality in college students. Data Sources Literature searches were conducted in 4 electronic databases. Study Selection Peer-reviewed publications that assessed the effect of nutrition label use on food choice or dietary intake in college students were included. Data Extraction Twenty-two randomized controlled trials, cohort studies, and pre–post studies were identified. Results Sixteen studies found label exposure to be associated with improved diet. Of the 13 studies reporting calories selected or consumed, 8 found that posting labels at the point of purchase decreased calories, 4 found no effect, and 1 found that calories consumed increased after posting labels. Nine of the 12 studies assessing noncaloric measures found that nutrition labels positively affected diet quality. Meta-analysis of pre–post studies found a decrease of 36 kcal (P < 0.05) with label exposure. Conclusions Nutrition labels had a moderate but positive effect on dietary intake of college students. college, diet, menu labels, nutrition labels, university INTRODUCTION Food purchase and consumption outside the home has risen in the last 30 years,1,2 now accounting for almost 50% of the food expenditures among Americans.3 Since greater frequency of eating outside the home has been associated with higher body weight,4 labeling of calorie and nutrition information on restaurant menus has emerged as a tool to enable consumers to make informed food selections. However, recent reviews have questioned the efficacy of posting calorie information, indicating the lack of effect on calorie purchase or consumption.5–7 The null effect could partially be due to differences in populations studied and the wide variety of study designs and outcomes assessed. Further, the motivation and reasoning behind label usage is unclear, as several survey-based studies found that individuals who reported using nutrition labels had higher nutrition knowledge or motivation to eat in a specific way8,9 or to lose weight,10 whereas others found nutrition labels to be effective across consumer characteristics,11–13 or even more effective for those with lower health consciousness.14 Besides being affected by consumer characteristics, research on the effect of nutrition labels may be confounded by the fact that certain types of labels tend to be more effective than others. A recent meta-analysis reported that, while standard calorie labeling did not affect calorie selection or consumption, both outcomes decreased when labels included contextual information such as daily intake or traffic light symbols.15 Interpretational aids and cues help consumers better understand and use labels,16 and studies have shown that traffic light labels in particular are more effective than simple textual labels at decreasing calorie selection.17,18 Another major gap in understanding the effect of nutrition labels lies within the limited number of dietary outcomes assessed. While several reviews have focused on calories selected and consumed,6,15,19 fewer have used broader definitions of dietary intake. Those that have observed broader measures are generally limited by including only a small number of studies.20,21 Since diet measures vary widely across studies, pooling results to assess the overall effect is often infeasible. College students are a population at an elevated risk of poor nutrition and eating habits. Dietary quality22 and fruit and vegetable intake tend to decrease during emerging adulthood.23 Further, a meta-analysis estimated that college students gain about 4 pounds on average during their freshman year.24 While previous studies have reviewed prevalence and predictors of nutrition label use9 and the effects of dietary interventions among college students,25–27 the effect of nutrition labels in particular is unknown. Reviewing the effectiveness of nutrition labels in a group with relatively homogenous age and education may allow more precise identification of factors interacting with label use and diet quality. A systematic review of the effect of nutrition label use on diet quality among college students was conducted. The hypothesis that nutrition label use could improve diet quality by helping college students make more healthful food choices was tested. In addition, a meta-analysis was conducted to quantify the influence of nutrition label use on dietary quality. METHODS Systematic review and meta-analysis procedures were conducted in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines.28 Study selection criteria Studies were included in this review if the following criteria were met: (1) study adopted a design of a randomized control trial, a cohort study, or a pre–post study; (2) study took place on a college campus; (3) study assessed nutrition label use (nutrition facts, nutrition labels specific to foods in cafeterias or dining units, or general nutrition labels) as a predictor of diet quality or food choice; (4) study assessed actual food choice or dietary intake of more than 1 food item as the outcome; (5) study was published in a peer-reviewed publication; (6) study was written in English; and (7) study was published on or before May 18, 2017. PICOS (population, intervention, comparator, outcome, setting) criteria are listed in Table 1. Studies adopting a qualitative, case report, case–control, or cross-sectional design, published in a language other than English, not peer reviewed, not occurring on a college campus, requiring participants to read labels as a prerequisite for participation (including labels that were not nutrition labels), not providing a direct test of the relationship between label use and dietary quality or food choice independent of other factors such as food availability or education, using experiments that measured intake of only 1 item (eg, a snack bar), or reporting hypothetical choices or intentions rather than actual food choice or diet quality were excluded. Additionally, studies wherein the average age of participants was over 30 years or the majority of participants were over 30 years old were excluded. In the case of studies that met the inclusion criteria but also had additional interventions such as price changes,29,30 downsizing offers,31 a social media campaign,32 or food taxes,33 only the results of the labeling intervention were summarized. Table 1 PICOS criteria for inclusion of studies Parameter  Inclusion criteria  Population  College students  Intervention  Nutrition labels on food and beverage products served on college campuses  Comparator  Exposure to nutrition labels compared with no exposure  Outcome  Diet and food choices  Setting  College, university, and tertiary education campuses  Parameter  Inclusion criteria  Population  College students  Intervention  Nutrition labels on food and beverage products served on college campuses  Comparator  Exposure to nutrition labels compared with no exposure  Outcome  Diet and food choices  Setting  College, university, and tertiary education campuses  Search strategy PubMed, EBSCO, PsycInfo, and Web of Science databases were searched using combinations of the following keywords: (1) “nutrition,” “calorie,” “food,” “diet,” or “menu”; (2) “label,” “labeling,” or “labelling”; (3) “dietary quality,” “diet,” “dietary intake,” “food intake,” “caloric intake,” “calorie intake,” “nutritional quality,” “nutritional intake,” “food choice,” “meal choice,” “food selection,” “food consumption,” “meal selection,” “meal consumption,” or “eating”; and (4) “college student(s),” “university student(s),” “young adult(s),” “university,” “college,” or “tertiary education.” The following keywords were used to exclude articles to limit the number of harvests: “supplement,” “pharmacology,” “medication,” “allerg*,” “mice,” “rat,” “choline,” “anemia,” “anorexia,” and “cigarette.” For example, the specific search terms used in PubMed are listed in Table 2. Table 2 Search terms used in PubMed Predictor terms  (“nutrition label” OR “nutrition labels” OR “nutrition labeling” OR “nutrition labelling” OR “calorie labels” OR “calorie label” OR “calorie labelling” OR “calorie labeling” OR “food label” OR “food labels” OR “menu label” OR “menu labels” OR “menu labeling” OR “menu labelling” OR “label usage” OR “label use”)  Outcome terms  AND (“dietary quality” OR “diet” OR “dietary intake” OR “food intake” OR “caloric intake” OR “calorie intake” OR “nutritional quality” OR “nutritional intake” OR “food choice” OR “meal choice” OR “food selection” OR “food consumption” OR “meal selection” OR “meal consumption” OR “eating”)  Population terms  AND (“college student” OR “college students” OR “university student” OR “university students” OR “young adult” OR “young adults” OR “university” OR “college” OR “tertiary education”)  Exclusionary terms  NOT supplement NOT pharmacology NOT medication NOT allerg* NOT mice NOT rat NOT cigarette NOT choline NOT anemia NOT anorexia  Predictor terms  (“nutrition label” OR “nutrition labels” OR “nutrition labeling” OR “nutrition labelling” OR “calorie labels” OR “calorie label” OR “calorie labelling” OR “calorie labeling” OR “food label” OR “food labels” OR “menu label” OR “menu labels” OR “menu labeling” OR “menu labelling” OR “label usage” OR “label use”)  Outcome terms  AND (“dietary quality” OR “diet” OR “dietary intake” OR “food intake” OR “caloric intake” OR “calorie intake” OR “nutritional quality” OR “nutritional intake” OR “food choice” OR “meal choice” OR “food selection” OR “food consumption” OR “meal selection” OR “meal consumption” OR “eating”)  Population terms  AND (“college student” OR “college students” OR “university student” OR “university students” OR “young adult” OR “young adults” OR “university” OR “college” OR “tertiary education”)  Exclusionary terms  NOT supplement NOT pharmacology NOT medication NOT allerg* NOT mice NOT rat NOT cigarette NOT choline NOT anemia NOT anorexia  Titles and abstracts of the articles identified through the keyword search were screened against the study selection criteria. Potentially relevant articles were retrieved for evaluation of the full text. A reference list search (ie, backward reference search) and cited reference search (ie, forward reference search) were conducted on the basis of the full-text articles that met the study selection criteria and were identified from the keyword search. Articles identified from the backward and forward reference search were further screened and evaluated using the same study selection criteria. The reference search was repeated on all newly identified articles until no additional relevant articles were found. Data extraction and synthesis A standardized data extraction form was used to collect the following methodological and outcome variables from each included study: author(s), publication year, study design, setting, sample size and demographics, response and/or completion rate, participant recruitment criteria, measures of nutrition label use and diet quality, main findings, and conclusions. Meta-analysis A meta-analysis was performed on studies that reported the mean number of calories consumed in the presence and absence of labels, standard deviations, and sample size for each group. When the number of calories selected or consumed was reported as an outcome but did not include all necessary information, authors were contacted. Effect size was calculated on the basis of the mean difference in calories selected or consumed between groups exposed and not exposed to nutrition labels. Pre–post studies (without a control group) and randomized controlled trials were analyzed in separate meta-analyses owing to differences in strength of the study design.19 Among studies that compared different label types, the simple textual labels were used when testing the effects of nutrition labels vs no labels. In addition, a meta-analysis was performed to test the effect of contextual labels vs simple textual labels. Four studies were included in the controlled experiment meta-analysis29,34–36, 6 in the pre–post meta-analysis,31,37–41 and 3 in the meta-analysis comparing contextual labels with simple textual labels.29,36,41 Study heterogeneity was assessed using the I2 index. The level of heterogeneity represented by I2 was interpreted as modest (I2 ≤ 25%), moderate (25% < I2 ≤ 50%), substantial (50% < I2 ≤ 75%), or considerable (I2 > 75%). Random-effects models were used for estimation since considerable heterogeneity was present. Publication bias was not assessed because of the variability in study designs and outcomes of interest. Meta-analysis was performed using Stata/IC software, version 13.1 (StataCorp, College Station, TX, USA). All analyses used 2-sided tests, and P < 0.05 was considered statistically significant. Study quality assessment The National Institutes of Health’s Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies was adapted to assess the quality of each included study.42 This assessment tool (Table 3) rates each study on the basis of 9 criteria. For each criterion, a score of 1 was assigned if “yes” was the response, whereas a score of 0 was assigned otherwise (ie, an answer of “no,” “not applicable,” “not reported,” or “cannot determine”). A study-specific global score, ranging from 0 to 9, was calculated by summing scores across all criteria. Criteria were as follows: (1) research question, study design, and data collection procedures were clearly documented; (2) sample size was sufficiently large to provide confidence in the findings; (3) reasons for selecting or recruiting the number of individuals were included, or statistical power was discussed; (4) there was a control (either a control group, control cafeteria, or a pre–post study in which participants served as their own controls) in the study; (5) either study participants or cafeterias were randomized; (6) dietary outcomes were observed rather than self-reported; (7) actual dietary intake (not simply selection) was assessed; (8) dietary outcome was assessed in a naturalistic eating setting; and (9) key potential confounding variables (eg, sex, body mass index) were measured and adjusted statistically for their effect on the relationship between nutrition label use and dietary intake. RESULTS Of the 798 unduplicated articles identified through the keyword and reference search, 722 were excluded by title and abstract screening (Figure 1). The remaining 76 articles were reviewed in full text, by which 61 studies were excluded because of the following reasons: age ineligibility (n = 31); inappropriate setting, such as a hospital or workplace cafeteria (n = 8); lack of quantitative assessment of food choices (n = 10); lack of assessment of label exposure in relation to diet, independent of other factors or interventions (n = 3); use of labels that were not nutrition labels (n = 2); manipulation of nutrition labels or priming of participants (n = 5); and assessment of hypothetical rather than actual dietary intake (n = 2). The remaining 15 articles were included in the review. An additional 6 articles were identified through reference search, resulting in a total of 21 articles (22 separate studies) included in the review. Table 3 Study quality assessment criteria, adapted from the US National Heart, Lung, and Blood Institute Quality Assessment Tool for pre–post studies and cross-sectional studies Item  Criterion of study quality  Mean score  1  Were the research question, the study design, and the data collection procedures clearly documented? (yes = 1, no = 0)a  1  2  Was the sample size sufficiently large to provide confidence in the findings?b (yes = 1, no = 0)  0.77  3  Were reasons for selecting or recruiting the number of individuals included, or was statistical power discussed? (yes = 1, no = 0)  0.14  4  Was there a control (a control group, a control cafeteria, or a pre–post study cohort study in which participants served as their controls)? (yes = 1, no = 0)  0.55  5  Were study participants or cafeterias randomized? (yes = 1, no = 0)  0.27  6  Were dietary outcome(s) observed, not self-reported? (yes = 1, no = 0)  0.86  7  Was actual intake, not just food selection, assessed? (yes = 1, no = 0)  0.41  8  Was dietary outcome assessed in a natural eating setting? (yes = 1, no = 0)  0.82  9  Were key potential confounding variables (eg, sex, body mass index) measured and adjusted statistically for their effect on the relationship between nutrition label use and dietary intake? (yes = 1, no = 0)  0.41    Overall study quality score  5.2  Item  Criterion of study quality  Mean score  1  Were the research question, the study design, and the data collection procedures clearly documented? (yes = 1, no = 0)a  1  2  Was the sample size sufficiently large to provide confidence in the findings?b (yes = 1, no = 0)  0.77  3  Were reasons for selecting or recruiting the number of individuals included, or was statistical power discussed? (yes = 1, no = 0)  0.14  4  Was there a control (a control group, a control cafeteria, or a pre–post study cohort study in which participants served as their controls)? (yes = 1, no = 0)  0.55  5  Were study participants or cafeterias randomized? (yes = 1, no = 0)  0.27  6  Were dietary outcome(s) observed, not self-reported? (yes = 1, no = 0)  0.86  7  Was actual intake, not just food selection, assessed? (yes = 1, no = 0)  0.41  8  Was dietary outcome assessed in a natural eating setting? (yes = 1, no = 0)  0.82  9  Were key potential confounding variables (eg, sex, body mass index) measured and adjusted statistically for their effect on the relationship between nutrition label use and dietary intake? (yes = 1, no = 0)  0.41    Overall study quality score  5.2  a At a minimum, recruitment, mode and setting for data collection, and study duration were indicated. b If power calculations were not detailed, sufficient size was at least 100 participants per group (ie, pre- and postintervention) or at least 100 observations per period (ie, sales pre- and postintervention). Figure 1 View largeDownload slide Flow diagram of the literature search process. Figure 1 View largeDownload slide Flow diagram of the literature search process. Basic study characteristics The basic characteristics of the included studies are reported in Table 4.29–41,43–50 Studies were conducted in 5 countries: the United States (n = 15), the United Kingdom (n = 3), Canada (n = 2), Belgium (n = 1), and Australia (n = 1). Of the 22 studies, 5 were experimental or quasi-experimental controlled trials wherein participants were exposed or not exposed to nutrition labels, and 17 were cohort or pre–post calorie-labeling interventions conducted in cafeterias or with vending machines. Thirteen studies measured calories selected or consumed, and 12 assessed dietary quality via noncaloric measures of specific food items or meals (some studies assessed multiple outcomes). Table 4 Study design, methods and characteristics of sample, dietary outcome assessed, and main findings of the studies included in the systematic review Reference  Design, setting, and location of study  Methods and sample  Dietary outcome  Main findings  Aaron et al. (1995)34  Quasiexperimental (pre–post + control), 2 cafeterias, Reading, UK  Duration: 2 wk, M–F lunch Week 1: No labels Week 2: Labels containing calories and percentage of calories from fat posted in cafeteria 1 Participants (N = 90: 65 intervention participants and 25 control participants) selected meals each day at lunch, recorded foods and portions selected and eaten, and gave investigators plates after eating. Intervention participants were surveyed on awareness, usage, and label understanding 1 wk after the study  Self-reported + objectively measured plate waste: energy and macronutrient intake/selection for 10 lunch meals  Most (92%) intervention participants noticed labels, but few used them (8% greatly or moderately, 18% slightly). Intervention participants ate more calories (927 ± 27) in week 2 than in week 1 (875 ± 23, P < 0.05), while selection among control participants did not differ (944 ± 51 vs. 906 ± 54 in wk 2, P > 0.05). Among intervention participants, unrestrained eaters and males consumed more calories and carbohydrates (grams) and fat (grams), and less protein (grams) and percent energy from protein, whereas Females and unrestrained eaters had similar intakes across both weeks  Chu et al. (2009)37  Quasiexperimental (pre–post), single group, interrupted time series, cafeteria, Columbus, OH, USA  Duration: 6 wk, 7 d/wk, all day Weeks 1–2: pretreatment, no labels Weeks 3–4: treatment, 3” × 5” label cards at point of selection included calories, serving size, and macronutrients for study entrées Weeks 5–6: post-treatment, no labels Sales data for 12 study entrées were collected during each period; N = 13 951 pretreatment, N = 14 199 treatment, N = 14 020 post-treatment  Sales data: daily average energy content of 12 entrées  Average entrée energy content of entrées decreased 12.0 kcal from the last day of baseline to the first day of treatment (P = 0.007). Energy content decreased over the treatment period (−0.3 kcal/d), but not significantly from pretreatment. After labels were removed, energy content increased at a rate of 1.5 kcal/d (P = 0.013) across post-treatment period  Cioffi et al. (2015)43  Quasiexperimental (pre–post), 20 retail dining units, Ithaca, NY, USA  Duration: 3 y (6 semesters) Semesters 1–3: baseline Semesters 4–6: nutrition labels with calories and nutrient composition posted Weekly sales data (including thousands of observations, but specific number not noted) for 45 prepackaged food items were collected from campus dining units. Items were categorized as high-calorie and low-calorie, and high-fat and low-fat  Sales data: weekly mean total calories and fat purchased per FreshTake item selected  Average energy and fat content of items purchased per week decreased 6.5% and 7.4%, respectively, after label introduction (both P < 0.001). Upon label introduction, sales of low-calorie foods increased from 2.9% to 3.2%, and sales of high-fat items decreased from 2.9% to 2.6% (both P < 0.001)  Cinciripini (1984)45  Quasiexperimental (pre–post), cafeteria, Galveston, TX, USA  Duration: ≈ 24–27 wk (3 periods, 8–9 wk each), lunch 3–5 d/wk Period 1: baseline 1 Period 2: 2 large signs were posted at cafeteria entrance listing items and caloric information; flyers describing the signs were distributed for 10 d Period 3: baseline 2 Unobtrusive observation of participants' body classification (lean, normal, obese) and food choices from a checklist of 97 food items offered while participants paid for lunch  Observed choice: nonstarchy vegetable/soup/fruit/low-fat dairy, high-fat, red meat, lean protein, regular dairy, starchy carbohydrates, salads  Frequency of choosing carbohydrates decreased for all participants from baseline to intervention, frequency of choosing red meat decreased for almost all groups, frequency of choosing regular dairy products decreased for normal-weight males and females, and frequency of choosing high-fat dessert/sauces decreased for 1 subgroup (all P < 0.05). Salad and vegetable/soup/fruit/low-fat dairy choices each increased for 1 subgroup. After label removal, frequency of choosing salad increased for 1 subgroup; decreases in frequency of choosing high-fat dessert/sauces persisted for 1 subgroup  Davis-Chervin et al. (1985)46  Quasiexperimental (pre–post + control), 2 cafeterias, Stanford, CA, USA  Duration: 1 y (3 trimesters), weekday lunch + dinner Trimester 1: 5-wk baseline, 5-wk intervention (cafeteria 1) with labels (4″×6″ 3-color cards with calories, percentage of calories from fat, milligrams of cholesterol), and nutrition education posters displayed Trimester 2: 2-wk baseline, 5-wk intervention (cafeteria 1), 3-wk baseline Trimester 3: 5-wk intervention (both cafeterias; in cafeteria 2 only, labels but no posters were displayed), 5-wk baseline Sales data were collected to identify the proportion of low-calorie, low-cholesterol, and low-fat entrées sold. Cafeteria 1 served 175–200 first-year students at each meal; cafeteria 2 served 450–500 undergraduates from all 4 classes  Sales data: proportion of entrées chosen with the lowest amount of cholesterol, fat, or calories of the entrées served on a given day  In cafeteria 1, selection of low-calorie entrées at lunch increased by 35% and 66% during intervention periods 1 and 2, respectively, while selection of low-cholesterol entrées at lunch or dinner increased by 28%–53% from baseline (all P < 0.05). Selection of low-cholesterol, low-fat, and low-calorie entrées increased as a proportion of total from baseline levels during each intervention period and during the final no-intervention phase. In cafeteria 2, nonsignificant increases in the selection of low-calorie and low-cholesterol entrées were observed from baseline to intervention  Dingman et al. (2015)35  Quasiexperimental (pre–post + control), residence hall vending machines, Greensboro, NC, USA  Duration: 8 wk 18 vending machines, containing 35–40 snacks each, produced usable data. At the start of week 5, posters listing Nutrition Facts for each snack and highlighting the 5 healthiest products with a Better Choice logo were posted beside the 9 intervention machines, with a note on the machine referencing the poster. Intervention hall residents were sent an email about Better Choice criteria. Students were emailed a survey upon study completion  Sales data: average calories per snack, proportion of Better Choice items sold  56% of residents in intervention halls (n = 364) noticed on-site nutrition information, but n = 192 (60% of those who answered both questions) reported it did not influence purchasing decisions. Neither the proportion of Better Choice snacks sold (6.2% before and 6.9% after in intervention machines; 8.2% before and 6.6% after in control machines) nor the average number of calories per snack (252 ± 24 before and 251 ± 21 after in intervention machines; 217 ± 55 before and 225 ± 56 after in control machines) differed between the intervention and control machines or after label introduction (both P > 0.05)  Ellison et al. (2014)29  Quasiexperimental comparative trial, full-service restaurant, Stillwater, OK, USA  Duration: 12 wk (out of a 19-wk intervention), lunch daily Diners at each table were given 1 of 3 menus: (1) conventional with food descriptions; (2) descriptions + calorie count; (3) descriptions + calorie count + traffic light symbol (red for entrées over 800 kcal, yellow for entrées 401–800 kcal, and green for those under 400 kcal). Receipts yielded 978 observations for the labeling portion of the study  Sales data: number of calories in main entrée  Labels resulted in more orders of low- and medium-calorie items (P < 0.01). Without labels, 30% of participants chose low-calorie, 36% medium-calorie, and 35% high-calorie entrées. With calorie-only labels, 32% chose low-calorie, 38% medium-calorie, and 30% high-calorie entrées. With traffic light labels, 39% chose low-calorie, 33% medium-calorie, and 28% high-calorie entrées. Calorie-only and traffic light labels reduced orders of high-calorie items by 4.4% and 6.4%, respectively. Those exposed to calorie-only and traffic light labels ordered 30 (3.9%) and 71 (9.4%) fewer kilocalories per meal, respectively, compared with those not exposed to labels. Using regression, traffic light labels predicted calories ordered (P < 0.01), while calorie-only labels did not  Freedman (2011)47  Quasiexperimental (pre–post), all-you-can-eat cafeteria, San Jose, CA, USA  Duration: 5 wk, MWF lunch Week 1: baseline (no labels) Weeks 2–5: intervention (point-of-service nutrition labels, 10″ × 7.5″ laminated color signs on sneeze guards, including portion sizes, photos, and slogans) Researchers unobtrusively observed students’ (N = 1675) choices of French fries (small portions ≤ 18 fries, large > 18 fries) and salad dressings in 1 cafeteria. Students (N = 377) were surveyed about label awareness and usage 1 wk after intervention  Observed choice: choices of salad dressing and French fries, portion size of fries  Of those who reported seeing nutrition information, one-third said it affected their choice (32%, N = 73) or portion size (38%, N = 84) of French fries, and one-fourth said it affected their choice (24%, N = 53) or portion size (26%, N = 58) of salad dressing. French fry selection did not change from baseline (24% of diners) to intervention (25%), but percentage of diners choosing large portion sizes decreased from 60% at baseline to 43% when labels were present (P < 0.05). The proportion of students selecting salad dressing at baseline (19%) and intervention (25%) remained similar, but the proportion selecting salad dressings with mid-range calories increased (P < 0.05), in part due to nonsignificant decreases in the proportion selecting the highest-calorie dressings  Freedman & Connors (2010)50  Quasiexperimental (pre–post), convenience store, San Jose, CA, USA  Duration: 11 wk, separated by several months Weeks 1–6: baseline, mid-fall 2008 Winter break 2008–2009: Eat Smart labels (1.25″ × 3″) and a poster were placed in the convenience store, and labels were placed directly beneath healthier food items Weeks 7–11: follow-up, mid-spring 2009 Sales data were collected for cereal, soup, cracker, and bread categories  Sales data: tagged items  Sales and the percentage of sales of tagged food items did not change significantly upon label introduction. Sales of tagged items increased by 3.6% in the intervention (P = 0.082), and the percentage of sales of tagged items in the cereal, soup, and cracker categories increased, while sales of tagged bread decreased (all P > 0.05)  Hammond et al. (2015)38  Quasiexperimental (pre–post cohort), cafeteria, Waterloo, ON, Canada  Duration: 2 wk, lunch and dinner, baseline and intervention separated by 6 wk so that menu offerings were the same Week 1: baseline (no labels) Week 7: posted labels included calories in red, food description in black, 24-point font Week 8: patrons (N = 159) were approached (using an intercept method) upon exiting and asked to complete a 10-min interviewer-administered survey on nutrition label use and food and beverage selections and intake that day  Self-reported (directly after meal) data: calories ordered and consumed  Reported nutrition label use increased from 9% at baseline to 29% at follow-up (P < 0.001). Calories ordered decreased by 11% (91 kcal, P = 0.013) and calories consumed by 13% (98 kcal, P = 0.006) from baseline to follow-up after adjustment for sex, BMI, race, general health, and weight aspirations and perceptions  Hoefkens et al. (2011)39  Quasiexperimental (pre–post cohort), cafeteria, Ghent, Belgium  Duration: 8 mo, 6 d per person, 3 d each at baseline and follow-up Months 1–2 (October/November 2008): baseline Month 6: labels posted in March included a star rating system, with meals given a star if they met recommended amounts for energy, saturated fat, sodium, and vegetables, for a total of 0–4 points and 0–3 stars. The 12 healthiest combinations, with ratings, were listed each day, and featured on posters and in the buffet line Months 7–8 (April/May 2009): follow-up A convenience sample (N = 224) completed 3-d food records and questionnaires at baseline and follow-up  Self-reported (concerning specific meals) data: energy intake from cafeteria meals (average of the 3 d from food records), food types, and macronutrients  Average calorie intake for the lunch cafeteria meal and over 24 h did not change between baseline and follow-up (P > 0.05). Participants consumed significantly more grams of vegetables at follow-up (both at the canteen meal and over 24 h, driven by the canteen meal) and fewer grams of carbohydrates over 24 h (P < 0.05). Protein, fat, percent energy from saturated fat, and sodium were similar across baseline and follow-up, as was the proportion of meals chosen in different star rating groups  Hoerr & Louden (1993)30  Quasiexperimental (pre–post), vending machines, East Lansing, MI, USA  Duration: 2 y, weeks 4–7 of each trimester Year 1: baseline Year 2: posted labels for each item included calories, protein, vitamin A, vitamin C, thiamin, riboflavin, niacin, calcium, and iron (in orange bar graphs) Sales of low-nutrient-density (chocolate candy bar, nuts, chocolate cookie), moderate-nutrient density (chocolate peanuts, granola bar, cheese popcorn), and high-nutrient-density (pretzels, peanut butter and crackers, peanuts) items were measured in 4 vending machines with 8 slots each in 4 academic buildings; N = 7174 in year 1 and N = 7742 in year 2  Sales data: proportion of low-, moderate-, and high-nutrient-density foods sold  Proportion of snacks sold in the low-, moderate-, and high-nutrient-density groups did not differ significantly between years 1 and 2  James et al. (2015)36  Quasiexperimental comparative trial, metabolic kitchen and graduate residence hall, Fort Worth, TX, USA  Duration: N/A Participants (N = 300) came to a laboratory kitchen (N = 278) or graduate residence hall (N = 22), had height and weight measured, and were seated alone, receiving 1 of 3 menus for a fast-food restaurant. Menus included the following: (1) no labels, (2) kilocalorie labels with a statement about daily caloric requirements, and (3) exercise labels showing minutes of brisk walking required to burn the energy from food items (specific to males and females). After ordering, participants were surveyed, and participants in labels groups were asked if they had noticed the labels. Food and beverages were unobtrusively weighed before and after the meal  Observed choice + objectively measured plate waste: calories ordered and eaten and calories from fat, protein, and carbohydrates  91% of those exposed noticed labels. The no-labels group ordered more calories than the exercise-labels group but not the kilocalorie-labels group (overall and exercise- vs no-labels groups both P<0.05). For consumption, the no-labels group again ate significantly more than the exercise-labels group but not the kcal labels (overall and exercise- vs no-labels groups both P<0.05). There were no differences for ordering and consumption between the exercise- and kcal-label groups and the kcal- and no-label groups (P>0.05). Exercise- and no-label groups also ordered and consumed differing percentages of calories from fat (both P<0.05), but not carbohydrates and protein  Larson-Brown (1978)49  Quasiexperimental (pre–post), vending machines, Provo, UT, USA  Duration: 2 mo Month 1: baseline Month 2: nutrition labels posted in front of items included calories and the percentage of US dietary recommended allowance for protein, calcium, thiamin, vitamin C, and iron (in colored bar graphs) Sales data were collected for vending machines in 2 adjoining campus buildings. Foods were categorized as more nutritious (milk, sandwiches, fruit, Welchade [Welch’s; Concord, MA, USA], yogurt, V-8 juice [Campbell Soup Co; Camden, NJ, USA], ice cream) or less nutritious (soft drinks, sweet rolls and brownies, gum and LifeSavers [Squibb Beech-Nut; New York, NY, USA], Hostess products [Continental Baking Co; New York, NY, USA], M&M’s [Mars; McLean, VA, USA] , Hershey’s chocolate [Hershey Foods Corp; Hershey, PA, USA] , candy, cookies). N = 26 558 sales in February and N = 30 371 in March  Sales data: more-nutritious and less-nutritious foods  Purchase of more-nutritious foods increased from 49.8% of total sales in February to 53.7% of total sales in March, a significant difference. For more-nutritious foods, sales of milk, sandwiches, fruits, Welchade, and yogurt increased, while sales of V-8 juice and ice cream decreased. For less-nutritious foods, sales of soft drinks increased, while sales of all others decreased (significance not noted)  Lillico et al. (2015)40  Quasiexperimental (pre–post), student residence cafeteria, Waterloo, ON, Canada  Duration: 2 wk, lunch and dinner, separated by 6 wk so that menu offerings were the same Week 1: baseline (no labels) Week 6: posted labels included calories and food description in 24-point font Week 7: students (N = 131 baseline, N = 168 follow-up) were approached (using an intercept method) upon exiting and asked to complete a 10-min interviewer-administered survey assessing food and beverage intake  Self-reported (directly after meal) data: calories consumed  Calorie consumption did not change significantly between baseline (661 ± 309 kcal) and follow-up (601 ± 282 kcal, P = 0.104) periods  Nikolaou et al. (2014)41  Quasiexperimental (pre–post, interrupted time series), residence hall cafeteria, Glasgow, Scotland, UK  Duration: three 14-d study periods, each separated by 4 wk, evening Weeks 1–2: no labels Weeks 3–4: calorie-only labels Weeks 5–6: calories + suggested daily intake labels The first 100 meal selections for 14-d periods within the 5-wk menu cycle were recorded, for a total of 4200 meals, including side dishes. Ingredient orders for evening meals placed by caterers were also recorded over the course of 2 y (2 mo each year)  Observed choice: calories, fat, saturated fat, vitamin C, iron, and calcium content of meal choices  Both males and females selected fewer calories when labels were present and even fewer when calories + suggested daily intake labels were present; selection during each period differed significantly from that during the other periods (P < 0.01). From period 1 (simple labels) to period 3 (contextual labels), mean calories per tray fell by 25% for females and 15% for males. Fat and saturated fat content of meals decreased after exposure to calorie labels + suggested daily intake; no differences were found in selection of vitamin C, iron, or calcium. Total calories ordered by caterers fell 18%, orders for ingredients used primarily for dessert preparation fell 60%, and oils used for frying fell 35% from years 1 to 2 when labels were present  Nikolaou et al. (2014)48  Quasiexperimental (pre–post + control), 3 food retail outlets, Glasgow, Scotland, UK  Duration: 2 mo Month 1: baseline Month 2: labels posted for the last 2 wk (first 2 wk were a university holiday, with reduced catering) Calorie labels (laminated 5.4 cm × 9.9 cm) containing the item name, calories, and the “Human Nutrition” department logo and university coat of arms were posted prominently in front of all sandwiches in 2 intervention food outlets; a control outlet did not have labels posted. Patrons were surveyed online and in outlets 1 wk into the intervention  Sales data of 19 sandwiches/rolls with a variety of fillings and caloric content  61% of female and 41% of male students reported that calorie information influenced choices. Between months 1 and 2, sales of all labeled items fell 17% in the intervention and 2% in the control outlets (P < 0.001). Sales of high-calorie (−30%) and low-calorie (−18%) items and high-fat (−21%) and low-fat (−23%) items decreased from months 1 to 2 in intervention outlets (P < 0.001), while sales of these items did not differ at the control outlet  Roy et al. (2016)32  Quasiexperimental (pre–post), quick-service food outlet, Sydney, Australia  Duration: 10 wk (5 wk each, 1 y apart) Sales data were collected at baseline for 5 wk. The next year, kilojoule content was posted on menus, including a reference statement listing an average adult serving of 8700 KJ. Students selected foods and beverages from a menu and then ordered food at a counter. Students (N = 318) were also surveyed during the intervention  Sales data of 9 items of varying caloric content  Only 5% of those surveyed reported being both aware of and influenced by labels. Compared with baseline, sales of a high-calorie entrée (grilled burger) decreased 35% while sales of 1 lower-calorie meal (chicken schnitzel and chips) increased 34% upon exposure to kilojoule labeling; sales of the remaining 7 items did not change  Schwartz et al. (2012)31  Quasiexperimental (pre–post), Chinese restaurant, Durham, NC, USA  Duration: two 3-wk periods (two 2-wk periods included in this analysis), M–Th, lunch Patrons ordered 1 of 4 side dishes (rice, fried rice, lo mein, or steamed vegetables) and then 1 of 16–20 stir fry entrées. After a 3-wk baseline and downsizing intervention with no labels present, there was a 2-wk break wherein calorie labels were posted on the sneeze shield above containers. A second 3-wk intervention was then conducted with labels and a downsizing intervention (not included in this review). Itemized receipts were collected to measure sales  Sales data: calories ordered  Customers ordered an average of 1020 ± 15 kcal when not exposed to labels and 1033 ± 16 kcal after label exposure, a nonsignificant difference (P >0 .05)  Temple et al. (2010)44  Randomized between-group experiment, laboratory, Buffalo, NY, USA  Duration: N/A Participants (N = 47) visited a lab at lunchtime and were randomly assigned to watch a movie on either the organic food movement or how to read nutrition labels. They then ate a buffet lunch of preweighed items either including or not including nutrition labels  Observed choice + plate waste: calories consumed, energy-dense foods consumed  Those exposed to labels ate fewer calories than those not exposed (P = 0.04). Energy density of chosen foods also differed; those not exposed to labels ate more of both high- and low-energy-density foods (both P < 0.05)  Temple et al. (2011)33 (A)  Within-subject experiment, laboratory, Buffalo, NY, USA  Duration: N/A Participants (N = 51) were surveyed and then ate a buffet lunch of preweighed items in the laboratory 3 times (≈ 1 h each time): once without labels present, once with standard labels present (4″ × 6″ labels resembling the manufacturer’s label), and once with traffic light labels present (in random order). Participants were given 25 min alone to eat  Observed choice + plate waste: calories consumed, proportion of green/yellow/red foods consumed  Label condition did not affect calories consumed, but there was a significant interaction between gender, labeling condition, and weight group for calories consumed. Lean females consumed fewer calories when standard or traffic labels were present (P < 0.05); all other groups consumed approximately the same number of calories in all 3 conditions (P > 0.05). All groups consumed more green foods in the presence of traffic light labels (P = 0.002)  Temple et al. (2011)33 (B)  Within-subject experiment, laboratory, Buffalo, NY, USA  Duration: N/A Participants (N = 41) were surveyed and then ate a buffet lunch of preweighed items in the laboratory 2 times (≈ 1 h each time): once without labels present, and once with traffic light labels present (in random order)  Observed choice + plate waste: calories consumed, green/yellow/red foods consumed  Participants exposed to traffic light labels were more likely to purchase green items (P = 0.01), but labeling did not affect energy intake. Intake of green foods increased and red foods decreased upon exposure to traffic light labels vs no labels (both P < 0.05)  Reference  Design, setting, and location of study  Methods and sample  Dietary outcome  Main findings  Aaron et al. (1995)34  Quasiexperimental (pre–post + control), 2 cafeterias, Reading, UK  Duration: 2 wk, M–F lunch Week 1: No labels Week 2: Labels containing calories and percentage of calories from fat posted in cafeteria 1 Participants (N = 90: 65 intervention participants and 25 control participants) selected meals each day at lunch, recorded foods and portions selected and eaten, and gave investigators plates after eating. Intervention participants were surveyed on awareness, usage, and label understanding 1 wk after the study  Self-reported + objectively measured plate waste: energy and macronutrient intake/selection for 10 lunch meals  Most (92%) intervention participants noticed labels, but few used them (8% greatly or moderately, 18% slightly). Intervention participants ate more calories (927 ± 27) in week 2 than in week 1 (875 ± 23, P < 0.05), while selection among control participants did not differ (944 ± 51 vs. 906 ± 54 in wk 2, P > 0.05). Among intervention participants, unrestrained eaters and males consumed more calories and carbohydrates (grams) and fat (grams), and less protein (grams) and percent energy from protein, whereas Females and unrestrained eaters had similar intakes across both weeks  Chu et al. (2009)37  Quasiexperimental (pre–post), single group, interrupted time series, cafeteria, Columbus, OH, USA  Duration: 6 wk, 7 d/wk, all day Weeks 1–2: pretreatment, no labels Weeks 3–4: treatment, 3” × 5” label cards at point of selection included calories, serving size, and macronutrients for study entrées Weeks 5–6: post-treatment, no labels Sales data for 12 study entrées were collected during each period; N = 13 951 pretreatment, N = 14 199 treatment, N = 14 020 post-treatment  Sales data: daily average energy content of 12 entrées  Average entrée energy content of entrées decreased 12.0 kcal from the last day of baseline to the first day of treatment (P = 0.007). Energy content decreased over the treatment period (−0.3 kcal/d), but not significantly from pretreatment. After labels were removed, energy content increased at a rate of 1.5 kcal/d (P = 0.013) across post-treatment period  Cioffi et al. (2015)43  Quasiexperimental (pre–post), 20 retail dining units, Ithaca, NY, USA  Duration: 3 y (6 semesters) Semesters 1–3: baseline Semesters 4–6: nutrition labels with calories and nutrient composition posted Weekly sales data (including thousands of observations, but specific number not noted) for 45 prepackaged food items were collected from campus dining units. Items were categorized as high-calorie and low-calorie, and high-fat and low-fat  Sales data: weekly mean total calories and fat purchased per FreshTake item selected  Average energy and fat content of items purchased per week decreased 6.5% and 7.4%, respectively, after label introduction (both P < 0.001). Upon label introduction, sales of low-calorie foods increased from 2.9% to 3.2%, and sales of high-fat items decreased from 2.9% to 2.6% (both P < 0.001)  Cinciripini (1984)45  Quasiexperimental (pre–post), cafeteria, Galveston, TX, USA  Duration: ≈ 24–27 wk (3 periods, 8–9 wk each), lunch 3–5 d/wk Period 1: baseline 1 Period 2: 2 large signs were posted at cafeteria entrance listing items and caloric information; flyers describing the signs were distributed for 10 d Period 3: baseline 2 Unobtrusive observation of participants' body classification (lean, normal, obese) and food choices from a checklist of 97 food items offered while participants paid for lunch  Observed choice: nonstarchy vegetable/soup/fruit/low-fat dairy, high-fat, red meat, lean protein, regular dairy, starchy carbohydrates, salads  Frequency of choosing carbohydrates decreased for all participants from baseline to intervention, frequency of choosing red meat decreased for almost all groups, frequency of choosing regular dairy products decreased for normal-weight males and females, and frequency of choosing high-fat dessert/sauces decreased for 1 subgroup (all P < 0.05). Salad and vegetable/soup/fruit/low-fat dairy choices each increased for 1 subgroup. After label removal, frequency of choosing salad increased for 1 subgroup; decreases in frequency of choosing high-fat dessert/sauces persisted for 1 subgroup  Davis-Chervin et al. (1985)46  Quasiexperimental (pre–post + control), 2 cafeterias, Stanford, CA, USA  Duration: 1 y (3 trimesters), weekday lunch + dinner Trimester 1: 5-wk baseline, 5-wk intervention (cafeteria 1) with labels (4″×6″ 3-color cards with calories, percentage of calories from fat, milligrams of cholesterol), and nutrition education posters displayed Trimester 2: 2-wk baseline, 5-wk intervention (cafeteria 1), 3-wk baseline Trimester 3: 5-wk intervention (both cafeterias; in cafeteria 2 only, labels but no posters were displayed), 5-wk baseline Sales data were collected to identify the proportion of low-calorie, low-cholesterol, and low-fat entrées sold. Cafeteria 1 served 175–200 first-year students at each meal; cafeteria 2 served 450–500 undergraduates from all 4 classes  Sales data: proportion of entrées chosen with the lowest amount of cholesterol, fat, or calories of the entrées served on a given day  In cafeteria 1, selection of low-calorie entrées at lunch increased by 35% and 66% during intervention periods 1 and 2, respectively, while selection of low-cholesterol entrées at lunch or dinner increased by 28%–53% from baseline (all P < 0.05). Selection of low-cholesterol, low-fat, and low-calorie entrées increased as a proportion of total from baseline levels during each intervention period and during the final no-intervention phase. In cafeteria 2, nonsignificant increases in the selection of low-calorie and low-cholesterol entrées were observed from baseline to intervention  Dingman et al. (2015)35  Quasiexperimental (pre–post + control), residence hall vending machines, Greensboro, NC, USA  Duration: 8 wk 18 vending machines, containing 35–40 snacks each, produced usable data. At the start of week 5, posters listing Nutrition Facts for each snack and highlighting the 5 healthiest products with a Better Choice logo were posted beside the 9 intervention machines, with a note on the machine referencing the poster. Intervention hall residents were sent an email about Better Choice criteria. Students were emailed a survey upon study completion  Sales data: average calories per snack, proportion of Better Choice items sold  56% of residents in intervention halls (n = 364) noticed on-site nutrition information, but n = 192 (60% of those who answered both questions) reported it did not influence purchasing decisions. Neither the proportion of Better Choice snacks sold (6.2% before and 6.9% after in intervention machines; 8.2% before and 6.6% after in control machines) nor the average number of calories per snack (252 ± 24 before and 251 ± 21 after in intervention machines; 217 ± 55 before and 225 ± 56 after in control machines) differed between the intervention and control machines or after label introduction (both P > 0.05)  Ellison et al. (2014)29  Quasiexperimental comparative trial, full-service restaurant, Stillwater, OK, USA  Duration: 12 wk (out of a 19-wk intervention), lunch daily Diners at each table were given 1 of 3 menus: (1) conventional with food descriptions; (2) descriptions + calorie count; (3) descriptions + calorie count + traffic light symbol (red for entrées over 800 kcal, yellow for entrées 401–800 kcal, and green for those under 400 kcal). Receipts yielded 978 observations for the labeling portion of the study  Sales data: number of calories in main entrée  Labels resulted in more orders of low- and medium-calorie items (P < 0.01). Without labels, 30% of participants chose low-calorie, 36% medium-calorie, and 35% high-calorie entrées. With calorie-only labels, 32% chose low-calorie, 38% medium-calorie, and 30% high-calorie entrées. With traffic light labels, 39% chose low-calorie, 33% medium-calorie, and 28% high-calorie entrées. Calorie-only and traffic light labels reduced orders of high-calorie items by 4.4% and 6.4%, respectively. Those exposed to calorie-only and traffic light labels ordered 30 (3.9%) and 71 (9.4%) fewer kilocalories per meal, respectively, compared with those not exposed to labels. Using regression, traffic light labels predicted calories ordered (P < 0.01), while calorie-only labels did not  Freedman (2011)47  Quasiexperimental (pre–post), all-you-can-eat cafeteria, San Jose, CA, USA  Duration: 5 wk, MWF lunch Week 1: baseline (no labels) Weeks 2–5: intervention (point-of-service nutrition labels, 10″ × 7.5″ laminated color signs on sneeze guards, including portion sizes, photos, and slogans) Researchers unobtrusively observed students’ (N = 1675) choices of French fries (small portions ≤ 18 fries, large > 18 fries) and salad dressings in 1 cafeteria. Students (N = 377) were surveyed about label awareness and usage 1 wk after intervention  Observed choice: choices of salad dressing and French fries, portion size of fries  Of those who reported seeing nutrition information, one-third said it affected their choice (32%, N = 73) or portion size (38%, N = 84) of French fries, and one-fourth said it affected their choice (24%, N = 53) or portion size (26%, N = 58) of salad dressing. French fry selection did not change from baseline (24% of diners) to intervention (25%), but percentage of diners choosing large portion sizes decreased from 60% at baseline to 43% when labels were present (P < 0.05). The proportion of students selecting salad dressing at baseline (19%) and intervention (25%) remained similar, but the proportion selecting salad dressings with mid-range calories increased (P < 0.05), in part due to nonsignificant decreases in the proportion selecting the highest-calorie dressings  Freedman & Connors (2010)50  Quasiexperimental (pre–post), convenience store, San Jose, CA, USA  Duration: 11 wk, separated by several months Weeks 1–6: baseline, mid-fall 2008 Winter break 2008–2009: Eat Smart labels (1.25″ × 3″) and a poster were placed in the convenience store, and labels were placed directly beneath healthier food items Weeks 7–11: follow-up, mid-spring 2009 Sales data were collected for cereal, soup, cracker, and bread categories  Sales data: tagged items  Sales and the percentage of sales of tagged food items did not change significantly upon label introduction. Sales of tagged items increased by 3.6% in the intervention (P = 0.082), and the percentage of sales of tagged items in the cereal, soup, and cracker categories increased, while sales of tagged bread decreased (all P > 0.05)  Hammond et al. (2015)38  Quasiexperimental (pre–post cohort), cafeteria, Waterloo, ON, Canada  Duration: 2 wk, lunch and dinner, baseline and intervention separated by 6 wk so that menu offerings were the same Week 1: baseline (no labels) Week 7: posted labels included calories in red, food description in black, 24-point font Week 8: patrons (N = 159) were approached (using an intercept method) upon exiting and asked to complete a 10-min interviewer-administered survey on nutrition label use and food and beverage selections and intake that day  Self-reported (directly after meal) data: calories ordered and consumed  Reported nutrition label use increased from 9% at baseline to 29% at follow-up (P < 0.001). Calories ordered decreased by 11% (91 kcal, P = 0.013) and calories consumed by 13% (98 kcal, P = 0.006) from baseline to follow-up after adjustment for sex, BMI, race, general health, and weight aspirations and perceptions  Hoefkens et al. (2011)39  Quasiexperimental (pre–post cohort), cafeteria, Ghent, Belgium  Duration: 8 mo, 6 d per person, 3 d each at baseline and follow-up Months 1–2 (October/November 2008): baseline Month 6: labels posted in March included a star rating system, with meals given a star if they met recommended amounts for energy, saturated fat, sodium, and vegetables, for a total of 0–4 points and 0–3 stars. The 12 healthiest combinations, with ratings, were listed each day, and featured on posters and in the buffet line Months 7–8 (April/May 2009): follow-up A convenience sample (N = 224) completed 3-d food records and questionnaires at baseline and follow-up  Self-reported (concerning specific meals) data: energy intake from cafeteria meals (average of the 3 d from food records), food types, and macronutrients  Average calorie intake for the lunch cafeteria meal and over 24 h did not change between baseline and follow-up (P > 0.05). Participants consumed significantly more grams of vegetables at follow-up (both at the canteen meal and over 24 h, driven by the canteen meal) and fewer grams of carbohydrates over 24 h (P < 0.05). Protein, fat, percent energy from saturated fat, and sodium were similar across baseline and follow-up, as was the proportion of meals chosen in different star rating groups  Hoerr & Louden (1993)30  Quasiexperimental (pre–post), vending machines, East Lansing, MI, USA  Duration: 2 y, weeks 4–7 of each trimester Year 1: baseline Year 2: posted labels for each item included calories, protein, vitamin A, vitamin C, thiamin, riboflavin, niacin, calcium, and iron (in orange bar graphs) Sales of low-nutrient-density (chocolate candy bar, nuts, chocolate cookie), moderate-nutrient density (chocolate peanuts, granola bar, cheese popcorn), and high-nutrient-density (pretzels, peanut butter and crackers, peanuts) items were measured in 4 vending machines with 8 slots each in 4 academic buildings; N = 7174 in year 1 and N = 7742 in year 2  Sales data: proportion of low-, moderate-, and high-nutrient-density foods sold  Proportion of snacks sold in the low-, moderate-, and high-nutrient-density groups did not differ significantly between years 1 and 2  James et al. (2015)36  Quasiexperimental comparative trial, metabolic kitchen and graduate residence hall, Fort Worth, TX, USA  Duration: N/A Participants (N = 300) came to a laboratory kitchen (N = 278) or graduate residence hall (N = 22), had height and weight measured, and were seated alone, receiving 1 of 3 menus for a fast-food restaurant. Menus included the following: (1) no labels, (2) kilocalorie labels with a statement about daily caloric requirements, and (3) exercise labels showing minutes of brisk walking required to burn the energy from food items (specific to males and females). After ordering, participants were surveyed, and participants in labels groups were asked if they had noticed the labels. Food and beverages were unobtrusively weighed before and after the meal  Observed choice + objectively measured plate waste: calories ordered and eaten and calories from fat, protein, and carbohydrates  91% of those exposed noticed labels. The no-labels group ordered more calories than the exercise-labels group but not the kilocalorie-labels group (overall and exercise- vs no-labels groups both P<0.05). For consumption, the no-labels group again ate significantly more than the exercise-labels group but not the kcal labels (overall and exercise- vs no-labels groups both P<0.05). There were no differences for ordering and consumption between the exercise- and kcal-label groups and the kcal- and no-label groups (P>0.05). Exercise- and no-label groups also ordered and consumed differing percentages of calories from fat (both P<0.05), but not carbohydrates and protein  Larson-Brown (1978)49  Quasiexperimental (pre–post), vending machines, Provo, UT, USA  Duration: 2 mo Month 1: baseline Month 2: nutrition labels posted in front of items included calories and the percentage of US dietary recommended allowance for protein, calcium, thiamin, vitamin C, and iron (in colored bar graphs) Sales data were collected for vending machines in 2 adjoining campus buildings. Foods were categorized as more nutritious (milk, sandwiches, fruit, Welchade [Welch’s; Concord, MA, USA], yogurt, V-8 juice [Campbell Soup Co; Camden, NJ, USA], ice cream) or less nutritious (soft drinks, sweet rolls and brownies, gum and LifeSavers [Squibb Beech-Nut; New York, NY, USA], Hostess products [Continental Baking Co; New York, NY, USA], M&M’s [Mars; McLean, VA, USA] , Hershey’s chocolate [Hershey Foods Corp; Hershey, PA, USA] , candy, cookies). N = 26 558 sales in February and N = 30 371 in March  Sales data: more-nutritious and less-nutritious foods  Purchase of more-nutritious foods increased from 49.8% of total sales in February to 53.7% of total sales in March, a significant difference. For more-nutritious foods, sales of milk, sandwiches, fruits, Welchade, and yogurt increased, while sales of V-8 juice and ice cream decreased. For less-nutritious foods, sales of soft drinks increased, while sales of all others decreased (significance not noted)  Lillico et al. (2015)40  Quasiexperimental (pre–post), student residence cafeteria, Waterloo, ON, Canada  Duration: 2 wk, lunch and dinner, separated by 6 wk so that menu offerings were the same Week 1: baseline (no labels) Week 6: posted labels included calories and food description in 24-point font Week 7: students (N = 131 baseline, N = 168 follow-up) were approached (using an intercept method) upon exiting and asked to complete a 10-min interviewer-administered survey assessing food and beverage intake  Self-reported (directly after meal) data: calories consumed  Calorie consumption did not change significantly between baseline (661 ± 309 kcal) and follow-up (601 ± 282 kcal, P = 0.104) periods  Nikolaou et al. (2014)41  Quasiexperimental (pre–post, interrupted time series), residence hall cafeteria, Glasgow, Scotland, UK  Duration: three 14-d study periods, each separated by 4 wk, evening Weeks 1–2: no labels Weeks 3–4: calorie-only labels Weeks 5–6: calories + suggested daily intake labels The first 100 meal selections for 14-d periods within the 5-wk menu cycle were recorded, for a total of 4200 meals, including side dishes. Ingredient orders for evening meals placed by caterers were also recorded over the course of 2 y (2 mo each year)  Observed choice: calories, fat, saturated fat, vitamin C, iron, and calcium content of meal choices  Both males and females selected fewer calories when labels were present and even fewer when calories + suggested daily intake labels were present; selection during each period differed significantly from that during the other periods (P < 0.01). From period 1 (simple labels) to period 3 (contextual labels), mean calories per tray fell by 25% for females and 15% for males. Fat and saturated fat content of meals decreased after exposure to calorie labels + suggested daily intake; no differences were found in selection of vitamin C, iron, or calcium. Total calories ordered by caterers fell 18%, orders for ingredients used primarily for dessert preparation fell 60%, and oils used for frying fell 35% from years 1 to 2 when labels were present  Nikolaou et al. (2014)48  Quasiexperimental (pre–post + control), 3 food retail outlets, Glasgow, Scotland, UK  Duration: 2 mo Month 1: baseline Month 2: labels posted for the last 2 wk (first 2 wk were a university holiday, with reduced catering) Calorie labels (laminated 5.4 cm × 9.9 cm) containing the item name, calories, and the “Human Nutrition” department logo and university coat of arms were posted prominently in front of all sandwiches in 2 intervention food outlets; a control outlet did not have labels posted. Patrons were surveyed online and in outlets 1 wk into the intervention  Sales data of 19 sandwiches/rolls with a variety of fillings and caloric content  61% of female and 41% of male students reported that calorie information influenced choices. Between months 1 and 2, sales of all labeled items fell 17% in the intervention and 2% in the control outlets (P < 0.001). Sales of high-calorie (−30%) and low-calorie (−18%) items and high-fat (−21%) and low-fat (−23%) items decreased from months 1 to 2 in intervention outlets (P < 0.001), while sales of these items did not differ at the control outlet  Roy et al. (2016)32  Quasiexperimental (pre–post), quick-service food outlet, Sydney, Australia  Duration: 10 wk (5 wk each, 1 y apart) Sales data were collected at baseline for 5 wk. The next year, kilojoule content was posted on menus, including a reference statement listing an average adult serving of 8700 KJ. Students selected foods and beverages from a menu and then ordered food at a counter. Students (N = 318) were also surveyed during the intervention  Sales data of 9 items of varying caloric content  Only 5% of those surveyed reported being both aware of and influenced by labels. Compared with baseline, sales of a high-calorie entrée (grilled burger) decreased 35% while sales of 1 lower-calorie meal (chicken schnitzel and chips) increased 34% upon exposure to kilojoule labeling; sales of the remaining 7 items did not change  Schwartz et al. (2012)31  Quasiexperimental (pre–post), Chinese restaurant, Durham, NC, USA  Duration: two 3-wk periods (two 2-wk periods included in this analysis), M–Th, lunch Patrons ordered 1 of 4 side dishes (rice, fried rice, lo mein, or steamed vegetables) and then 1 of 16–20 stir fry entrées. After a 3-wk baseline and downsizing intervention with no labels present, there was a 2-wk break wherein calorie labels were posted on the sneeze shield above containers. A second 3-wk intervention was then conducted with labels and a downsizing intervention (not included in this review). Itemized receipts were collected to measure sales  Sales data: calories ordered  Customers ordered an average of 1020 ± 15 kcal when not exposed to labels and 1033 ± 16 kcal after label exposure, a nonsignificant difference (P >0 .05)  Temple et al. (2010)44  Randomized between-group experiment, laboratory, Buffalo, NY, USA  Duration: N/A Participants (N = 47) visited a lab at lunchtime and were randomly assigned to watch a movie on either the organic food movement or how to read nutrition labels. They then ate a buffet lunch of preweighed items either including or not including nutrition labels  Observed choice + plate waste: calories consumed, energy-dense foods consumed  Those exposed to labels ate fewer calories than those not exposed (P = 0.04). Energy density of chosen foods also differed; those not exposed to labels ate more of both high- and low-energy-density foods (both P < 0.05)  Temple et al. (2011)33 (A)  Within-subject experiment, laboratory, Buffalo, NY, USA  Duration: N/A Participants (N = 51) were surveyed and then ate a buffet lunch of preweighed items in the laboratory 3 times (≈ 1 h each time): once without labels present, once with standard labels present (4″ × 6″ labels resembling the manufacturer’s label), and once with traffic light labels present (in random order). Participants were given 25 min alone to eat  Observed choice + plate waste: calories consumed, proportion of green/yellow/red foods consumed  Label condition did not affect calories consumed, but there was a significant interaction between gender, labeling condition, and weight group for calories consumed. Lean females consumed fewer calories when standard or traffic labels were present (P < 0.05); all other groups consumed approximately the same number of calories in all 3 conditions (P > 0.05). All groups consumed more green foods in the presence of traffic light labels (P = 0.002)  Temple et al. (2011)33 (B)  Within-subject experiment, laboratory, Buffalo, NY, USA  Duration: N/A Participants (N = 41) were surveyed and then ate a buffet lunch of preweighed items in the laboratory 2 times (≈ 1 h each time): once without labels present, and once with traffic light labels present (in random order)  Observed choice + plate waste: calories consumed, green/yellow/red foods consumed  Participants exposed to traffic light labels were more likely to purchase green items (P = 0.01), but labeling did not affect energy intake. Intake of green foods increased and red foods decreased upon exposure to traffic light labels vs no labels (both P < 0.05)  Abbreviations: BMI, body mass index; M−F, Monday through Friday; MWF, Monday, Wednesday, Friday; N/A, not available. College students included in the study samples were aged 19 to 29.9 years, and 34% to 75% were female (except for 1 study that exclusively recruited females). The majority of the samples were normal weight, though 1 study reported an average body mass index considered overweight (25.9), and another purposefully recruited participants with a weight status classified as obese for about half of the sample. Dietary outcomes: calories selected or consumed Dietary outcomes and main findings are reported in Table 4. Eight of the 13 studies investigating caloric selection or intake as an outcome found positive effects of posting labels; of the 5 remaining studies, 1 found a negative effect of introducing labels and 4 found no effect on energy selection and/or intake. Of the studies showing a statistically significant positive effect, Chu et al.37 in a cafeteria and Cioffi et al43 in retail dining units found that the average calorie content of items sold decreased after nutrition labels were introduced. In cafeterias, Hammond et al.38 reported that students ordered and consumed fewer calories after labels were posted, while Nikolaou et al.41 found that calories selected decreased upon posting of both simple nutrition labels and labels that included suggested daily intake. In an on-campus restaurant, Ellison et al.29 reported that exposure to traffic light or numeric menu labels resulted in decreased calories ordered, with traffic light labels being especially effective. In laboratory settings, James et al.36 found that exposure to exercise labels specifying how much physical activity was needed to burn the calories in a food item effectively decreased calories ordered and consumed, and Temple et al.33,44 found in 2 separate studies that label exposure decreased calories consumed in a buffet lunch among 1 subgroup and overall. In contrast, Aaron et al.34 found a negative effect of labeling: students eating in an intervention cafeteria consumed more calories after label introduction and more calories than the control group. Of the studies showing no effect, Dingman et al.35 reported that posting labels on vending machines did not affect the average number of calories sold. Hoefkens et al.39 and Lillico et al.40 in campus cafeterias and Schwartz et al.31 in a quick-service restaurant found that posting nutrition labels did not affect calorie intake. Dietary outcomes: noncaloric measures Five studies34,36,39,41,43 assessed the relationship between label exposure and macronutrient selection or intake, summarized in Table 4. Of the 12 studies reporting noncaloric and nonmacronutrient outcomes (eg, portion size, proportion of low-energy-density foods chosen), 9 reported that nutrition labels had some statistically significant positive effect, although a few identified an effect only in a subsample of participants. In pre–post studies in cafeterias where labels were implemented, Cinciripini45 found improved food group selections; Davis-Chervin et al.46 reported higher selection of low-calorie and low-cholesterol entrées in a cafeteria where posters were also displayed; Freedman47 found that portion sizes for fries decreased and salad dressing choices changed; and Hoefkens et al.39 reported higher vegetable consumption. In quick-service outlets, Nikolaou et al.48 found that posting labels resulted in decreased sales of both low-fat and high-fat items as well as decreases in sales of low-calorie items and much larger decreases in sales of high-calorie items; Roy et al32 found that sales of a high-calorie entrée decreased and sales of a lower-calorie meal increased. In a vending machine study, Larson-Brown49 found that sales of more-nutritious foods increased after labels were posted. In a laboratory, Temple et al.33,44 found that participants exposed to calorie labels ate fewer high-energy-density and low-energy-density foods and that participants exposed to traffic light labels were more likely to purchase green (or healthier) items. Three studies that examined noncaloric outcomes showed no statistically significant differences after label introduction. In a convenience store, Freedman and Connors50 reported a small increase in the percentage of tagged healthy items sold. In vending machine studies, Dingman et al.35 found that the proportion of “Better Choice” snacks purchased remained similar, and Hoerr and Louden30 found that the proportion of snacks in low-, moderate-, and high-nutrient-density groups did not differ upon label implementation. Meta-analysis Figure 229,31,34–41 shows the forest plots for the meta-analyses conducted on the 4 controlled experimental studies29,34–36 and the 6 pre–post studies.31,37–41 Meta-analyses results are shown in Table 5.29,31,34–41 Among controlled studies, exposure to simple textual nutrition labels was not associated with change in calories ordered or consumed (P = 0.4). Among pre–post studies, posting nutrition labels was associated with a decrease in the number of calories ordered or consumed by 36.0 kcal (95%CI, −60.2 to −11.8 kcal). Contextual labels (eg, traffic light, exercise equivalents, or list of daily suggested requirements) were more effective than simple calorie labels29,36,41 at improving dietary intake in all but 1 study.33 All 3 studies29,36,41 pooled for meta-analysis comparing textual vs contextual labels found contextual labels to be more effective, leading to a pooled estimated reduction of calories selected or consumed by 66.9 kcal (95%CI, −86.7 to −47.2 kcal). Table 5 Modeling results from the meta-analysis Reference  Study design  Experimental group kcal ± SD (group n)  Control group kcal ± SD (group n)  I2 index (%)  Pooled effect size (95%CI)  Model used  P value for dose–response effect from meta-regression  Aaron et al. (1995)34  Controlled  927 ± 27 (65)  906 ± 54 (25)  98.2  β = −20.8 (−69.3, 27.7)  Random-effects  0.374  Dingman et al. (2015)35  251 ± 21 (6170)  225 ± 56 (5538)  Ellison et al. (2014)29  724 ± 333 (312)  754 ± 339 (311)  James et al. (2015)36: selection  827 ± 61 (99)  902 ± 62 (99)  James et al. (2015)36: intake  722 ± 54 (99)  770 ± 53 (99)      Postlabeling  Prelabeling          Chu et al. (2009)37  Pre–post  635 ± 152 (14199)  648 ± 152 (13951)  98.6  β = −36.0 (−60.2, −11.8)  Random-effects  0.038  Hammond et al. (2015)38: selection  734 ± 331 (156)  825 ± 336 (149)  Hammond et al. (2015)38: intake  671 ± 327 (156)  769 ± 342 (149)  Hoefkens et al. (2011)39  598 ± 98 (224)  597 ± 114 (224)  Lillico et al. (2015)40  601 ± 282 (168)  661 ± 309 (131)  Nikolaou et al. (2014)41: men  692 ± 105 (507)  734 ± 101 (499)  Nikolaou et al. (2014)41: women  628 ± 105 (893)  709 ± 101 (901)  Schwartz et al. (2009)31  1033 ± 16 (294)  1020 ± 15 (299)      Contextual labels  Simple labels          Ellison et al. (2014)29  Contextual labels vs simple labels  683 ± 318 (355)  724 ± 333 (312)  86.4  β = −66.9 (−86.7, −47.2)  Random-effects  0.002  James et al. (2015)36: selection  763 ± 61 (102)  827 ± 61 (99)  James et al. (2015)36: intake  673 ± 53 (102)  722 ± 54 (99)  Nikolaou et al. (2014)41: men  534 ± 116 (952)  628 ± 105 (893)  Nikolaou et al. (2014)41: women  622 ± 116 (448)  692 ± 105 (507)  Reference  Study design  Experimental group kcal ± SD (group n)  Control group kcal ± SD (group n)  I2 index (%)  Pooled effect size (95%CI)  Model used  P value for dose–response effect from meta-regression  Aaron et al. (1995)34  Controlled  927 ± 27 (65)  906 ± 54 (25)  98.2  β = −20.8 (−69.3, 27.7)  Random-effects  0.374  Dingman et al. (2015)35  251 ± 21 (6170)  225 ± 56 (5538)  Ellison et al. (2014)29  724 ± 333 (312)  754 ± 339 (311)  James et al. (2015)36: selection  827 ± 61 (99)  902 ± 62 (99)  James et al. (2015)36: intake  722 ± 54 (99)  770 ± 53 (99)      Postlabeling  Prelabeling          Chu et al. (2009)37  Pre–post  635 ± 152 (14199)  648 ± 152 (13951)  98.6  β = −36.0 (−60.2, −11.8)  Random-effects  0.038  Hammond et al. (2015)38: selection  734 ± 331 (156)  825 ± 336 (149)  Hammond et al. (2015)38: intake  671 ± 327 (156)  769 ± 342 (149)  Hoefkens et al. (2011)39  598 ± 98 (224)  597 ± 114 (224)  Lillico et al. (2015)40  601 ± 282 (168)  661 ± 309 (131)  Nikolaou et al. (2014)41: men  692 ± 105 (507)  734 ± 101 (499)  Nikolaou et al. (2014)41: women  628 ± 105 (893)  709 ± 101 (901)  Schwartz et al. (2009)31  1033 ± 16 (294)  1020 ± 15 (299)      Contextual labels  Simple labels          Ellison et al. (2014)29  Contextual labels vs simple labels  683 ± 318 (355)  724 ± 333 (312)  86.4  β = −66.9 (−86.7, −47.2)  Random-effects  0.002  James et al. (2015)36: selection  763 ± 61 (102)  827 ± 61 (99)  James et al. (2015)36: intake  673 ± 53 (102)  722 ± 54 (99)  Nikolaou et al. (2014)41: men  534 ± 116 (952)  628 ± 105 (893)  Nikolaou et al. (2014)41: women  622 ± 116 (448)  692 ± 105 (507)  Figure 2 View largeDownload slide Forest plots showing calories selected or consumed by labeling condition for controlled studies (top) and by pre–post studies (bottom) reporting calories as an outcome.Note: Axes have different scales. Abbreviation: WMD, weighted mean difference. Figure 2 View largeDownload slide Forest plots showing calories selected or consumed by labeling condition for controlled studies (top) and by pre–post studies (bottom) reporting calories as an outcome.Note: Axes have different scales. Abbreviation: WMD, weighted mean difference. Study quality Studies included in the review on average scored 5.2 out of 9 points (Table 629–41,43–50). Studies scored particularly low (an average of 0.14 out of 1 point) on including a power analysis or reasoning for the sample size, and on randomization of participants or study setting (0.27 out of 1 point). Studies scored substantially higher on objectively observing dietary intake (0.86 out of 1 point) rather than relying on surveys, and on performing the study in a natural setting (0.82 out of 1 point). Of the 9 studies with the highest quality, 7 indicated that label exposure was related to better dietary intake for at least some groups. Of the 6 studies with the lowest quality, 5 showed some dietary improvement upon label exposure. Table 6 Scoring of study quality. Points awarded on the basis of criteria met for each study included in the review Reference  Documented procedures  Sample size  Reasoning or power analysis  Control group or cafeteria  Randomized participants or settings  Observed dietary outcome  Assessed intake  Natural setting  Accounted for potential confounders  Total points  Aaron et al. (1995)34  1  0  0  1  0  1  1  1  0  5  Chu et al. (2009)37  1  1  0  0  0  1  0  1  0  4  Cioffi et al. (2015)43  1  1  0  0  0  1  0  1  0  4  Cinciripini (1984)45  1  1  0  0  0  1  0  1  1  5  Davis-Chervin et al. (1985)46  1  1  0  1  0  1  0  1  0  5  Dingman et al. (2015)35  1  1  0  1  1  1  0  1  1  7  Ellison et al. (2014)29  1  1  0  1  1  1  0  1  0  6  Freedman (2011)47  1  1  0  0  0  1  0  1  0  4  Freedman et al. (2010)50,a  1  0  0  0  0  1  0  1  0  3  Hammond et al. (2015)38,b  1  1  0  1  0  0  1  1  1  6  Hoefkens et al. (2011)39,b  1  1  1  1  0  0  1  1  0  6  Hoerr & Louden (1993)30  1  1  1  0  0  1  0  1  0  5  James et al. (2015)36  1  1  1  1  1  1  1  0  1  8  Larson-Brown (1978)49  1  1  0  0  0  1  0  1  0  4  Lillico et al. (2015)40,b  1  1  0  1  0  0  1  1  1  6  Nikolaou et al. (2014)A41  1  1  0  0  0  1  0  1  1  5  Nikolaou et al. (2014)B48  1  1  0  1  0  1  0  1  0  5  Roy et al. (2016)32  1  1  0  0  0  1  0  1  0  4  Schwartz et al. (2012)31,c  1  1  0  0  0  1  1  1  0  5  Temple et al. (2010)44  1  0  0  1  1  1  1  0  1  6  Temple et al. (2011)A33  1  0  0  1  1  1  1  0  1  6  Temple et al. (2011)B33  1  0  0  1  1  1  1  0  1  6  Average of all studies  1  0.77  0.14  0.55  0.27  0.86  0.41  0.82  0.41  5.2  Reference  Documented procedures  Sample size  Reasoning or power analysis  Control group or cafeteria  Randomized participants or settings  Observed dietary outcome  Assessed intake  Natural setting  Accounted for potential confounders  Total points  Aaron et al. (1995)34  1  0  0  1  0  1  1  1  0  5  Chu et al. (2009)37  1  1  0  0  0  1  0  1  0  4  Cioffi et al. (2015)43  1  1  0  0  0  1  0  1  0  4  Cinciripini (1984)45  1  1  0  0  0  1  0  1  1  5  Davis-Chervin et al. (1985)46  1  1  0  1  0  1  0  1  0  5  Dingman et al. (2015)35  1  1  0  1  1  1  0  1  1  7  Ellison et al. (2014)29  1  1  0  1  1  1  0  1  0  6  Freedman (2011)47  1  1  0  0  0  1  0  1  0  4  Freedman et al. (2010)50,a  1  0  0  0  0  1  0  1  0  3  Hammond et al. (2015)38,b  1  1  0  1  0  0  1  1  1  6  Hoefkens et al. (2011)39,b  1  1  1  1  0  0  1  1  0  6  Hoerr & Louden (1993)30  1  1  1  0  0  1  0  1  0  5  James et al. (2015)36  1  1  1  1  1  1  1  0  1  8  Larson-Brown (1978)49  1  1  0  0  0  1  0  1  0  4  Lillico et al. (2015)40,b  1  1  0  1  0  0  1  1  1  6  Nikolaou et al. (2014)A41  1  1  0  0  0  1  0  1  1  5  Nikolaou et al. (2014)B48  1  1  0  1  0  1  0  1  0  5  Roy et al. (2016)32  1  1  0  0  0  1  0  1  0  4  Schwartz et al. (2012)31,c  1  1  0  0  0  1  1  1  0  5  Temple et al. (2010)44  1  0  0  1  1  1  1  0  1  6  Temple et al. (2011)A33  1  0  0  1  1  1  1  0  1  6  Temple et al. (2011)B33  1  0  0  1  1  1  1  0  1  6  Average of all studies  1  0.77  0.14  0.55  0.27  0.86  0.41  0.82  0.41  5.2  a Number of observations not reported. b Studies utilizing pre–post cohort designs were considered as having controls, though participants served as their own controls. c While the Schwartz et al. (2012)31 study assessed intake in another experiment in the same article, the experiment included in this review did not assess intake. DISCUSSION The present systematic review examined the effect of nutrition label use on diet among college students. Overall, 16 of the 22 studies included in the review reported that exposure to nutrition labels led to improved dietary choices. Eight of the 13 studies involving caloric outcomes found that posting nutrition labels at the point of purchase decreased calorie selection or consumption. Nine of the 12 studies measuring noncaloric measures of dietary quality such as food group choices found that introducing labels improved dietary quality. In the 10 studies pooled for meta-analysis, controlled experimental studies showed a nonsignificant decrease in calories selected and/or consumed in the presence of labels, whereas pre–post studies showed a significant decrease of 36 kcal in calories selected and/or consumed in the presence of labels. Studies of both relatively low and relatively high quality produced similar results, with a majority of both higher-quality and lower-quality studies showing that nutrition label exposure improved dietary intake in at least some groups. Setting might be crucial to measuring the effectiveness of nutrition labels. Of the 12 studies conducted in college cafeterias in the present review, 10 found positive effects,32,37–41,43,45–48 1 found no effect,40 and 1 found a negative effect34 of nutrition labels. Studies in laboratories generally showed positive effects, those in quick-service outlets and convenience stores were more mixed but overall positive, and those in vending machines showed few effects. The magnitude of effects was often small, and even those studies reporting overall significant effects had subgroups that were not always affected by the intervention. Prior systematic reviews in general populations have also shown that setting is crucial: Long et al.19 reported that labels significantly decreased calories ordered in nonrestaurant but not restaurant settings, and Fernandes et al.7 found menu labeling was more effective in cafeterias than in restaurants. In the latter study, the authors hypothesized that this effect could be related to educational level and the daily nature of cafeteria usage contrasted with the special-occasion nature of restaurant visits.7 The potential interaction between setting, nutrition label use, and dietary outcomes should be investigated, particularly for the effect on daily food patterns and noncaloric outcomes. In addition, it is important to consider barriers such as hunger and food cost51 as well as how nutrition labels may act together with other nutrition interventions in college settings, including price incentives, changes in food offerings, and control of portion size.25 In the present review, when compared with simple calorie labels, contextual labels (traffic light, exercise, and those containing daily intake recommendations) tended to be more effective at improving dietary intake, resulting in an average of 67 fewer calories ordered or consumed. Prior reviews and studies in adults have reported that labels containing contextual information are better understood16 and more effective at reducing intake of calories,15,17,52 total fat, saturated fat, and sodium52 and at improving food choices.21 Thus, the results of both the present review and prior studies indicate that contextual or interpretive labels such as traffic light labels or exercise equivalents are more effective at improving dietary intake. Contextual labels that include several components, such as the star ratings employed by Hoefkens et al.,39 may also have an added benefit of providing a more holistic approach to dietary quality that does not focus solely on calories. The wide variety of dietary outcomes assessed in the reviewed studies included calories, macronutrients, micronutrients, sales of items deemed healthful using different standards, cholesterol, energy density, and food groups. This array of outcomes was crucial for obtaining a comprehensive view of the effect of labels on overall dietary quality. However, this also meant that study results were difficult to pool, and estimates of the overall effect of labels were nontrivial. Future research should investigate the effect of nutrition labels on comprehensive measures of dietary quality rather than on calories alone. To ensure results can be compared and pooled across studies, standardized measures should be used, such as the Healthy Eating Index,53 the Mediterranean Diet Score,54 dietary quality indices,55,56 or nutrients present on nutrition labels.57 The results of this review support menu-labeling policies such as the US Food and Drug Administration’s menu-labeling rule, which will require restaurants and food retail establishments with 20 or more locations to post calorie labels on menus starting in May 2018.58 Posting point-of-purchase information is critical for providing consumers adequate information to make dietary choices, and the results of this review suggest that, among college students eating on campus, exposure to nutrition labels is likely to improve dietary intake. However, in the United States, college and university cafeterias and restaurants will not be required to post menu labels unless they have 20 or more locations offering similar products. Educational institutions should consider proactively implementing nutrition labeling, especially using interpretive labels to help students compare dietary options quickly and easily. In addition to potentially improving dietary intake, menu labeling may also decrease food costs; Nikolaou et al.41 found that, compared with data from the prior year, catering orders for overall calories, ingredients used primarily for desserts, and oils for frying all decreased substantially when labels were posted. Another potential benefit to nutrition labeling is that it may encourage product and recipe reformulation,59 which could improve dietary intake, even for consumers who do not consciously use nutrition labels. A few limitations of the review and included studies should be noted. The wide variety of study designs, outcomes, and even label presentation formats limits the ability to pool results. While it was crucial to aggregate studies separately on the basis of study design and label type, this meant that 2 of the meta-analyses included fewer than 5 studies, which could limit the generalizability, as meta-analyses are stronger when they encompass a larger number of studies with similar designs. Only 9 of the 22 included studies had comparison groups where participants did not serve as their own controls, although 17 of the 22 studies included a pre–post comparison. The need for studies with comparison groups has been highlighted in a previous review, which reported that nutrition-label interventions in real-world environments with comparison groups did not produce a significant decrease in calories ordered.6 Additionally, a few studies included only subgroup effects rather than overall effects, 2 studies38,40 may have had sample overlap, and several studies included relatively small samples for testing a population-level intervention. Lastly, title and abstract reviews were conducted by only 1 investigator. One strength of this review compared with other recent reviews6,15,60 is that, by limiting the population to college students, multiple outcomes of dietary quality beyond calories selected or consumed were assessed. This is an important distinction, as some data suggest that, while nutrition label users may eat similarly to nonusers in terms of food amount, there are meaningful differences in the foods selected.61 Thus, this review is able to comment on overall dietary quality, which has been shown to relate to long-term health outcomes.62 In addition, this review compared the relationship between nutrition label use and dietary quality across different settings, thus showing that the majority of studies in some settings (eg, cafeterias and laboratories) showed a positive effect, whereas studies in other settings (eg, vending machines) largely showed mixed results or no effect on dietary quality. Finally, this review considered study design within the meta-analysis, which is important for reviewing studies with vastly different designs. CONCLUSION The present systematic review and meta-analysis examined the effect of nutrition labels on diet among college students. Among the 22 studies included in the review, nutrition labels were found to have a moderate but significant positive effect on dietary choices in college students. These effects were modified by individual sociodemographics, setting, and type of labels used. Studies in cafeterias and laboratories generally produced more positive effects than those in quick-service restaurants or vending machines. Contextual labels listing daily recommended intake or including traffic lights or exercise equivalents displayed higher efficacy in this population. Both higher-quality and lower-quality studies generally showed positive effects of labeling. Field experiments, particularly with large representative samples and adequate controls, are warranted to assess the effect of nutrition labels among college students. The results of this study support nutrition-labeling policies, suggesting that implementing nutrition labels may improve dietary intake among college students. Colleges, universities, and other institutions should consider implementing nutrition labeling, particularly using contextual formats that allow for quick comparisons across food choices. Acknowledgments Author contributions. M.J.C. conceptualized the study, performed the systematic review and meta-analysis, and drafted the manuscript. R.A. oversaw the methods and contributed to each manuscript draft. Funding/support. M.J.C. is supported by the Health Resources and Services Administration (HRSA) of the US Department of Health and Human Services (HHS) under the National Research Service Award (NRSA) in Primary Medical Care, grant no. T32HP22239 (PI: Borowsky). This information or content and conclusions are those of the authors and should not be construed as the official position or policy of, nor should any endorsements be inferred by, the HRSA, the HHS, or the US government. Funding agencies were not involved in the preparation of this manuscript. Declaration of interest. The authors have no relevant interests to declare. References 1 Guthrie JF, Lin BH, Frazao E. Role of food prepared away from home in the American diet, 1977-78 versus 1994-96: changes and consequences. J Nutr Educ Behav.  2002; 34: 140– 150. http://dx.doi.org/10.1016/S1499-4046(06)60083-3 Google Scholar CrossRef Search ADS PubMed  2 Kant AK, Graubard BI. Eating out in America, 1987–2000: trends and nutritional correlates. Prev Med.  2004; 38: 243– 249. http://dx.doi.org/10.1016/j.ypmed.2003.10.004 Google Scholar CrossRef Search ADS PubMed  3 National Restaurant Association. Facts at a Glance. News and Research 2017. http://www.restaurant.org/News-Research/Research/Facts-at-a-Glance. Updated April 28, 2017. Accessed September 13, 2017. 4 Bezerra IN, Curioni C, Sichieri R. Association between eating out of home and body weight. Nutr Rev.  2012; 70: 65– 79. http://dx.doi.org/10.1111/j.1753-4887.2011.00459.x Google Scholar CrossRef Search ADS PubMed  5 Swartz J, Braxton D, Viera A. Calorie menu labeling on quick-service restaurant menus: an updated systematic review of the literature. Int J Behav Nutr Phys Act.  2011; 8: 135. doi:10.1186/1479-5868-8-135 Google Scholar CrossRef Search ADS PubMed  6 Kiszko K, Martinez O, Abrams C, et al.   The influence of calorie labeling on food orders and consumption: a review of the literature. J Community Health.  2014; 39: 1248– 1269. http://dx.doi.org/10.1007/s10900-014-9876-0 Google Scholar CrossRef Search ADS PubMed  7 Fernandes AC, Oliveira RC, Proenca RP, et al.   Influence of menu labeling on food choices in real-life settings: a systematic review. Nutr Rev.  2016; 74: 534– 548. http://dx.doi.org/10.1093/nutrit/nuw013 Google Scholar CrossRef Search ADS PubMed  8 Kim SY, Nayga RM, Capps O. The effect of food label use on nutrient intakes: an endogenous switching regression analysis. J Agri Resource Econ . 2000; 215– 231. 9 Christoph MJ, An R, Ellison B. Correlates of nutrition label use among college students and young adults: a review. Public Health Nutr.  2016; 19: 2135– 2148. http://dx.doi.org/10.1017/S1368980015003183 Google Scholar CrossRef Search ADS PubMed  10 Bleich SN, Wolfson JA. Differences in consumer use of food labels by weight loss strategies and demographic characteristics. BMC Public Health . 2015; 15: 1275. doi: 10.1186/s12889-015-2651-z Google Scholar CrossRef Search ADS PubMed  11 Tandon PS, Wright J, Chuan Z, et al.   Nutrition menu labeling may lead to lower-calorie restaurant meal choices for children. Pediatrics . 2010; 125: 244– 248. http://dx.doi.org/10.1542/peds.2009-1117 Google Scholar CrossRef Search ADS PubMed  12 Pratt NS, Ellison BD, Benjamin AS, et al.   Improvements in recall and food choices using a graphical method to deliver information of select nutrients. Nutr Res.  2016; 36: 44– 56. http://dx.doi.org/10.1016/j.nutres.2015.10.009 Google Scholar CrossRef Search ADS PubMed  13 Roberto CA, Larsen PD, Agnew H, et al.   Evaluating the impact of menu labeling on food choices and intake. Am J Public Health.  2010; 100: 312– 318. http://dx.doi.org/10.2105/AJPH.2009.160226 Google Scholar CrossRef Search ADS PubMed  14 Ellison B, Lusk JL, Davis D. Looking at the label and beyond: the effects of calorie labels, health consciousness, and demographics on caloric intake in restaurants. Int J Behav Nutr Phys Act.  2013; 10: 21. doi:10.1186/1479-5868-10-21 Google Scholar CrossRef Search ADS PubMed  15 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. http://dx.doi.org/10.1016/j.jand.2014.05.014 Google Scholar CrossRef Search ADS PubMed  16 Cowburn G, Stockley L. Consumer understanding and use of nutrition labelling: a systematic review. Public Health Nutr.  2005; 8: 21– 28. http://dx.doi.org/10.1079/PHN2005666 Google Scholar CrossRef Search ADS PubMed  17 Liu PJ, Roberto CA, Liu LJ, et al.   A test of different menu labeling presentations. Appetite . 2012; 59: 770– 777. http://dx.doi.org/10.1016/j.appet.2012.08.011 Google Scholar CrossRef Search ADS PubMed  18 van Herpen E, Trijp HC. Front-of-pack nutrition labels. Their effect on attention and choices when consumers have varying goals and time constraints. Appetite . 2011; 57: 148– 160. Google Scholar CrossRef Search ADS PubMed  19 Long MW, Tobias DK, Cradock AL, et al.   Systematic review and meta-analysis of the impact of restaurant menu calorie labeling. Am J Public Health.  2015; 105: e11– e24. Google Scholar CrossRef Search ADS PubMed  20 Harnack L, French S. Effect of point-of-purchase calorie labeling on restaurant and cafeteria food choices: a review of the literature. Int J Behav Nutr Phys Act.  2008; 5: 51. doi:10.1186/1479-5868-5-51 Google Scholar CrossRef Search ADS PubMed  21 Cecchini M, Warin L. Impact of food labelling systems on food choices and eating behaviours: a systematic review and meta-analysis of randomized studies. Obes Rev.  2016; 17: 201– 210. http://dx.doi.org/10.1111/obr.12364 Google Scholar CrossRef Search ADS PubMed  22 Harris K, Gordon-Larsen P, Chantala K, et al.   Longitudinal trends in race/ethnic disparities in leading health indicators from adolescence to young adulthood. Arch Pediatr Adolesc Med.  2006; 160: 74– 81. http://dx.doi.org/10.1001/archpedi.160.1.74 Google Scholar CrossRef Search ADS PubMed  23 Nelson MC, Story M, Larson NI, et al.   Emerging adulthood and college-aged youth: an overlooked age for weight-related behavior change. Obesity . 2008; 16: 2205– 2211. http://dx.doi.org/10.1038/oby.2008.365 Google Scholar CrossRef Search ADS PubMed  24 Vella-Zarb RA, Elgar FJ. The ‘Freshman 5’: a meta-analysis of weight gain in the freshman year of college. J Am Coll Health.  2009; 58: 161– 166. Google Scholar CrossRef Search ADS PubMed  25 Roy R, Kelly B, Rangan A, et al.   Food environment interventions to improve the dietary behavior of young adults in tertiary education settings: a systematic literature review. J Acad Nutr Diet.  2015; 115: 1647– 1681. http://dx.doi.org/10.1016/j.jand.2015.06.380 Google Scholar CrossRef Search ADS PubMed  26 Kelly NR, Mazzeo SE, Bean MK. Systematic review of dietary interventions with college students: directions for future research and practice. J Nutr Educ Behav . 2013; 45: 304– 313. http://dx.doi.org/10.1016/j.jneb.2012.10.012 Google Scholar CrossRef Search ADS PubMed  27 Deliens T, Van Crombruggen R, Verbruggen S, et al.   Dietary interventions among university students: a systematic review. Appetite . 2016; 105: 14– 26. http://dx.doi.org/10.1016/j.appet.2016.05.003 Google Scholar CrossRef Search ADS PubMed  28 Liberati A, Altman DG, Tetzlaff J, et al.   The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. PLoS Med . 2009; 6: e1000100. 10.1371/journal.pmed.1000100 Google Scholar CrossRef Search ADS PubMed  29 Ellison B, Lusk JL, Davis D. The impact of restaurant calorie labels on food choice: results from a field experiment. Econ Inquiry . 2014; 52: 666– 681. http://dx.doi.org/10.1111/ecin.12069 Google Scholar CrossRef Search ADS   30 Hoerr SM, Louden VA. Can nutrition information increase sales of healthful vended snacks? J Sch Health.  1993; 63: 386– 390. Google Scholar CrossRef Search ADS PubMed  31 Schwartz J, Riis J, Elbel B, et al.   Inviting consumers to downsize fast-food portions significantly reduces calorie consumption. Health Aff . 2012; 31: 399– 407. http://dx.doi.org/10.1377/hlthaff.2011.0224 Google Scholar CrossRef Search ADS   32 Roy R, Beattie-Bowers J, Ang SM, et al.   The effect of energy labelling on menus and a social marketing campaign on food-purchasing behaviours of university students. BMC Public Health . 2016; 16: 727. doi:10.1186/s12889-016-3426-x Google Scholar CrossRef Search ADS PubMed  33 Temple JL, Johnson KM, Archer K, et al.   Influence of simplified nutrition labeling and taxation on laboratory energy intake in adults. Appetite . 2011; 57: 184– 192. http://dx.doi.org/10.1016/j.appet.2011.04.018 Google Scholar CrossRef Search ADS PubMed  34 Aaron JI, Evans RE, Mela DJ. Paradoxical effect of a nutrition labeling scheme in a student cafeteria. Nutr Res . 1995; 15: 1251– 1261. http://dx.doi.org/10.1016/0271-5317(95)02001-C Google Scholar CrossRef Search ADS   35 Dingman DA, Schulz MR, Wyrick DL, et al.   Does providing nutrition information at vending machines reduce calories per item sold? J Public Health Pol.  2015; 36: 110– 122. Google Scholar CrossRef Search ADS   36 James A, Adams-Huet B, Shah M. Menu labels displaying the kilocalorie content or the exercise equivalent: effects on energy ordered and consumed in young adults. Am J Health Promot.  2015; 29: 294– 302. http://dx.doi.org/10.4278/ajhp.130522-QUAN-267 Google Scholar CrossRef Search ADS PubMed  37 Chu Y, Frongillo E, Jones S, et al.   Improving patrons' meal selections through the use of point-of-selection nutrition labels. Am J Public Health.  2009; 99: 2001– 2005. http://dx.doi.org/10.2105/AJPH.2008.153205 Google Scholar CrossRef Search ADS PubMed  38 Hammond D, Lillico HG, Vanderlee L, et al.   The impact of nutrition labeling on menus: a naturalistic cohort study. Am J Health Behav.  2015; 39: 540– 548. http://dx.doi.org/10.5993/AJHB.39.4.10 Google Scholar CrossRef Search ADS PubMed  39 Hoefkens C, Lachat C, Kolsteren P, et al.   Posting point-of-purchase nutrition information in university canteens does not influence meal choice and nutrient intake. Am J Clin Nutr.  2011; 94: 562– 570. http://dx.doi.org/10.3945/ajcn.111.013417 Google Scholar CrossRef Search ADS PubMed  40 Lillico HG, Hanning R, Findlay S, et al.   The effects of calorie labels on those at high-risk of eating pathologies: a pre-post intervention study in a University cafeteria. Public Health . 2015; 129: 732– 739. http://dx.doi.org/10.1016/j.puhe.2015.03.005 Google Scholar CrossRef Search ADS PubMed  41 Nikolaou CK, Hankey CR, Lean MEJ. Preventing weight gain with calorie-labeling. Obesity (Silver Spring).  2014; 22: 2277– 2283. http://dx.doi.org/10.1002/oby.20885 Google Scholar CrossRef Search ADS PubMed  42 National Heart, Blood, and Lung Institute. Study Quality Assessment Tools. 2014. https://www.nhlbi.nih.gov/health-topics/study-quality-assessment-tools. Accessed December 28, 2017. . 43 Cioffi CE, Levitsky DA, Pacanowski CR, et al.   A nudge in a healthy direction. The effect of nutrition labels on food purchasing behaviors in university dining facilities. Appetite . 2015; 92: 7– 14. http://dx.doi.org/10.1016/j.appet.2015.04.053 Google Scholar CrossRef Search ADS PubMed  44 Temple JL, Johnson K, Recupero K, et al.   Nutrition labels decrease energy intake in adults consuming lunch in the laboratory. J Am Diet Assoc . 2010; 110: 1094– 1097. http://dx.doi.org/10.1016/j.jada.2010.04.006 Google Scholar CrossRef Search ADS PubMed  45 Cinciripini PM. Changing food selections in a public cafeteria: an applied behavior analysis. Behav Modif.  1984; 8: 520– 539. http://dx.doi.org/10.1177/01454455840084004 Google Scholar CrossRef Search ADS   46 Davis-Chervin D, Rogers T, Clark M. Influencing food selection with point-of-choice nutrition information. J Nutr Educ . 1985; 17: 18– 22. http://dx.doi.org/10.1016/S0022-3182(85)80014-5 Google Scholar CrossRef Search ADS   47 Freedman MR. Point-of-selection nutrition information influences choice of portion size in an all-you-can-eat university dining hall. J Foodservice Business Res . 2011; 14: 86– 98. http://dx.doi.org/10.1080/15378020.2011.548228 Google Scholar CrossRef Search ADS   48 Nikolaou CK, Lean MEJ, Hankey CR. Calorie-labelling in catering outlets: acceptability and impacts on food sales. Prev Med.  2014; 67: 160– 165. http://dx.doi.org/10.1016/j.ypmed.2014.07.027 Google Scholar CrossRef Search ADS PubMed  49 Larson-Brown LB. Point-of-purchase information on vended foods. J Nutr Educ . 1978; 10: 116– 118. http://dx.doi.org/10.1016/S0022-3182(78)80053-3 Google Scholar CrossRef Search ADS   50 Freedman MR, Connors R. Point-of-purchase nutrition information influences food-purchasing behaviors of college students: a pilot study. J Am Diet Assoc . 2010; 111: S42– S46. Google Scholar CrossRef Search ADS   51 Stran KA, Knol LL, Turner LW, et al.   College students must overcome barriers to use calorie labels in fast-food restaurants. J Nutr Educ Behav.  2016; 48: 122.e1– 130.e121. Google Scholar CrossRef Search ADS   52 Emrich TE, Qi Y, Lou WY, et al.   Traffic-light labels could reduce population intakes of calories, total fat, saturated fat, and sodium. PLoS One.  2017; 12: e0171188. doi:10.1016/j.jneb.2015.09.009 Google Scholar CrossRef Search ADS PubMed  53 Guenther PM, Casavale KO, Reedy J, et al.   Update of the Healthy Eating Index: HEI-2010. J Acad Nutr Diet.  2013; 113: 569– 580. http://dx.doi.org/10.1016/j.jand.2012.12.016 Google Scholar CrossRef Search ADS PubMed  54 Panagiotakos DB, Pitsavos C, Stefanadis C. Dietary patterns: a Mediterranean diet score and its relation to clinical and biological markers of cardiovascular disease risk. Nutr Metab Cardiovasc Dis . 2006; 16: 559– 568. http://dx.doi.org/10.1016/j.numecd.2005.08.006 Google Scholar CrossRef Search ADS PubMed  55 Cooke R, Papadaki A. Nutrition label use mediates the positive relationship between nutrition knowledge and attitudes towards healthy eating with dietary quality among university students in the UK. Appetite . 2014; 83: 297– 303. http://dx.doi.org/10.1016/j.appet.2014.08.039 Google Scholar CrossRef Search ADS PubMed  56 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. http://dx.doi.org/10.1016/j.jada.2011.08.047 Google Scholar CrossRef Search ADS PubMed  57 Ollberding NJ, Wolf RL, Contento I. Food label use and its relation to dietary intake among US adults. J Am Diet Assoc . 2010; 110: 1233– 1237. http://dx.doi.org/10.1016/j.jada.2010.05.007 Google Scholar CrossRef Search ADS PubMed  58 Menu labeling requirements. Final rule. Fed Regist.  2014; 79 230: 71155– 71259. PubMed  59 Hawkes C, Smith TG, Jewell J, et al.   Smart food policies for obesity prevention. Lancet.  2015; 385: 2410– 2421. http://dx.doi.org/10.1016/S0140-6736(14)61745-1 Google Scholar CrossRef Search ADS PubMed  60 Nikolaou CK, Hankey CR, Lean MEJ. Calorie-labelling: does it impact on calorie purchase in catering outlets and the views of young adults? Int J Obes (Lond).  2015; 39: 542– 545. Google Scholar CrossRef Search ADS PubMed  61 Christoph MJ, Ellison B. A cross-sectional study of the relationship between nutrition label use and food selection, servings, and consumption in a university dining setting. J Acad Nutr Diet . 2017; 117: 1528– 1537. http://dx.doi.org/10.1016/j.jand.2017.01.027 Google Scholar CrossRef Search ADS PubMed  62 Steffen LM, Van Horn L, Daviglus ML, et al.   A modified Mediterranean diet score is associated with a lower risk of incident metabolic syndrome over 25 years among young adults: the CARDIA (Coronary Artery Risk Development in Young Adults) study. Br J Nutr.  2014; 112: 1654– 1661. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press on behalf of the International Life Sciences Institute. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Nutrition Reviews Oxford University Press

Effect of nutrition labels on dietary quality among college students: a systematic review and meta-analysis

Loading next page...
 
/lp/ou_press/effect-of-nutrition-labels-on-dietary-quality-among-college-students-a-jpzQf5VilU
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the International Life Sciences Institute. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com.
ISSN
0029-6643
eISSN
1753-4887
D.O.I.
10.1093/nutrit/nux069
Publisher site
See Article on Publisher Site

Abstract

Abstract Context College students are at an elevated risk of poor nutrition and eating habits. Objective The aim of this systematic review was to examine and quantify the effect of nutrition labels on diet quality in college students. Data Sources Literature searches were conducted in 4 electronic databases. Study Selection Peer-reviewed publications that assessed the effect of nutrition label use on food choice or dietary intake in college students were included. Data Extraction Twenty-two randomized controlled trials, cohort studies, and pre–post studies were identified. Results Sixteen studies found label exposure to be associated with improved diet. Of the 13 studies reporting calories selected or consumed, 8 found that posting labels at the point of purchase decreased calories, 4 found no effect, and 1 found that calories consumed increased after posting labels. Nine of the 12 studies assessing noncaloric measures found that nutrition labels positively affected diet quality. Meta-analysis of pre–post studies found a decrease of 36 kcal (P < 0.05) with label exposure. Conclusions Nutrition labels had a moderate but positive effect on dietary intake of college students. college, diet, menu labels, nutrition labels, university INTRODUCTION Food purchase and consumption outside the home has risen in the last 30 years,1,2 now accounting for almost 50% of the food expenditures among Americans.3 Since greater frequency of eating outside the home has been associated with higher body weight,4 labeling of calorie and nutrition information on restaurant menus has emerged as a tool to enable consumers to make informed food selections. However, recent reviews have questioned the efficacy of posting calorie information, indicating the lack of effect on calorie purchase or consumption.5–7 The null effect could partially be due to differences in populations studied and the wide variety of study designs and outcomes assessed. Further, the motivation and reasoning behind label usage is unclear, as several survey-based studies found that individuals who reported using nutrition labels had higher nutrition knowledge or motivation to eat in a specific way8,9 or to lose weight,10 whereas others found nutrition labels to be effective across consumer characteristics,11–13 or even more effective for those with lower health consciousness.14 Besides being affected by consumer characteristics, research on the effect of nutrition labels may be confounded by the fact that certain types of labels tend to be more effective than others. A recent meta-analysis reported that, while standard calorie labeling did not affect calorie selection or consumption, both outcomes decreased when labels included contextual information such as daily intake or traffic light symbols.15 Interpretational aids and cues help consumers better understand and use labels,16 and studies have shown that traffic light labels in particular are more effective than simple textual labels at decreasing calorie selection.17,18 Another major gap in understanding the effect of nutrition labels lies within the limited number of dietary outcomes assessed. While several reviews have focused on calories selected and consumed,6,15,19 fewer have used broader definitions of dietary intake. Those that have observed broader measures are generally limited by including only a small number of studies.20,21 Since diet measures vary widely across studies, pooling results to assess the overall effect is often infeasible. College students are a population at an elevated risk of poor nutrition and eating habits. Dietary quality22 and fruit and vegetable intake tend to decrease during emerging adulthood.23 Further, a meta-analysis estimated that college students gain about 4 pounds on average during their freshman year.24 While previous studies have reviewed prevalence and predictors of nutrition label use9 and the effects of dietary interventions among college students,25–27 the effect of nutrition labels in particular is unknown. Reviewing the effectiveness of nutrition labels in a group with relatively homogenous age and education may allow more precise identification of factors interacting with label use and diet quality. A systematic review of the effect of nutrition label use on diet quality among college students was conducted. The hypothesis that nutrition label use could improve diet quality by helping college students make more healthful food choices was tested. In addition, a meta-analysis was conducted to quantify the influence of nutrition label use on dietary quality. METHODS Systematic review and meta-analysis procedures were conducted in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines.28 Study selection criteria Studies were included in this review if the following criteria were met: (1) study adopted a design of a randomized control trial, a cohort study, or a pre–post study; (2) study took place on a college campus; (3) study assessed nutrition label use (nutrition facts, nutrition labels specific to foods in cafeterias or dining units, or general nutrition labels) as a predictor of diet quality or food choice; (4) study assessed actual food choice or dietary intake of more than 1 food item as the outcome; (5) study was published in a peer-reviewed publication; (6) study was written in English; and (7) study was published on or before May 18, 2017. PICOS (population, intervention, comparator, outcome, setting) criteria are listed in Table 1. Studies adopting a qualitative, case report, case–control, or cross-sectional design, published in a language other than English, not peer reviewed, not occurring on a college campus, requiring participants to read labels as a prerequisite for participation (including labels that were not nutrition labels), not providing a direct test of the relationship between label use and dietary quality or food choice independent of other factors such as food availability or education, using experiments that measured intake of only 1 item (eg, a snack bar), or reporting hypothetical choices or intentions rather than actual food choice or diet quality were excluded. Additionally, studies wherein the average age of participants was over 30 years or the majority of participants were over 30 years old were excluded. In the case of studies that met the inclusion criteria but also had additional interventions such as price changes,29,30 downsizing offers,31 a social media campaign,32 or food taxes,33 only the results of the labeling intervention were summarized. Table 1 PICOS criteria for inclusion of studies Parameter  Inclusion criteria  Population  College students  Intervention  Nutrition labels on food and beverage products served on college campuses  Comparator  Exposure to nutrition labels compared with no exposure  Outcome  Diet and food choices  Setting  College, university, and tertiary education campuses  Parameter  Inclusion criteria  Population  College students  Intervention  Nutrition labels on food and beverage products served on college campuses  Comparator  Exposure to nutrition labels compared with no exposure  Outcome  Diet and food choices  Setting  College, university, and tertiary education campuses  Search strategy PubMed, EBSCO, PsycInfo, and Web of Science databases were searched using combinations of the following keywords: (1) “nutrition,” “calorie,” “food,” “diet,” or “menu”; (2) “label,” “labeling,” or “labelling”; (3) “dietary quality,” “diet,” “dietary intake,” “food intake,” “caloric intake,” “calorie intake,” “nutritional quality,” “nutritional intake,” “food choice,” “meal choice,” “food selection,” “food consumption,” “meal selection,” “meal consumption,” or “eating”; and (4) “college student(s),” “university student(s),” “young adult(s),” “university,” “college,” or “tertiary education.” The following keywords were used to exclude articles to limit the number of harvests: “supplement,” “pharmacology,” “medication,” “allerg*,” “mice,” “rat,” “choline,” “anemia,” “anorexia,” and “cigarette.” For example, the specific search terms used in PubMed are listed in Table 2. Table 2 Search terms used in PubMed Predictor terms  (“nutrition label” OR “nutrition labels” OR “nutrition labeling” OR “nutrition labelling” OR “calorie labels” OR “calorie label” OR “calorie labelling” OR “calorie labeling” OR “food label” OR “food labels” OR “menu label” OR “menu labels” OR “menu labeling” OR “menu labelling” OR “label usage” OR “label use”)  Outcome terms  AND (“dietary quality” OR “diet” OR “dietary intake” OR “food intake” OR “caloric intake” OR “calorie intake” OR “nutritional quality” OR “nutritional intake” OR “food choice” OR “meal choice” OR “food selection” OR “food consumption” OR “meal selection” OR “meal consumption” OR “eating”)  Population terms  AND (“college student” OR “college students” OR “university student” OR “university students” OR “young adult” OR “young adults” OR “university” OR “college” OR “tertiary education”)  Exclusionary terms  NOT supplement NOT pharmacology NOT medication NOT allerg* NOT mice NOT rat NOT cigarette NOT choline NOT anemia NOT anorexia  Predictor terms  (“nutrition label” OR “nutrition labels” OR “nutrition labeling” OR “nutrition labelling” OR “calorie labels” OR “calorie label” OR “calorie labelling” OR “calorie labeling” OR “food label” OR “food labels” OR “menu label” OR “menu labels” OR “menu labeling” OR “menu labelling” OR “label usage” OR “label use”)  Outcome terms  AND (“dietary quality” OR “diet” OR “dietary intake” OR “food intake” OR “caloric intake” OR “calorie intake” OR “nutritional quality” OR “nutritional intake” OR “food choice” OR “meal choice” OR “food selection” OR “food consumption” OR “meal selection” OR “meal consumption” OR “eating”)  Population terms  AND (“college student” OR “college students” OR “university student” OR “university students” OR “young adult” OR “young adults” OR “university” OR “college” OR “tertiary education”)  Exclusionary terms  NOT supplement NOT pharmacology NOT medication NOT allerg* NOT mice NOT rat NOT cigarette NOT choline NOT anemia NOT anorexia  Titles and abstracts of the articles identified through the keyword search were screened against the study selection criteria. Potentially relevant articles were retrieved for evaluation of the full text. A reference list search (ie, backward reference search) and cited reference search (ie, forward reference search) were conducted on the basis of the full-text articles that met the study selection criteria and were identified from the keyword search. Articles identified from the backward and forward reference search were further screened and evaluated using the same study selection criteria. The reference search was repeated on all newly identified articles until no additional relevant articles were found. Data extraction and synthesis A standardized data extraction form was used to collect the following methodological and outcome variables from each included study: author(s), publication year, study design, setting, sample size and demographics, response and/or completion rate, participant recruitment criteria, measures of nutrition label use and diet quality, main findings, and conclusions. Meta-analysis A meta-analysis was performed on studies that reported the mean number of calories consumed in the presence and absence of labels, standard deviations, and sample size for each group. When the number of calories selected or consumed was reported as an outcome but did not include all necessary information, authors were contacted. Effect size was calculated on the basis of the mean difference in calories selected or consumed between groups exposed and not exposed to nutrition labels. Pre–post studies (without a control group) and randomized controlled trials were analyzed in separate meta-analyses owing to differences in strength of the study design.19 Among studies that compared different label types, the simple textual labels were used when testing the effects of nutrition labels vs no labels. In addition, a meta-analysis was performed to test the effect of contextual labels vs simple textual labels. Four studies were included in the controlled experiment meta-analysis29,34–36, 6 in the pre–post meta-analysis,31,37–41 and 3 in the meta-analysis comparing contextual labels with simple textual labels.29,36,41 Study heterogeneity was assessed using the I2 index. The level of heterogeneity represented by I2 was interpreted as modest (I2 ≤ 25%), moderate (25% < I2 ≤ 50%), substantial (50% < I2 ≤ 75%), or considerable (I2 > 75%). Random-effects models were used for estimation since considerable heterogeneity was present. Publication bias was not assessed because of the variability in study designs and outcomes of interest. Meta-analysis was performed using Stata/IC software, version 13.1 (StataCorp, College Station, TX, USA). All analyses used 2-sided tests, and P < 0.05 was considered statistically significant. Study quality assessment The National Institutes of Health’s Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies was adapted to assess the quality of each included study.42 This assessment tool (Table 3) rates each study on the basis of 9 criteria. For each criterion, a score of 1 was assigned if “yes” was the response, whereas a score of 0 was assigned otherwise (ie, an answer of “no,” “not applicable,” “not reported,” or “cannot determine”). A study-specific global score, ranging from 0 to 9, was calculated by summing scores across all criteria. Criteria were as follows: (1) research question, study design, and data collection procedures were clearly documented; (2) sample size was sufficiently large to provide confidence in the findings; (3) reasons for selecting or recruiting the number of individuals were included, or statistical power was discussed; (4) there was a control (either a control group, control cafeteria, or a pre–post study in which participants served as their own controls) in the study; (5) either study participants or cafeterias were randomized; (6) dietary outcomes were observed rather than self-reported; (7) actual dietary intake (not simply selection) was assessed; (8) dietary outcome was assessed in a naturalistic eating setting; and (9) key potential confounding variables (eg, sex, body mass index) were measured and adjusted statistically for their effect on the relationship between nutrition label use and dietary intake. RESULTS Of the 798 unduplicated articles identified through the keyword and reference search, 722 were excluded by title and abstract screening (Figure 1). The remaining 76 articles were reviewed in full text, by which 61 studies were excluded because of the following reasons: age ineligibility (n = 31); inappropriate setting, such as a hospital or workplace cafeteria (n = 8); lack of quantitative assessment of food choices (n = 10); lack of assessment of label exposure in relation to diet, independent of other factors or interventions (n = 3); use of labels that were not nutrition labels (n = 2); manipulation of nutrition labels or priming of participants (n = 5); and assessment of hypothetical rather than actual dietary intake (n = 2). The remaining 15 articles were included in the review. An additional 6 articles were identified through reference search, resulting in a total of 21 articles (22 separate studies) included in the review. Table 3 Study quality assessment criteria, adapted from the US National Heart, Lung, and Blood Institute Quality Assessment Tool for pre–post studies and cross-sectional studies Item  Criterion of study quality  Mean score  1  Were the research question, the study design, and the data collection procedures clearly documented? (yes = 1, no = 0)a  1  2  Was the sample size sufficiently large to provide confidence in the findings?b (yes = 1, no = 0)  0.77  3  Were reasons for selecting or recruiting the number of individuals included, or was statistical power discussed? (yes = 1, no = 0)  0.14  4  Was there a control (a control group, a control cafeteria, or a pre–post study cohort study in which participants served as their controls)? (yes = 1, no = 0)  0.55  5  Were study participants or cafeterias randomized? (yes = 1, no = 0)  0.27  6  Were dietary outcome(s) observed, not self-reported? (yes = 1, no = 0)  0.86  7  Was actual intake, not just food selection, assessed? (yes = 1, no = 0)  0.41  8  Was dietary outcome assessed in a natural eating setting? (yes = 1, no = 0)  0.82  9  Were key potential confounding variables (eg, sex, body mass index) measured and adjusted statistically for their effect on the relationship between nutrition label use and dietary intake? (yes = 1, no = 0)  0.41    Overall study quality score  5.2  Item  Criterion of study quality  Mean score  1  Were the research question, the study design, and the data collection procedures clearly documented? (yes = 1, no = 0)a  1  2  Was the sample size sufficiently large to provide confidence in the findings?b (yes = 1, no = 0)  0.77  3  Were reasons for selecting or recruiting the number of individuals included, or was statistical power discussed? (yes = 1, no = 0)  0.14  4  Was there a control (a control group, a control cafeteria, or a pre–post study cohort study in which participants served as their controls)? (yes = 1, no = 0)  0.55  5  Were study participants or cafeterias randomized? (yes = 1, no = 0)  0.27  6  Were dietary outcome(s) observed, not self-reported? (yes = 1, no = 0)  0.86  7  Was actual intake, not just food selection, assessed? (yes = 1, no = 0)  0.41  8  Was dietary outcome assessed in a natural eating setting? (yes = 1, no = 0)  0.82  9  Were key potential confounding variables (eg, sex, body mass index) measured and adjusted statistically for their effect on the relationship between nutrition label use and dietary intake? (yes = 1, no = 0)  0.41    Overall study quality score  5.2  a At a minimum, recruitment, mode and setting for data collection, and study duration were indicated. b If power calculations were not detailed, sufficient size was at least 100 participants per group (ie, pre- and postintervention) or at least 100 observations per period (ie, sales pre- and postintervention). Figure 1 View largeDownload slide Flow diagram of the literature search process. Figure 1 View largeDownload slide Flow diagram of the literature search process. Basic study characteristics The basic characteristics of the included studies are reported in Table 4.29–41,43–50 Studies were conducted in 5 countries: the United States (n = 15), the United Kingdom (n = 3), Canada (n = 2), Belgium (n = 1), and Australia (n = 1). Of the 22 studies, 5 were experimental or quasi-experimental controlled trials wherein participants were exposed or not exposed to nutrition labels, and 17 were cohort or pre–post calorie-labeling interventions conducted in cafeterias or with vending machines. Thirteen studies measured calories selected or consumed, and 12 assessed dietary quality via noncaloric measures of specific food items or meals (some studies assessed multiple outcomes). Table 4 Study design, methods and characteristics of sample, dietary outcome assessed, and main findings of the studies included in the systematic review Reference  Design, setting, and location of study  Methods and sample  Dietary outcome  Main findings  Aaron et al. (1995)34  Quasiexperimental (pre–post + control), 2 cafeterias, Reading, UK  Duration: 2 wk, M–F lunch Week 1: No labels Week 2: Labels containing calories and percentage of calories from fat posted in cafeteria 1 Participants (N = 90: 65 intervention participants and 25 control participants) selected meals each day at lunch, recorded foods and portions selected and eaten, and gave investigators plates after eating. Intervention participants were surveyed on awareness, usage, and label understanding 1 wk after the study  Self-reported + objectively measured plate waste: energy and macronutrient intake/selection for 10 lunch meals  Most (92%) intervention participants noticed labels, but few used them (8% greatly or moderately, 18% slightly). Intervention participants ate more calories (927 ± 27) in week 2 than in week 1 (875 ± 23, P < 0.05), while selection among control participants did not differ (944 ± 51 vs. 906 ± 54 in wk 2, P > 0.05). Among intervention participants, unrestrained eaters and males consumed more calories and carbohydrates (grams) and fat (grams), and less protein (grams) and percent energy from protein, whereas Females and unrestrained eaters had similar intakes across both weeks  Chu et al. (2009)37  Quasiexperimental (pre–post), single group, interrupted time series, cafeteria, Columbus, OH, USA  Duration: 6 wk, 7 d/wk, all day Weeks 1–2: pretreatment, no labels Weeks 3–4: treatment, 3” × 5” label cards at point of selection included calories, serving size, and macronutrients for study entrées Weeks 5–6: post-treatment, no labels Sales data for 12 study entrées were collected during each period; N = 13 951 pretreatment, N = 14 199 treatment, N = 14 020 post-treatment  Sales data: daily average energy content of 12 entrées  Average entrée energy content of entrées decreased 12.0 kcal from the last day of baseline to the first day of treatment (P = 0.007). Energy content decreased over the treatment period (−0.3 kcal/d), but not significantly from pretreatment. After labels were removed, energy content increased at a rate of 1.5 kcal/d (P = 0.013) across post-treatment period  Cioffi et al. (2015)43  Quasiexperimental (pre–post), 20 retail dining units, Ithaca, NY, USA  Duration: 3 y (6 semesters) Semesters 1–3: baseline Semesters 4–6: nutrition labels with calories and nutrient composition posted Weekly sales data (including thousands of observations, but specific number not noted) for 45 prepackaged food items were collected from campus dining units. Items were categorized as high-calorie and low-calorie, and high-fat and low-fat  Sales data: weekly mean total calories and fat purchased per FreshTake item selected  Average energy and fat content of items purchased per week decreased 6.5% and 7.4%, respectively, after label introduction (both P < 0.001). Upon label introduction, sales of low-calorie foods increased from 2.9% to 3.2%, and sales of high-fat items decreased from 2.9% to 2.6% (both P < 0.001)  Cinciripini (1984)45  Quasiexperimental (pre–post), cafeteria, Galveston, TX, USA  Duration: ≈ 24–27 wk (3 periods, 8–9 wk each), lunch 3–5 d/wk Period 1: baseline 1 Period 2: 2 large signs were posted at cafeteria entrance listing items and caloric information; flyers describing the signs were distributed for 10 d Period 3: baseline 2 Unobtrusive observation of participants' body classification (lean, normal, obese) and food choices from a checklist of 97 food items offered while participants paid for lunch  Observed choice: nonstarchy vegetable/soup/fruit/low-fat dairy, high-fat, red meat, lean protein, regular dairy, starchy carbohydrates, salads  Frequency of choosing carbohydrates decreased for all participants from baseline to intervention, frequency of choosing red meat decreased for almost all groups, frequency of choosing regular dairy products decreased for normal-weight males and females, and frequency of choosing high-fat dessert/sauces decreased for 1 subgroup (all P < 0.05). Salad and vegetable/soup/fruit/low-fat dairy choices each increased for 1 subgroup. After label removal, frequency of choosing salad increased for 1 subgroup; decreases in frequency of choosing high-fat dessert/sauces persisted for 1 subgroup  Davis-Chervin et al. (1985)46  Quasiexperimental (pre–post + control), 2 cafeterias, Stanford, CA, USA  Duration: 1 y (3 trimesters), weekday lunch + dinner Trimester 1: 5-wk baseline, 5-wk intervention (cafeteria 1) with labels (4″×6″ 3-color cards with calories, percentage of calories from fat, milligrams of cholesterol), and nutrition education posters displayed Trimester 2: 2-wk baseline, 5-wk intervention (cafeteria 1), 3-wk baseline Trimester 3: 5-wk intervention (both cafeterias; in cafeteria 2 only, labels but no posters were displayed), 5-wk baseline Sales data were collected to identify the proportion of low-calorie, low-cholesterol, and low-fat entrées sold. Cafeteria 1 served 175–200 first-year students at each meal; cafeteria 2 served 450–500 undergraduates from all 4 classes  Sales data: proportion of entrées chosen with the lowest amount of cholesterol, fat, or calories of the entrées served on a given day  In cafeteria 1, selection of low-calorie entrées at lunch increased by 35% and 66% during intervention periods 1 and 2, respectively, while selection of low-cholesterol entrées at lunch or dinner increased by 28%–53% from baseline (all P < 0.05). Selection of low-cholesterol, low-fat, and low-calorie entrées increased as a proportion of total from baseline levels during each intervention period and during the final no-intervention phase. In cafeteria 2, nonsignificant increases in the selection of low-calorie and low-cholesterol entrées were observed from baseline to intervention  Dingman et al. (2015)35  Quasiexperimental (pre–post + control), residence hall vending machines, Greensboro, NC, USA  Duration: 8 wk 18 vending machines, containing 35–40 snacks each, produced usable data. At the start of week 5, posters listing Nutrition Facts for each snack and highlighting the 5 healthiest products with a Better Choice logo were posted beside the 9 intervention machines, with a note on the machine referencing the poster. Intervention hall residents were sent an email about Better Choice criteria. Students were emailed a survey upon study completion  Sales data: average calories per snack, proportion of Better Choice items sold  56% of residents in intervention halls (n = 364) noticed on-site nutrition information, but n = 192 (60% of those who answered both questions) reported it did not influence purchasing decisions. Neither the proportion of Better Choice snacks sold (6.2% before and 6.9% after in intervention machines; 8.2% before and 6.6% after in control machines) nor the average number of calories per snack (252 ± 24 before and 251 ± 21 after in intervention machines; 217 ± 55 before and 225 ± 56 after in control machines) differed between the intervention and control machines or after label introduction (both P > 0.05)  Ellison et al. (2014)29  Quasiexperimental comparative trial, full-service restaurant, Stillwater, OK, USA  Duration: 12 wk (out of a 19-wk intervention), lunch daily Diners at each table were given 1 of 3 menus: (1) conventional with food descriptions; (2) descriptions + calorie count; (3) descriptions + calorie count + traffic light symbol (red for entrées over 800 kcal, yellow for entrées 401–800 kcal, and green for those under 400 kcal). Receipts yielded 978 observations for the labeling portion of the study  Sales data: number of calories in main entrée  Labels resulted in more orders of low- and medium-calorie items (P < 0.01). Without labels, 30% of participants chose low-calorie, 36% medium-calorie, and 35% high-calorie entrées. With calorie-only labels, 32% chose low-calorie, 38% medium-calorie, and 30% high-calorie entrées. With traffic light labels, 39% chose low-calorie, 33% medium-calorie, and 28% high-calorie entrées. Calorie-only and traffic light labels reduced orders of high-calorie items by 4.4% and 6.4%, respectively. Those exposed to calorie-only and traffic light labels ordered 30 (3.9%) and 71 (9.4%) fewer kilocalories per meal, respectively, compared with those not exposed to labels. Using regression, traffic light labels predicted calories ordered (P < 0.01), while calorie-only labels did not  Freedman (2011)47  Quasiexperimental (pre–post), all-you-can-eat cafeteria, San Jose, CA, USA  Duration: 5 wk, MWF lunch Week 1: baseline (no labels) Weeks 2–5: intervention (point-of-service nutrition labels, 10″ × 7.5″ laminated color signs on sneeze guards, including portion sizes, photos, and slogans) Researchers unobtrusively observed students’ (N = 1675) choices of French fries (small portions ≤ 18 fries, large > 18 fries) and salad dressings in 1 cafeteria. Students (N = 377) were surveyed about label awareness and usage 1 wk after intervention  Observed choice: choices of salad dressing and French fries, portion size of fries  Of those who reported seeing nutrition information, one-third said it affected their choice (32%, N = 73) or portion size (38%, N = 84) of French fries, and one-fourth said it affected their choice (24%, N = 53) or portion size (26%, N = 58) of salad dressing. French fry selection did not change from baseline (24% of diners) to intervention (25%), but percentage of diners choosing large portion sizes decreased from 60% at baseline to 43% when labels were present (P < 0.05). The proportion of students selecting salad dressing at baseline (19%) and intervention (25%) remained similar, but the proportion selecting salad dressings with mid-range calories increased (P < 0.05), in part due to nonsignificant decreases in the proportion selecting the highest-calorie dressings  Freedman & Connors (2010)50  Quasiexperimental (pre–post), convenience store, San Jose, CA, USA  Duration: 11 wk, separated by several months Weeks 1–6: baseline, mid-fall 2008 Winter break 2008–2009: Eat Smart labels (1.25″ × 3″) and a poster were placed in the convenience store, and labels were placed directly beneath healthier food items Weeks 7–11: follow-up, mid-spring 2009 Sales data were collected for cereal, soup, cracker, and bread categories  Sales data: tagged items  Sales and the percentage of sales of tagged food items did not change significantly upon label introduction. Sales of tagged items increased by 3.6% in the intervention (P = 0.082), and the percentage of sales of tagged items in the cereal, soup, and cracker categories increased, while sales of tagged bread decreased (all P > 0.05)  Hammond et al. (2015)38  Quasiexperimental (pre–post cohort), cafeteria, Waterloo, ON, Canada  Duration: 2 wk, lunch and dinner, baseline and intervention separated by 6 wk so that menu offerings were the same Week 1: baseline (no labels) Week 7: posted labels included calories in red, food description in black, 24-point font Week 8: patrons (N = 159) were approached (using an intercept method) upon exiting and asked to complete a 10-min interviewer-administered survey on nutrition label use and food and beverage selections and intake that day  Self-reported (directly after meal) data: calories ordered and consumed  Reported nutrition label use increased from 9% at baseline to 29% at follow-up (P < 0.001). Calories ordered decreased by 11% (91 kcal, P = 0.013) and calories consumed by 13% (98 kcal, P = 0.006) from baseline to follow-up after adjustment for sex, BMI, race, general health, and weight aspirations and perceptions  Hoefkens et al. (2011)39  Quasiexperimental (pre–post cohort), cafeteria, Ghent, Belgium  Duration: 8 mo, 6 d per person, 3 d each at baseline and follow-up Months 1–2 (October/November 2008): baseline Month 6: labels posted in March included a star rating system, with meals given a star if they met recommended amounts for energy, saturated fat, sodium, and vegetables, for a total of 0–4 points and 0–3 stars. The 12 healthiest combinations, with ratings, were listed each day, and featured on posters and in the buffet line Months 7–8 (April/May 2009): follow-up A convenience sample (N = 224) completed 3-d food records and questionnaires at baseline and follow-up  Self-reported (concerning specific meals) data: energy intake from cafeteria meals (average of the 3 d from food records), food types, and macronutrients  Average calorie intake for the lunch cafeteria meal and over 24 h did not change between baseline and follow-up (P > 0.05). Participants consumed significantly more grams of vegetables at follow-up (both at the canteen meal and over 24 h, driven by the canteen meal) and fewer grams of carbohydrates over 24 h (P < 0.05). Protein, fat, percent energy from saturated fat, and sodium were similar across baseline and follow-up, as was the proportion of meals chosen in different star rating groups  Hoerr & Louden (1993)30  Quasiexperimental (pre–post), vending machines, East Lansing, MI, USA  Duration: 2 y, weeks 4–7 of each trimester Year 1: baseline Year 2: posted labels for each item included calories, protein, vitamin A, vitamin C, thiamin, riboflavin, niacin, calcium, and iron (in orange bar graphs) Sales of low-nutrient-density (chocolate candy bar, nuts, chocolate cookie), moderate-nutrient density (chocolate peanuts, granola bar, cheese popcorn), and high-nutrient-density (pretzels, peanut butter and crackers, peanuts) items were measured in 4 vending machines with 8 slots each in 4 academic buildings; N = 7174 in year 1 and N = 7742 in year 2  Sales data: proportion of low-, moderate-, and high-nutrient-density foods sold  Proportion of snacks sold in the low-, moderate-, and high-nutrient-density groups did not differ significantly between years 1 and 2  James et al. (2015)36  Quasiexperimental comparative trial, metabolic kitchen and graduate residence hall, Fort Worth, TX, USA  Duration: N/A Participants (N = 300) came to a laboratory kitchen (N = 278) or graduate residence hall (N = 22), had height and weight measured, and were seated alone, receiving 1 of 3 menus for a fast-food restaurant. Menus included the following: (1) no labels, (2) kilocalorie labels with a statement about daily caloric requirements, and (3) exercise labels showing minutes of brisk walking required to burn the energy from food items (specific to males and females). After ordering, participants were surveyed, and participants in labels groups were asked if they had noticed the labels. Food and beverages were unobtrusively weighed before and after the meal  Observed choice + objectively measured plate waste: calories ordered and eaten and calories from fat, protein, and carbohydrates  91% of those exposed noticed labels. The no-labels group ordered more calories than the exercise-labels group but not the kilocalorie-labels group (overall and exercise- vs no-labels groups both P<0.05). For consumption, the no-labels group again ate significantly more than the exercise-labels group but not the kcal labels (overall and exercise- vs no-labels groups both P<0.05). There were no differences for ordering and consumption between the exercise- and kcal-label groups and the kcal- and no-label groups (P>0.05). Exercise- and no-label groups also ordered and consumed differing percentages of calories from fat (both P<0.05), but not carbohydrates and protein  Larson-Brown (1978)49  Quasiexperimental (pre–post), vending machines, Provo, UT, USA  Duration: 2 mo Month 1: baseline Month 2: nutrition labels posted in front of items included calories and the percentage of US dietary recommended allowance for protein, calcium, thiamin, vitamin C, and iron (in colored bar graphs) Sales data were collected for vending machines in 2 adjoining campus buildings. Foods were categorized as more nutritious (milk, sandwiches, fruit, Welchade [Welch’s; Concord, MA, USA], yogurt, V-8 juice [Campbell Soup Co; Camden, NJ, USA], ice cream) or less nutritious (soft drinks, sweet rolls and brownies, gum and LifeSavers [Squibb Beech-Nut; New York, NY, USA], Hostess products [Continental Baking Co; New York, NY, USA], M&M’s [Mars; McLean, VA, USA] , Hershey’s chocolate [Hershey Foods Corp; Hershey, PA, USA] , candy, cookies). N = 26 558 sales in February and N = 30 371 in March  Sales data: more-nutritious and less-nutritious foods  Purchase of more-nutritious foods increased from 49.8% of total sales in February to 53.7% of total sales in March, a significant difference. For more-nutritious foods, sales of milk, sandwiches, fruits, Welchade, and yogurt increased, while sales of V-8 juice and ice cream decreased. For less-nutritious foods, sales of soft drinks increased, while sales of all others decreased (significance not noted)  Lillico et al. (2015)40  Quasiexperimental (pre–post), student residence cafeteria, Waterloo, ON, Canada  Duration: 2 wk, lunch and dinner, separated by 6 wk so that menu offerings were the same Week 1: baseline (no labels) Week 6: posted labels included calories and food description in 24-point font Week 7: students (N = 131 baseline, N = 168 follow-up) were approached (using an intercept method) upon exiting and asked to complete a 10-min interviewer-administered survey assessing food and beverage intake  Self-reported (directly after meal) data: calories consumed  Calorie consumption did not change significantly between baseline (661 ± 309 kcal) and follow-up (601 ± 282 kcal, P = 0.104) periods  Nikolaou et al. (2014)41  Quasiexperimental (pre–post, interrupted time series), residence hall cafeteria, Glasgow, Scotland, UK  Duration: three 14-d study periods, each separated by 4 wk, evening Weeks 1–2: no labels Weeks 3–4: calorie-only labels Weeks 5–6: calories + suggested daily intake labels The first 100 meal selections for 14-d periods within the 5-wk menu cycle were recorded, for a total of 4200 meals, including side dishes. Ingredient orders for evening meals placed by caterers were also recorded over the course of 2 y (2 mo each year)  Observed choice: calories, fat, saturated fat, vitamin C, iron, and calcium content of meal choices  Both males and females selected fewer calories when labels were present and even fewer when calories + suggested daily intake labels were present; selection during each period differed significantly from that during the other periods (P < 0.01). From period 1 (simple labels) to period 3 (contextual labels), mean calories per tray fell by 25% for females and 15% for males. Fat and saturated fat content of meals decreased after exposure to calorie labels + suggested daily intake; no differences were found in selection of vitamin C, iron, or calcium. Total calories ordered by caterers fell 18%, orders for ingredients used primarily for dessert preparation fell 60%, and oils used for frying fell 35% from years 1 to 2 when labels were present  Nikolaou et al. (2014)48  Quasiexperimental (pre–post + control), 3 food retail outlets, Glasgow, Scotland, UK  Duration: 2 mo Month 1: baseline Month 2: labels posted for the last 2 wk (first 2 wk were a university holiday, with reduced catering) Calorie labels (laminated 5.4 cm × 9.9 cm) containing the item name, calories, and the “Human Nutrition” department logo and university coat of arms were posted prominently in front of all sandwiches in 2 intervention food outlets; a control outlet did not have labels posted. Patrons were surveyed online and in outlets 1 wk into the intervention  Sales data of 19 sandwiches/rolls with a variety of fillings and caloric content  61% of female and 41% of male students reported that calorie information influenced choices. Between months 1 and 2, sales of all labeled items fell 17% in the intervention and 2% in the control outlets (P < 0.001). Sales of high-calorie (−30%) and low-calorie (−18%) items and high-fat (−21%) and low-fat (−23%) items decreased from months 1 to 2 in intervention outlets (P < 0.001), while sales of these items did not differ at the control outlet  Roy et al. (2016)32  Quasiexperimental (pre–post), quick-service food outlet, Sydney, Australia  Duration: 10 wk (5 wk each, 1 y apart) Sales data were collected at baseline for 5 wk. The next year, kilojoule content was posted on menus, including a reference statement listing an average adult serving of 8700 KJ. Students selected foods and beverages from a menu and then ordered food at a counter. Students (N = 318) were also surveyed during the intervention  Sales data of 9 items of varying caloric content  Only 5% of those surveyed reported being both aware of and influenced by labels. Compared with baseline, sales of a high-calorie entrée (grilled burger) decreased 35% while sales of 1 lower-calorie meal (chicken schnitzel and chips) increased 34% upon exposure to kilojoule labeling; sales of the remaining 7 items did not change  Schwartz et al. (2012)31  Quasiexperimental (pre–post), Chinese restaurant, Durham, NC, USA  Duration: two 3-wk periods (two 2-wk periods included in this analysis), M–Th, lunch Patrons ordered 1 of 4 side dishes (rice, fried rice, lo mein, or steamed vegetables) and then 1 of 16–20 stir fry entrées. After a 3-wk baseline and downsizing intervention with no labels present, there was a 2-wk break wherein calorie labels were posted on the sneeze shield above containers. A second 3-wk intervention was then conducted with labels and a downsizing intervention (not included in this review). Itemized receipts were collected to measure sales  Sales data: calories ordered  Customers ordered an average of 1020 ± 15 kcal when not exposed to labels and 1033 ± 16 kcal after label exposure, a nonsignificant difference (P >0 .05)  Temple et al. (2010)44  Randomized between-group experiment, laboratory, Buffalo, NY, USA  Duration: N/A Participants (N = 47) visited a lab at lunchtime and were randomly assigned to watch a movie on either the organic food movement or how to read nutrition labels. They then ate a buffet lunch of preweighed items either including or not including nutrition labels  Observed choice + plate waste: calories consumed, energy-dense foods consumed  Those exposed to labels ate fewer calories than those not exposed (P = 0.04). Energy density of chosen foods also differed; those not exposed to labels ate more of both high- and low-energy-density foods (both P < 0.05)  Temple et al. (2011)33 (A)  Within-subject experiment, laboratory, Buffalo, NY, USA  Duration: N/A Participants (N = 51) were surveyed and then ate a buffet lunch of preweighed items in the laboratory 3 times (≈ 1 h each time): once without labels present, once with standard labels present (4″ × 6″ labels resembling the manufacturer’s label), and once with traffic light labels present (in random order). Participants were given 25 min alone to eat  Observed choice + plate waste: calories consumed, proportion of green/yellow/red foods consumed  Label condition did not affect calories consumed, but there was a significant interaction between gender, labeling condition, and weight group for calories consumed. Lean females consumed fewer calories when standard or traffic labels were present (P < 0.05); all other groups consumed approximately the same number of calories in all 3 conditions (P > 0.05). All groups consumed more green foods in the presence of traffic light labels (P = 0.002)  Temple et al. (2011)33 (B)  Within-subject experiment, laboratory, Buffalo, NY, USA  Duration: N/A Participants (N = 41) were surveyed and then ate a buffet lunch of preweighed items in the laboratory 2 times (≈ 1 h each time): once without labels present, and once with traffic light labels present (in random order)  Observed choice + plate waste: calories consumed, green/yellow/red foods consumed  Participants exposed to traffic light labels were more likely to purchase green items (P = 0.01), but labeling did not affect energy intake. Intake of green foods increased and red foods decreased upon exposure to traffic light labels vs no labels (both P < 0.05)  Reference  Design, setting, and location of study  Methods and sample  Dietary outcome  Main findings  Aaron et al. (1995)34  Quasiexperimental (pre–post + control), 2 cafeterias, Reading, UK  Duration: 2 wk, M–F lunch Week 1: No labels Week 2: Labels containing calories and percentage of calories from fat posted in cafeteria 1 Participants (N = 90: 65 intervention participants and 25 control participants) selected meals each day at lunch, recorded foods and portions selected and eaten, and gave investigators plates after eating. Intervention participants were surveyed on awareness, usage, and label understanding 1 wk after the study  Self-reported + objectively measured plate waste: energy and macronutrient intake/selection for 10 lunch meals  Most (92%) intervention participants noticed labels, but few used them (8% greatly or moderately, 18% slightly). Intervention participants ate more calories (927 ± 27) in week 2 than in week 1 (875 ± 23, P < 0.05), while selection among control participants did not differ (944 ± 51 vs. 906 ± 54 in wk 2, P > 0.05). Among intervention participants, unrestrained eaters and males consumed more calories and carbohydrates (grams) and fat (grams), and less protein (grams) and percent energy from protein, whereas Females and unrestrained eaters had similar intakes across both weeks  Chu et al. (2009)37  Quasiexperimental (pre–post), single group, interrupted time series, cafeteria, Columbus, OH, USA  Duration: 6 wk, 7 d/wk, all day Weeks 1–2: pretreatment, no labels Weeks 3–4: treatment, 3” × 5” label cards at point of selection included calories, serving size, and macronutrients for study entrées Weeks 5–6: post-treatment, no labels Sales data for 12 study entrées were collected during each period; N = 13 951 pretreatment, N = 14 199 treatment, N = 14 020 post-treatment  Sales data: daily average energy content of 12 entrées  Average entrée energy content of entrées decreased 12.0 kcal from the last day of baseline to the first day of treatment (P = 0.007). Energy content decreased over the treatment period (−0.3 kcal/d), but not significantly from pretreatment. After labels were removed, energy content increased at a rate of 1.5 kcal/d (P = 0.013) across post-treatment period  Cioffi et al. (2015)43  Quasiexperimental (pre–post), 20 retail dining units, Ithaca, NY, USA  Duration: 3 y (6 semesters) Semesters 1–3: baseline Semesters 4–6: nutrition labels with calories and nutrient composition posted Weekly sales data (including thousands of observations, but specific number not noted) for 45 prepackaged food items were collected from campus dining units. Items were categorized as high-calorie and low-calorie, and high-fat and low-fat  Sales data: weekly mean total calories and fat purchased per FreshTake item selected  Average energy and fat content of items purchased per week decreased 6.5% and 7.4%, respectively, after label introduction (both P < 0.001). Upon label introduction, sales of low-calorie foods increased from 2.9% to 3.2%, and sales of high-fat items decreased from 2.9% to 2.6% (both P < 0.001)  Cinciripini (1984)45  Quasiexperimental (pre–post), cafeteria, Galveston, TX, USA  Duration: ≈ 24–27 wk (3 periods, 8–9 wk each), lunch 3–5 d/wk Period 1: baseline 1 Period 2: 2 large signs were posted at cafeteria entrance listing items and caloric information; flyers describing the signs were distributed for 10 d Period 3: baseline 2 Unobtrusive observation of participants' body classification (lean, normal, obese) and food choices from a checklist of 97 food items offered while participants paid for lunch  Observed choice: nonstarchy vegetable/soup/fruit/low-fat dairy, high-fat, red meat, lean protein, regular dairy, starchy carbohydrates, salads  Frequency of choosing carbohydrates decreased for all participants from baseline to intervention, frequency of choosing red meat decreased for almost all groups, frequency of choosing regular dairy products decreased for normal-weight males and females, and frequency of choosing high-fat dessert/sauces decreased for 1 subgroup (all P < 0.05). Salad and vegetable/soup/fruit/low-fat dairy choices each increased for 1 subgroup. After label removal, frequency of choosing salad increased for 1 subgroup; decreases in frequency of choosing high-fat dessert/sauces persisted for 1 subgroup  Davis-Chervin et al. (1985)46  Quasiexperimental (pre–post + control), 2 cafeterias, Stanford, CA, USA  Duration: 1 y (3 trimesters), weekday lunch + dinner Trimester 1: 5-wk baseline, 5-wk intervention (cafeteria 1) with labels (4″×6″ 3-color cards with calories, percentage of calories from fat, milligrams of cholesterol), and nutrition education posters displayed Trimester 2: 2-wk baseline, 5-wk intervention (cafeteria 1), 3-wk baseline Trimester 3: 5-wk intervention (both cafeterias; in cafeteria 2 only, labels but no posters were displayed), 5-wk baseline Sales data were collected to identify the proportion of low-calorie, low-cholesterol, and low-fat entrées sold. Cafeteria 1 served 175–200 first-year students at each meal; cafeteria 2 served 450–500 undergraduates from all 4 classes  Sales data: proportion of entrées chosen with the lowest amount of cholesterol, fat, or calories of the entrées served on a given day  In cafeteria 1, selection of low-calorie entrées at lunch increased by 35% and 66% during intervention periods 1 and 2, respectively, while selection of low-cholesterol entrées at lunch or dinner increased by 28%–53% from baseline (all P < 0.05). Selection of low-cholesterol, low-fat, and low-calorie entrées increased as a proportion of total from baseline levels during each intervention period and during the final no-intervention phase. In cafeteria 2, nonsignificant increases in the selection of low-calorie and low-cholesterol entrées were observed from baseline to intervention  Dingman et al. (2015)35  Quasiexperimental (pre–post + control), residence hall vending machines, Greensboro, NC, USA  Duration: 8 wk 18 vending machines, containing 35–40 snacks each, produced usable data. At the start of week 5, posters listing Nutrition Facts for each snack and highlighting the 5 healthiest products with a Better Choice logo were posted beside the 9 intervention machines, with a note on the machine referencing the poster. Intervention hall residents were sent an email about Better Choice criteria. Students were emailed a survey upon study completion  Sales data: average calories per snack, proportion of Better Choice items sold  56% of residents in intervention halls (n = 364) noticed on-site nutrition information, but n = 192 (60% of those who answered both questions) reported it did not influence purchasing decisions. Neither the proportion of Better Choice snacks sold (6.2% before and 6.9% after in intervention machines; 8.2% before and 6.6% after in control machines) nor the average number of calories per snack (252 ± 24 before and 251 ± 21 after in intervention machines; 217 ± 55 before and 225 ± 56 after in control machines) differed between the intervention and control machines or after label introduction (both P > 0.05)  Ellison et al. (2014)29  Quasiexperimental comparative trial, full-service restaurant, Stillwater, OK, USA  Duration: 12 wk (out of a 19-wk intervention), lunch daily Diners at each table were given 1 of 3 menus: (1) conventional with food descriptions; (2) descriptions + calorie count; (3) descriptions + calorie count + traffic light symbol (red for entrées over 800 kcal, yellow for entrées 401–800 kcal, and green for those under 400 kcal). Receipts yielded 978 observations for the labeling portion of the study  Sales data: number of calories in main entrée  Labels resulted in more orders of low- and medium-calorie items (P < 0.01). Without labels, 30% of participants chose low-calorie, 36% medium-calorie, and 35% high-calorie entrées. With calorie-only labels, 32% chose low-calorie, 38% medium-calorie, and 30% high-calorie entrées. With traffic light labels, 39% chose low-calorie, 33% medium-calorie, and 28% high-calorie entrées. Calorie-only and traffic light labels reduced orders of high-calorie items by 4.4% and 6.4%, respectively. Those exposed to calorie-only and traffic light labels ordered 30 (3.9%) and 71 (9.4%) fewer kilocalories per meal, respectively, compared with those not exposed to labels. Using regression, traffic light labels predicted calories ordered (P < 0.01), while calorie-only labels did not  Freedman (2011)47  Quasiexperimental (pre–post), all-you-can-eat cafeteria, San Jose, CA, USA  Duration: 5 wk, MWF lunch Week 1: baseline (no labels) Weeks 2–5: intervention (point-of-service nutrition labels, 10″ × 7.5″ laminated color signs on sneeze guards, including portion sizes, photos, and slogans) Researchers unobtrusively observed students’ (N = 1675) choices of French fries (small portions ≤ 18 fries, large > 18 fries) and salad dressings in 1 cafeteria. Students (N = 377) were surveyed about label awareness and usage 1 wk after intervention  Observed choice: choices of salad dressing and French fries, portion size of fries  Of those who reported seeing nutrition information, one-third said it affected their choice (32%, N = 73) or portion size (38%, N = 84) of French fries, and one-fourth said it affected their choice (24%, N = 53) or portion size (26%, N = 58) of salad dressing. French fry selection did not change from baseline (24% of diners) to intervention (25%), but percentage of diners choosing large portion sizes decreased from 60% at baseline to 43% when labels were present (P < 0.05). The proportion of students selecting salad dressing at baseline (19%) and intervention (25%) remained similar, but the proportion selecting salad dressings with mid-range calories increased (P < 0.05), in part due to nonsignificant decreases in the proportion selecting the highest-calorie dressings  Freedman & Connors (2010)50  Quasiexperimental (pre–post), convenience store, San Jose, CA, USA  Duration: 11 wk, separated by several months Weeks 1–6: baseline, mid-fall 2008 Winter break 2008–2009: Eat Smart labels (1.25″ × 3″) and a poster were placed in the convenience store, and labels were placed directly beneath healthier food items Weeks 7–11: follow-up, mid-spring 2009 Sales data were collected for cereal, soup, cracker, and bread categories  Sales data: tagged items  Sales and the percentage of sales of tagged food items did not change significantly upon label introduction. Sales of tagged items increased by 3.6% in the intervention (P = 0.082), and the percentage of sales of tagged items in the cereal, soup, and cracker categories increased, while sales of tagged bread decreased (all P > 0.05)  Hammond et al. (2015)38  Quasiexperimental (pre–post cohort), cafeteria, Waterloo, ON, Canada  Duration: 2 wk, lunch and dinner, baseline and intervention separated by 6 wk so that menu offerings were the same Week 1: baseline (no labels) Week 7: posted labels included calories in red, food description in black, 24-point font Week 8: patrons (N = 159) were approached (using an intercept method) upon exiting and asked to complete a 10-min interviewer-administered survey on nutrition label use and food and beverage selections and intake that day  Self-reported (directly after meal) data: calories ordered and consumed  Reported nutrition label use increased from 9% at baseline to 29% at follow-up (P < 0.001). Calories ordered decreased by 11% (91 kcal, P = 0.013) and calories consumed by 13% (98 kcal, P = 0.006) from baseline to follow-up after adjustment for sex, BMI, race, general health, and weight aspirations and perceptions  Hoefkens et al. (2011)39  Quasiexperimental (pre–post cohort), cafeteria, Ghent, Belgium  Duration: 8 mo, 6 d per person, 3 d each at baseline and follow-up Months 1–2 (October/November 2008): baseline Month 6: labels posted in March included a star rating system, with meals given a star if they met recommended amounts for energy, saturated fat, sodium, and vegetables, for a total of 0–4 points and 0–3 stars. The 12 healthiest combinations, with ratings, were listed each day, and featured on posters and in the buffet line Months 7–8 (April/May 2009): follow-up A convenience sample (N = 224) completed 3-d food records and questionnaires at baseline and follow-up  Self-reported (concerning specific meals) data: energy intake from cafeteria meals (average of the 3 d from food records), food types, and macronutrients  Average calorie intake for the lunch cafeteria meal and over 24 h did not change between baseline and follow-up (P > 0.05). Participants consumed significantly more grams of vegetables at follow-up (both at the canteen meal and over 24 h, driven by the canteen meal) and fewer grams of carbohydrates over 24 h (P < 0.05). Protein, fat, percent energy from saturated fat, and sodium were similar across baseline and follow-up, as was the proportion of meals chosen in different star rating groups  Hoerr & Louden (1993)30  Quasiexperimental (pre–post), vending machines, East Lansing, MI, USA  Duration: 2 y, weeks 4–7 of each trimester Year 1: baseline Year 2: posted labels for each item included calories, protein, vitamin A, vitamin C, thiamin, riboflavin, niacin, calcium, and iron (in orange bar graphs) Sales of low-nutrient-density (chocolate candy bar, nuts, chocolate cookie), moderate-nutrient density (chocolate peanuts, granola bar, cheese popcorn), and high-nutrient-density (pretzels, peanut butter and crackers, peanuts) items were measured in 4 vending machines with 8 slots each in 4 academic buildings; N = 7174 in year 1 and N = 7742 in year 2  Sales data: proportion of low-, moderate-, and high-nutrient-density foods sold  Proportion of snacks sold in the low-, moderate-, and high-nutrient-density groups did not differ significantly between years 1 and 2  James et al. (2015)36  Quasiexperimental comparative trial, metabolic kitchen and graduate residence hall, Fort Worth, TX, USA  Duration: N/A Participants (N = 300) came to a laboratory kitchen (N = 278) or graduate residence hall (N = 22), had height and weight measured, and were seated alone, receiving 1 of 3 menus for a fast-food restaurant. Menus included the following: (1) no labels, (2) kilocalorie labels with a statement about daily caloric requirements, and (3) exercise labels showing minutes of brisk walking required to burn the energy from food items (specific to males and females). After ordering, participants were surveyed, and participants in labels groups were asked if they had noticed the labels. Food and beverages were unobtrusively weighed before and after the meal  Observed choice + objectively measured plate waste: calories ordered and eaten and calories from fat, protein, and carbohydrates  91% of those exposed noticed labels. The no-labels group ordered more calories than the exercise-labels group but not the kilocalorie-labels group (overall and exercise- vs no-labels groups both P<0.05). For consumption, the no-labels group again ate significantly more than the exercise-labels group but not the kcal labels (overall and exercise- vs no-labels groups both P<0.05). There were no differences for ordering and consumption between the exercise- and kcal-label groups and the kcal- and no-label groups (P>0.05). Exercise- and no-label groups also ordered and consumed differing percentages of calories from fat (both P<0.05), but not carbohydrates and protein  Larson-Brown (1978)49  Quasiexperimental (pre–post), vending machines, Provo, UT, USA  Duration: 2 mo Month 1: baseline Month 2: nutrition labels posted in front of items included calories and the percentage of US dietary recommended allowance for protein, calcium, thiamin, vitamin C, and iron (in colored bar graphs) Sales data were collected for vending machines in 2 adjoining campus buildings. Foods were categorized as more nutritious (milk, sandwiches, fruit, Welchade [Welch’s; Concord, MA, USA], yogurt, V-8 juice [Campbell Soup Co; Camden, NJ, USA], ice cream) or less nutritious (soft drinks, sweet rolls and brownies, gum and LifeSavers [Squibb Beech-Nut; New York, NY, USA], Hostess products [Continental Baking Co; New York, NY, USA], M&M’s [Mars; McLean, VA, USA] , Hershey’s chocolate [Hershey Foods Corp; Hershey, PA, USA] , candy, cookies). N = 26 558 sales in February and N = 30 371 in March  Sales data: more-nutritious and less-nutritious foods  Purchase of more-nutritious foods increased from 49.8% of total sales in February to 53.7% of total sales in March, a significant difference. For more-nutritious foods, sales of milk, sandwiches, fruits, Welchade, and yogurt increased, while sales of V-8 juice and ice cream decreased. For less-nutritious foods, sales of soft drinks increased, while sales of all others decreased (significance not noted)  Lillico et al. (2015)40  Quasiexperimental (pre–post), student residence cafeteria, Waterloo, ON, Canada  Duration: 2 wk, lunch and dinner, separated by 6 wk so that menu offerings were the same Week 1: baseline (no labels) Week 6: posted labels included calories and food description in 24-point font Week 7: students (N = 131 baseline, N = 168 follow-up) were approached (using an intercept method) upon exiting and asked to complete a 10-min interviewer-administered survey assessing food and beverage intake  Self-reported (directly after meal) data: calories consumed  Calorie consumption did not change significantly between baseline (661 ± 309 kcal) and follow-up (601 ± 282 kcal, P = 0.104) periods  Nikolaou et al. (2014)41  Quasiexperimental (pre–post, interrupted time series), residence hall cafeteria, Glasgow, Scotland, UK  Duration: three 14-d study periods, each separated by 4 wk, evening Weeks 1–2: no labels Weeks 3–4: calorie-only labels Weeks 5–6: calories + suggested daily intake labels The first 100 meal selections for 14-d periods within the 5-wk menu cycle were recorded, for a total of 4200 meals, including side dishes. Ingredient orders for evening meals placed by caterers were also recorded over the course of 2 y (2 mo each year)  Observed choice: calories, fat, saturated fat, vitamin C, iron, and calcium content of meal choices  Both males and females selected fewer calories when labels were present and even fewer when calories + suggested daily intake labels were present; selection during each period differed significantly from that during the other periods (P < 0.01). From period 1 (simple labels) to period 3 (contextual labels), mean calories per tray fell by 25% for females and 15% for males. Fat and saturated fat content of meals decreased after exposure to calorie labels + suggested daily intake; no differences were found in selection of vitamin C, iron, or calcium. Total calories ordered by caterers fell 18%, orders for ingredients used primarily for dessert preparation fell 60%, and oils used for frying fell 35% from years 1 to 2 when labels were present  Nikolaou et al. (2014)48  Quasiexperimental (pre–post + control), 3 food retail outlets, Glasgow, Scotland, UK  Duration: 2 mo Month 1: baseline Month 2: labels posted for the last 2 wk (first 2 wk were a university holiday, with reduced catering) Calorie labels (laminated 5.4 cm × 9.9 cm) containing the item name, calories, and the “Human Nutrition” department logo and university coat of arms were posted prominently in front of all sandwiches in 2 intervention food outlets; a control outlet did not have labels posted. Patrons were surveyed online and in outlets 1 wk into the intervention  Sales data of 19 sandwiches/rolls with a variety of fillings and caloric content  61% of female and 41% of male students reported that calorie information influenced choices. Between months 1 and 2, sales of all labeled items fell 17% in the intervention and 2% in the control outlets (P < 0.001). Sales of high-calorie (−30%) and low-calorie (−18%) items and high-fat (−21%) and low-fat (−23%) items decreased from months 1 to 2 in intervention outlets (P < 0.001), while sales of these items did not differ at the control outlet  Roy et al. (2016)32  Quasiexperimental (pre–post), quick-service food outlet, Sydney, Australia  Duration: 10 wk (5 wk each, 1 y apart) Sales data were collected at baseline for 5 wk. The next year, kilojoule content was posted on menus, including a reference statement listing an average adult serving of 8700 KJ. Students selected foods and beverages from a menu and then ordered food at a counter. Students (N = 318) were also surveyed during the intervention  Sales data of 9 items of varying caloric content  Only 5% of those surveyed reported being both aware of and influenced by labels. Compared with baseline, sales of a high-calorie entrée (grilled burger) decreased 35% while sales of 1 lower-calorie meal (chicken schnitzel and chips) increased 34% upon exposure to kilojoule labeling; sales of the remaining 7 items did not change  Schwartz et al. (2012)31  Quasiexperimental (pre–post), Chinese restaurant, Durham, NC, USA  Duration: two 3-wk periods (two 2-wk periods included in this analysis), M–Th, lunch Patrons ordered 1 of 4 side dishes (rice, fried rice, lo mein, or steamed vegetables) and then 1 of 16–20 stir fry entrées. After a 3-wk baseline and downsizing intervention with no labels present, there was a 2-wk break wherein calorie labels were posted on the sneeze shield above containers. A second 3-wk intervention was then conducted with labels and a downsizing intervention (not included in this review). Itemized receipts were collected to measure sales  Sales data: calories ordered  Customers ordered an average of 1020 ± 15 kcal when not exposed to labels and 1033 ± 16 kcal after label exposure, a nonsignificant difference (P >0 .05)  Temple et al. (2010)44  Randomized between-group experiment, laboratory, Buffalo, NY, USA  Duration: N/A Participants (N = 47) visited a lab at lunchtime and were randomly assigned to watch a movie on either the organic food movement or how to read nutrition labels. They then ate a buffet lunch of preweighed items either including or not including nutrition labels  Observed choice + plate waste: calories consumed, energy-dense foods consumed  Those exposed to labels ate fewer calories than those not exposed (P = 0.04). Energy density of chosen foods also differed; those not exposed to labels ate more of both high- and low-energy-density foods (both P < 0.05)  Temple et al. (2011)33 (A)  Within-subject experiment, laboratory, Buffalo, NY, USA  Duration: N/A Participants (N = 51) were surveyed and then ate a buffet lunch of preweighed items in the laboratory 3 times (≈ 1 h each time): once without labels present, once with standard labels present (4″ × 6″ labels resembling the manufacturer’s label), and once with traffic light labels present (in random order). Participants were given 25 min alone to eat  Observed choice + plate waste: calories consumed, proportion of green/yellow/red foods consumed  Label condition did not affect calories consumed, but there was a significant interaction between gender, labeling condition, and weight group for calories consumed. Lean females consumed fewer calories when standard or traffic labels were present (P < 0.05); all other groups consumed approximately the same number of calories in all 3 conditions (P > 0.05). All groups consumed more green foods in the presence of traffic light labels (P = 0.002)  Temple et al. (2011)33 (B)  Within-subject experiment, laboratory, Buffalo, NY, USA  Duration: N/A Participants (N = 41) were surveyed and then ate a buffet lunch of preweighed items in the laboratory 2 times (≈ 1 h each time): once without labels present, and once with traffic light labels present (in random order)  Observed choice + plate waste: calories consumed, green/yellow/red foods consumed  Participants exposed to traffic light labels were more likely to purchase green items (P = 0.01), but labeling did not affect energy intake. Intake of green foods increased and red foods decreased upon exposure to traffic light labels vs no labels (both P < 0.05)  Abbreviations: BMI, body mass index; M−F, Monday through Friday; MWF, Monday, Wednesday, Friday; N/A, not available. College students included in the study samples were aged 19 to 29.9 years, and 34% to 75% were female (except for 1 study that exclusively recruited females). The majority of the samples were normal weight, though 1 study reported an average body mass index considered overweight (25.9), and another purposefully recruited participants with a weight status classified as obese for about half of the sample. Dietary outcomes: calories selected or consumed Dietary outcomes and main findings are reported in Table 4. Eight of the 13 studies investigating caloric selection or intake as an outcome found positive effects of posting labels; of the 5 remaining studies, 1 found a negative effect of introducing labels and 4 found no effect on energy selection and/or intake. Of the studies showing a statistically significant positive effect, Chu et al.37 in a cafeteria and Cioffi et al43 in retail dining units found that the average calorie content of items sold decreased after nutrition labels were introduced. In cafeterias, Hammond et al.38 reported that students ordered and consumed fewer calories after labels were posted, while Nikolaou et al.41 found that calories selected decreased upon posting of both simple nutrition labels and labels that included suggested daily intake. In an on-campus restaurant, Ellison et al.29 reported that exposure to traffic light or numeric menu labels resulted in decreased calories ordered, with traffic light labels being especially effective. In laboratory settings, James et al.36 found that exposure to exercise labels specifying how much physical activity was needed to burn the calories in a food item effectively decreased calories ordered and consumed, and Temple et al.33,44 found in 2 separate studies that label exposure decreased calories consumed in a buffet lunch among 1 subgroup and overall. In contrast, Aaron et al.34 found a negative effect of labeling: students eating in an intervention cafeteria consumed more calories after label introduction and more calories than the control group. Of the studies showing no effect, Dingman et al.35 reported that posting labels on vending machines did not affect the average number of calories sold. Hoefkens et al.39 and Lillico et al.40 in campus cafeterias and Schwartz et al.31 in a quick-service restaurant found that posting nutrition labels did not affect calorie intake. Dietary outcomes: noncaloric measures Five studies34,36,39,41,43 assessed the relationship between label exposure and macronutrient selection or intake, summarized in Table 4. Of the 12 studies reporting noncaloric and nonmacronutrient outcomes (eg, portion size, proportion of low-energy-density foods chosen), 9 reported that nutrition labels had some statistically significant positive effect, although a few identified an effect only in a subsample of participants. In pre–post studies in cafeterias where labels were implemented, Cinciripini45 found improved food group selections; Davis-Chervin et al.46 reported higher selection of low-calorie and low-cholesterol entrées in a cafeteria where posters were also displayed; Freedman47 found that portion sizes for fries decreased and salad dressing choices changed; and Hoefkens et al.39 reported higher vegetable consumption. In quick-service outlets, Nikolaou et al.48 found that posting labels resulted in decreased sales of both low-fat and high-fat items as well as decreases in sales of low-calorie items and much larger decreases in sales of high-calorie items; Roy et al32 found that sales of a high-calorie entrée decreased and sales of a lower-calorie meal increased. In a vending machine study, Larson-Brown49 found that sales of more-nutritious foods increased after labels were posted. In a laboratory, Temple et al.33,44 found that participants exposed to calorie labels ate fewer high-energy-density and low-energy-density foods and that participants exposed to traffic light labels were more likely to purchase green (or healthier) items. Three studies that examined noncaloric outcomes showed no statistically significant differences after label introduction. In a convenience store, Freedman and Connors50 reported a small increase in the percentage of tagged healthy items sold. In vending machine studies, Dingman et al.35 found that the proportion of “Better Choice” snacks purchased remained similar, and Hoerr and Louden30 found that the proportion of snacks in low-, moderate-, and high-nutrient-density groups did not differ upon label implementation. Meta-analysis Figure 229,31,34–41 shows the forest plots for the meta-analyses conducted on the 4 controlled experimental studies29,34–36 and the 6 pre–post studies.31,37–41 Meta-analyses results are shown in Table 5.29,31,34–41 Among controlled studies, exposure to simple textual nutrition labels was not associated with change in calories ordered or consumed (P = 0.4). Among pre–post studies, posting nutrition labels was associated with a decrease in the number of calories ordered or consumed by 36.0 kcal (95%CI, −60.2 to −11.8 kcal). Contextual labels (eg, traffic light, exercise equivalents, or list of daily suggested requirements) were more effective than simple calorie labels29,36,41 at improving dietary intake in all but 1 study.33 All 3 studies29,36,41 pooled for meta-analysis comparing textual vs contextual labels found contextual labels to be more effective, leading to a pooled estimated reduction of calories selected or consumed by 66.9 kcal (95%CI, −86.7 to −47.2 kcal). Table 5 Modeling results from the meta-analysis Reference  Study design  Experimental group kcal ± SD (group n)  Control group kcal ± SD (group n)  I2 index (%)  Pooled effect size (95%CI)  Model used  P value for dose–response effect from meta-regression  Aaron et al. (1995)34  Controlled  927 ± 27 (65)  906 ± 54 (25)  98.2  β = −20.8 (−69.3, 27.7)  Random-effects  0.374  Dingman et al. (2015)35  251 ± 21 (6170)  225 ± 56 (5538)  Ellison et al. (2014)29  724 ± 333 (312)  754 ± 339 (311)  James et al. (2015)36: selection  827 ± 61 (99)  902 ± 62 (99)  James et al. (2015)36: intake  722 ± 54 (99)  770 ± 53 (99)      Postlabeling  Prelabeling          Chu et al. (2009)37  Pre–post  635 ± 152 (14199)  648 ± 152 (13951)  98.6  β = −36.0 (−60.2, −11.8)  Random-effects  0.038  Hammond et al. (2015)38: selection  734 ± 331 (156)  825 ± 336 (149)  Hammond et al. (2015)38: intake  671 ± 327 (156)  769 ± 342 (149)  Hoefkens et al. (2011)39  598 ± 98 (224)  597 ± 114 (224)  Lillico et al. (2015)40  601 ± 282 (168)  661 ± 309 (131)  Nikolaou et al. (2014)41: men  692 ± 105 (507)  734 ± 101 (499)  Nikolaou et al. (2014)41: women  628 ± 105 (893)  709 ± 101 (901)  Schwartz et al. (2009)31  1033 ± 16 (294)  1020 ± 15 (299)      Contextual labels  Simple labels          Ellison et al. (2014)29  Contextual labels vs simple labels  683 ± 318 (355)  724 ± 333 (312)  86.4  β = −66.9 (−86.7, −47.2)  Random-effects  0.002  James et al. (2015)36: selection  763 ± 61 (102)  827 ± 61 (99)  James et al. (2015)36: intake  673 ± 53 (102)  722 ± 54 (99)  Nikolaou et al. (2014)41: men  534 ± 116 (952)  628 ± 105 (893)  Nikolaou et al. (2014)41: women  622 ± 116 (448)  692 ± 105 (507)  Reference  Study design  Experimental group kcal ± SD (group n)  Control group kcal ± SD (group n)  I2 index (%)  Pooled effect size (95%CI)  Model used  P value for dose–response effect from meta-regression  Aaron et al. (1995)34  Controlled  927 ± 27 (65)  906 ± 54 (25)  98.2  β = −20.8 (−69.3, 27.7)  Random-effects  0.374  Dingman et al. (2015)35  251 ± 21 (6170)  225 ± 56 (5538)  Ellison et al. (2014)29  724 ± 333 (312)  754 ± 339 (311)  James et al. (2015)36: selection  827 ± 61 (99)  902 ± 62 (99)  James et al. (2015)36: intake  722 ± 54 (99)  770 ± 53 (99)      Postlabeling  Prelabeling          Chu et al. (2009)37  Pre–post  635 ± 152 (14199)  648 ± 152 (13951)  98.6  β = −36.0 (−60.2, −11.8)  Random-effects  0.038  Hammond et al. (2015)38: selection  734 ± 331 (156)  825 ± 336 (149)  Hammond et al. (2015)38: intake  671 ± 327 (156)  769 ± 342 (149)  Hoefkens et al. (2011)39  598 ± 98 (224)  597 ± 114 (224)  Lillico et al. (2015)40  601 ± 282 (168)  661 ± 309 (131)  Nikolaou et al. (2014)41: men  692 ± 105 (507)  734 ± 101 (499)  Nikolaou et al. (2014)41: women  628 ± 105 (893)  709 ± 101 (901)  Schwartz et al. (2009)31  1033 ± 16 (294)  1020 ± 15 (299)      Contextual labels  Simple labels          Ellison et al. (2014)29  Contextual labels vs simple labels  683 ± 318 (355)  724 ± 333 (312)  86.4  β = −66.9 (−86.7, −47.2)  Random-effects  0.002  James et al. (2015)36: selection  763 ± 61 (102)  827 ± 61 (99)  James et al. (2015)36: intake  673 ± 53 (102)  722 ± 54 (99)  Nikolaou et al. (2014)41: men  534 ± 116 (952)  628 ± 105 (893)  Nikolaou et al. (2014)41: women  622 ± 116 (448)  692 ± 105 (507)  Figure 2 View largeDownload slide Forest plots showing calories selected or consumed by labeling condition for controlled studies (top) and by pre–post studies (bottom) reporting calories as an outcome.Note: Axes have different scales. Abbreviation: WMD, weighted mean difference. Figure 2 View largeDownload slide Forest plots showing calories selected or consumed by labeling condition for controlled studies (top) and by pre–post studies (bottom) reporting calories as an outcome.Note: Axes have different scales. Abbreviation: WMD, weighted mean difference. Study quality Studies included in the review on average scored 5.2 out of 9 points (Table 629–41,43–50). Studies scored particularly low (an average of 0.14 out of 1 point) on including a power analysis or reasoning for the sample size, and on randomization of participants or study setting (0.27 out of 1 point). Studies scored substantially higher on objectively observing dietary intake (0.86 out of 1 point) rather than relying on surveys, and on performing the study in a natural setting (0.82 out of 1 point). Of the 9 studies with the highest quality, 7 indicated that label exposure was related to better dietary intake for at least some groups. Of the 6 studies with the lowest quality, 5 showed some dietary improvement upon label exposure. Table 6 Scoring of study quality. Points awarded on the basis of criteria met for each study included in the review Reference  Documented procedures  Sample size  Reasoning or power analysis  Control group or cafeteria  Randomized participants or settings  Observed dietary outcome  Assessed intake  Natural setting  Accounted for potential confounders  Total points  Aaron et al. (1995)34  1  0  0  1  0  1  1  1  0  5  Chu et al. (2009)37  1  1  0  0  0  1  0  1  0  4  Cioffi et al. (2015)43  1  1  0  0  0  1  0  1  0  4  Cinciripini (1984)45  1  1  0  0  0  1  0  1  1  5  Davis-Chervin et al. (1985)46  1  1  0  1  0  1  0  1  0  5  Dingman et al. (2015)35  1  1  0  1  1  1  0  1  1  7  Ellison et al. (2014)29  1  1  0  1  1  1  0  1  0  6  Freedman (2011)47  1  1  0  0  0  1  0  1  0  4  Freedman et al. (2010)50,a  1  0  0  0  0  1  0  1  0  3  Hammond et al. (2015)38,b  1  1  0  1  0  0  1  1  1  6  Hoefkens et al. (2011)39,b  1  1  1  1  0  0  1  1  0  6  Hoerr & Louden (1993)30  1  1  1  0  0  1  0  1  0  5  James et al. (2015)36  1  1  1  1  1  1  1  0  1  8  Larson-Brown (1978)49  1  1  0  0  0  1  0  1  0  4  Lillico et al. (2015)40,b  1  1  0  1  0  0  1  1  1  6  Nikolaou et al. (2014)A41  1  1  0  0  0  1  0  1  1  5  Nikolaou et al. (2014)B48  1  1  0  1  0  1  0  1  0  5  Roy et al. (2016)32  1  1  0  0  0  1  0  1  0  4  Schwartz et al. (2012)31,c  1  1  0  0  0  1  1  1  0  5  Temple et al. (2010)44  1  0  0  1  1  1  1  0  1  6  Temple et al. (2011)A33  1  0  0  1  1  1  1  0  1  6  Temple et al. (2011)B33  1  0  0  1  1  1  1  0  1  6  Average of all studies  1  0.77  0.14  0.55  0.27  0.86  0.41  0.82  0.41  5.2  Reference  Documented procedures  Sample size  Reasoning or power analysis  Control group or cafeteria  Randomized participants or settings  Observed dietary outcome  Assessed intake  Natural setting  Accounted for potential confounders  Total points  Aaron et al. (1995)34  1  0  0  1  0  1  1  1  0  5  Chu et al. (2009)37  1  1  0  0  0  1  0  1  0  4  Cioffi et al. (2015)43  1  1  0  0  0  1  0  1  0  4  Cinciripini (1984)45  1  1  0  0  0  1  0  1  1  5  Davis-Chervin et al. (1985)46  1  1  0  1  0  1  0  1  0  5  Dingman et al. (2015)35  1  1  0  1  1  1  0  1  1  7  Ellison et al. (2014)29  1  1  0  1  1  1  0  1  0  6  Freedman (2011)47  1  1  0  0  0  1  0  1  0  4  Freedman et al. (2010)50,a  1  0  0  0  0  1  0  1  0  3  Hammond et al. (2015)38,b  1  1  0  1  0  0  1  1  1  6  Hoefkens et al. (2011)39,b  1  1  1  1  0  0  1  1  0  6  Hoerr & Louden (1993)30  1  1  1  0  0  1  0  1  0  5  James et al. (2015)36  1  1  1  1  1  1  1  0  1  8  Larson-Brown (1978)49  1  1  0  0  0  1  0  1  0  4  Lillico et al. (2015)40,b  1  1  0  1  0  0  1  1  1  6  Nikolaou et al. (2014)A41  1  1  0  0  0  1  0  1  1  5  Nikolaou et al. (2014)B48  1  1  0  1  0  1  0  1  0  5  Roy et al. (2016)32  1  1  0  0  0  1  0  1  0  4  Schwartz et al. (2012)31,c  1  1  0  0  0  1  1  1  0  5  Temple et al. (2010)44  1  0  0  1  1  1  1  0  1  6  Temple et al. (2011)A33  1  0  0  1  1  1  1  0  1  6  Temple et al. (2011)B33  1  0  0  1  1  1  1  0  1  6  Average of all studies  1  0.77  0.14  0.55  0.27  0.86  0.41  0.82  0.41  5.2  a Number of observations not reported. b Studies utilizing pre–post cohort designs were considered as having controls, though participants served as their own controls. c While the Schwartz et al. (2012)31 study assessed intake in another experiment in the same article, the experiment included in this review did not assess intake. DISCUSSION The present systematic review examined the effect of nutrition label use on diet among college students. Overall, 16 of the 22 studies included in the review reported that exposure to nutrition labels led to improved dietary choices. Eight of the 13 studies involving caloric outcomes found that posting nutrition labels at the point of purchase decreased calorie selection or consumption. Nine of the 12 studies measuring noncaloric measures of dietary quality such as food group choices found that introducing labels improved dietary quality. In the 10 studies pooled for meta-analysis, controlled experimental studies showed a nonsignificant decrease in calories selected and/or consumed in the presence of labels, whereas pre–post studies showed a significant decrease of 36 kcal in calories selected and/or consumed in the presence of labels. Studies of both relatively low and relatively high quality produced similar results, with a majority of both higher-quality and lower-quality studies showing that nutrition label exposure improved dietary intake in at least some groups. Setting might be crucial to measuring the effectiveness of nutrition labels. Of the 12 studies conducted in college cafeterias in the present review, 10 found positive effects,32,37–41,43,45–48 1 found no effect,40 and 1 found a negative effect34 of nutrition labels. Studies in laboratories generally showed positive effects, those in quick-service outlets and convenience stores were more mixed but overall positive, and those in vending machines showed few effects. The magnitude of effects was often small, and even those studies reporting overall significant effects had subgroups that were not always affected by the intervention. Prior systematic reviews in general populations have also shown that setting is crucial: Long et al.19 reported that labels significantly decreased calories ordered in nonrestaurant but not restaurant settings, and Fernandes et al.7 found menu labeling was more effective in cafeterias than in restaurants. In the latter study, the authors hypothesized that this effect could be related to educational level and the daily nature of cafeteria usage contrasted with the special-occasion nature of restaurant visits.7 The potential interaction between setting, nutrition label use, and dietary outcomes should be investigated, particularly for the effect on daily food patterns and noncaloric outcomes. In addition, it is important to consider barriers such as hunger and food cost51 as well as how nutrition labels may act together with other nutrition interventions in college settings, including price incentives, changes in food offerings, and control of portion size.25 In the present review, when compared with simple calorie labels, contextual labels (traffic light, exercise, and those containing daily intake recommendations) tended to be more effective at improving dietary intake, resulting in an average of 67 fewer calories ordered or consumed. Prior reviews and studies in adults have reported that labels containing contextual information are better understood16 and more effective at reducing intake of calories,15,17,52 total fat, saturated fat, and sodium52 and at improving food choices.21 Thus, the results of both the present review and prior studies indicate that contextual or interpretive labels such as traffic light labels or exercise equivalents are more effective at improving dietary intake. Contextual labels that include several components, such as the star ratings employed by Hoefkens et al.,39 may also have an added benefit of providing a more holistic approach to dietary quality that does not focus solely on calories. The wide variety of dietary outcomes assessed in the reviewed studies included calories, macronutrients, micronutrients, sales of items deemed healthful using different standards, cholesterol, energy density, and food groups. This array of outcomes was crucial for obtaining a comprehensive view of the effect of labels on overall dietary quality. However, this also meant that study results were difficult to pool, and estimates of the overall effect of labels were nontrivial. Future research should investigate the effect of nutrition labels on comprehensive measures of dietary quality rather than on calories alone. To ensure results can be compared and pooled across studies, standardized measures should be used, such as the Healthy Eating Index,53 the Mediterranean Diet Score,54 dietary quality indices,55,56 or nutrients present on nutrition labels.57 The results of this review support menu-labeling policies such as the US Food and Drug Administration’s menu-labeling rule, which will require restaurants and food retail establishments with 20 or more locations to post calorie labels on menus starting in May 2018.58 Posting point-of-purchase information is critical for providing consumers adequate information to make dietary choices, and the results of this review suggest that, among college students eating on campus, exposure to nutrition labels is likely to improve dietary intake. However, in the United States, college and university cafeterias and restaurants will not be required to post menu labels unless they have 20 or more locations offering similar products. Educational institutions should consider proactively implementing nutrition labeling, especially using interpretive labels to help students compare dietary options quickly and easily. In addition to potentially improving dietary intake, menu labeling may also decrease food costs; Nikolaou et al.41 found that, compared with data from the prior year, catering orders for overall calories, ingredients used primarily for desserts, and oils for frying all decreased substantially when labels were posted. Another potential benefit to nutrition labeling is that it may encourage product and recipe reformulation,59 which could improve dietary intake, even for consumers who do not consciously use nutrition labels. A few limitations of the review and included studies should be noted. The wide variety of study designs, outcomes, and even label presentation formats limits the ability to pool results. While it was crucial to aggregate studies separately on the basis of study design and label type, this meant that 2 of the meta-analyses included fewer than 5 studies, which could limit the generalizability, as meta-analyses are stronger when they encompass a larger number of studies with similar designs. Only 9 of the 22 included studies had comparison groups where participants did not serve as their own controls, although 17 of the 22 studies included a pre–post comparison. The need for studies with comparison groups has been highlighted in a previous review, which reported that nutrition-label interventions in real-world environments with comparison groups did not produce a significant decrease in calories ordered.6 Additionally, a few studies included only subgroup effects rather than overall effects, 2 studies38,40 may have had sample overlap, and several studies included relatively small samples for testing a population-level intervention. Lastly, title and abstract reviews were conducted by only 1 investigator. One strength of this review compared with other recent reviews6,15,60 is that, by limiting the population to college students, multiple outcomes of dietary quality beyond calories selected or consumed were assessed. This is an important distinction, as some data suggest that, while nutrition label users may eat similarly to nonusers in terms of food amount, there are meaningful differences in the foods selected.61 Thus, this review is able to comment on overall dietary quality, which has been shown to relate to long-term health outcomes.62 In addition, this review compared the relationship between nutrition label use and dietary quality across different settings, thus showing that the majority of studies in some settings (eg, cafeterias and laboratories) showed a positive effect, whereas studies in other settings (eg, vending machines) largely showed mixed results or no effect on dietary quality. Finally, this review considered study design within the meta-analysis, which is important for reviewing studies with vastly different designs. CONCLUSION The present systematic review and meta-analysis examined the effect of nutrition labels on diet among college students. Among the 22 studies included in the review, nutrition labels were found to have a moderate but significant positive effect on dietary choices in college students. These effects were modified by individual sociodemographics, setting, and type of labels used. Studies in cafeterias and laboratories generally produced more positive effects than those in quick-service restaurants or vending machines. Contextual labels listing daily recommended intake or including traffic lights or exercise equivalents displayed higher efficacy in this population. Both higher-quality and lower-quality studies generally showed positive effects of labeling. Field experiments, particularly with large representative samples and adequate controls, are warranted to assess the effect of nutrition labels among college students. The results of this study support nutrition-labeling policies, suggesting that implementing nutrition labels may improve dietary intake among college students. Colleges, universities, and other institutions should consider implementing nutrition labeling, particularly using contextual formats that allow for quick comparisons across food choices. Acknowledgments Author contributions. M.J.C. conceptualized the study, performed the systematic review and meta-analysis, and drafted the manuscript. R.A. oversaw the methods and contributed to each manuscript draft. Funding/support. M.J.C. is supported by the Health Resources and Services Administration (HRSA) of the US Department of Health and Human Services (HHS) under the National Research Service Award (NRSA) in Primary Medical Care, grant no. T32HP22239 (PI: Borowsky). This information or content and conclusions are those of the authors and should not be construed as the official position or policy of, nor should any endorsements be inferred by, the HRSA, the HHS, or the US government. Funding agencies were not involved in the preparation of this manuscript. Declaration of interest. The authors have no relevant interests to declare. References 1 Guthrie JF, Lin BH, Frazao E. Role of food prepared away from home in the American diet, 1977-78 versus 1994-96: changes and consequences. J Nutr Educ Behav.  2002; 34: 140– 150. http://dx.doi.org/10.1016/S1499-4046(06)60083-3 Google Scholar CrossRef Search ADS PubMed  2 Kant AK, Graubard BI. Eating out in America, 1987–2000: trends and nutritional correlates. Prev Med.  2004; 38: 243– 249. http://dx.doi.org/10.1016/j.ypmed.2003.10.004 Google Scholar CrossRef Search ADS PubMed  3 National Restaurant Association. Facts at a Glance. News and Research 2017. http://www.restaurant.org/News-Research/Research/Facts-at-a-Glance. Updated April 28, 2017. Accessed September 13, 2017. 4 Bezerra IN, Curioni C, Sichieri R. Association between eating out of home and body weight. Nutr Rev.  2012; 70: 65– 79. http://dx.doi.org/10.1111/j.1753-4887.2011.00459.x Google Scholar CrossRef Search ADS PubMed  5 Swartz J, Braxton D, Viera A. Calorie menu labeling on quick-service restaurant menus: an updated systematic review of the literature. Int J Behav Nutr Phys Act.  2011; 8: 135. doi:10.1186/1479-5868-8-135 Google Scholar CrossRef Search ADS PubMed  6 Kiszko K, Martinez O, Abrams C, et al.   The influence of calorie labeling on food orders and consumption: a review of the literature. J Community Health.  2014; 39: 1248– 1269. http://dx.doi.org/10.1007/s10900-014-9876-0 Google Scholar CrossRef Search ADS PubMed  7 Fernandes AC, Oliveira RC, Proenca RP, et al.   Influence of menu labeling on food choices in real-life settings: a systematic review. Nutr Rev.  2016; 74: 534– 548. http://dx.doi.org/10.1093/nutrit/nuw013 Google Scholar CrossRef Search ADS PubMed  8 Kim SY, Nayga RM, Capps O. The effect of food label use on nutrient intakes: an endogenous switching regression analysis. J Agri Resource Econ . 2000; 215– 231. 9 Christoph MJ, An R, Ellison B. Correlates of nutrition label use among college students and young adults: a review. Public Health Nutr.  2016; 19: 2135– 2148. http://dx.doi.org/10.1017/S1368980015003183 Google Scholar CrossRef Search ADS PubMed  10 Bleich SN, Wolfson JA. Differences in consumer use of food labels by weight loss strategies and demographic characteristics. BMC Public Health . 2015; 15: 1275. doi: 10.1186/s12889-015-2651-z Google Scholar CrossRef Search ADS PubMed  11 Tandon PS, Wright J, Chuan Z, et al.   Nutrition menu labeling may lead to lower-calorie restaurant meal choices for children. Pediatrics . 2010; 125: 244– 248. http://dx.doi.org/10.1542/peds.2009-1117 Google Scholar CrossRef Search ADS PubMed  12 Pratt NS, Ellison BD, Benjamin AS, et al.   Improvements in recall and food choices using a graphical method to deliver information of select nutrients. Nutr Res.  2016; 36: 44– 56. http://dx.doi.org/10.1016/j.nutres.2015.10.009 Google Scholar CrossRef Search ADS PubMed  13 Roberto CA, Larsen PD, Agnew H, et al.   Evaluating the impact of menu labeling on food choices and intake. Am J Public Health.  2010; 100: 312– 318. http://dx.doi.org/10.2105/AJPH.2009.160226 Google Scholar CrossRef Search ADS PubMed  14 Ellison B, Lusk JL, Davis D. Looking at the label and beyond: the effects of calorie labels, health consciousness, and demographics on caloric intake in restaurants. Int J Behav Nutr Phys Act.  2013; 10: 21. doi:10.1186/1479-5868-10-21 Google Scholar CrossRef Search ADS PubMed  15 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. http://dx.doi.org/10.1016/j.jand.2014.05.014 Google Scholar CrossRef Search ADS PubMed  16 Cowburn G, Stockley L. Consumer understanding and use of nutrition labelling: a systematic review. Public Health Nutr.  2005; 8: 21– 28. http://dx.doi.org/10.1079/PHN2005666 Google Scholar CrossRef Search ADS PubMed  17 Liu PJ, Roberto CA, Liu LJ, et al.   A test of different menu labeling presentations. Appetite . 2012; 59: 770– 777. http://dx.doi.org/10.1016/j.appet.2012.08.011 Google Scholar CrossRef Search ADS PubMed  18 van Herpen E, Trijp HC. Front-of-pack nutrition labels. Their effect on attention and choices when consumers have varying goals and time constraints. Appetite . 2011; 57: 148– 160. Google Scholar CrossRef Search ADS PubMed  19 Long MW, Tobias DK, Cradock AL, et al.   Systematic review and meta-analysis of the impact of restaurant menu calorie labeling. Am J Public Health.  2015; 105: e11– e24. Google Scholar CrossRef Search ADS PubMed  20 Harnack L, French S. Effect of point-of-purchase calorie labeling on restaurant and cafeteria food choices: a review of the literature. Int J Behav Nutr Phys Act.  2008; 5: 51. doi:10.1186/1479-5868-5-51 Google Scholar CrossRef Search ADS PubMed  21 Cecchini M, Warin L. Impact of food labelling systems on food choices and eating behaviours: a systematic review and meta-analysis of randomized studies. Obes Rev.  2016; 17: 201– 210. http://dx.doi.org/10.1111/obr.12364 Google Scholar CrossRef Search ADS PubMed  22 Harris K, Gordon-Larsen P, Chantala K, et al.   Longitudinal trends in race/ethnic disparities in leading health indicators from adolescence to young adulthood. Arch Pediatr Adolesc Med.  2006; 160: 74– 81. http://dx.doi.org/10.1001/archpedi.160.1.74 Google Scholar CrossRef Search ADS PubMed  23 Nelson MC, Story M, Larson NI, et al.   Emerging adulthood and college-aged youth: an overlooked age for weight-related behavior change. Obesity . 2008; 16: 2205– 2211. http://dx.doi.org/10.1038/oby.2008.365 Google Scholar CrossRef Search ADS PubMed  24 Vella-Zarb RA, Elgar FJ. The ‘Freshman 5’: a meta-analysis of weight gain in the freshman year of college. J Am Coll Health.  2009; 58: 161– 166. Google Scholar CrossRef Search ADS PubMed  25 Roy R, Kelly B, Rangan A, et al.   Food environment interventions to improve the dietary behavior of young adults in tertiary education settings: a systematic literature review. J Acad Nutr Diet.  2015; 115: 1647– 1681. http://dx.doi.org/10.1016/j.jand.2015.06.380 Google Scholar CrossRef Search ADS PubMed  26 Kelly NR, Mazzeo SE, Bean MK. Systematic review of dietary interventions with college students: directions for future research and practice. J Nutr Educ Behav . 2013; 45: 304– 313. http://dx.doi.org/10.1016/j.jneb.2012.10.012 Google Scholar CrossRef Search ADS PubMed  27 Deliens T, Van Crombruggen R, Verbruggen S, et al.   Dietary interventions among university students: a systematic review. Appetite . 2016; 105: 14– 26. http://dx.doi.org/10.1016/j.appet.2016.05.003 Google Scholar CrossRef Search ADS PubMed  28 Liberati A, Altman DG, Tetzlaff J, et al.   The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. PLoS Med . 2009; 6: e1000100. 10.1371/journal.pmed.1000100 Google Scholar CrossRef Search ADS PubMed  29 Ellison B, Lusk JL, Davis D. The impact of restaurant calorie labels on food choice: results from a field experiment. Econ Inquiry . 2014; 52: 666– 681. http://dx.doi.org/10.1111/ecin.12069 Google Scholar CrossRef Search ADS   30 Hoerr SM, Louden VA. Can nutrition information increase sales of healthful vended snacks? J Sch Health.  1993; 63: 386– 390. Google Scholar CrossRef Search ADS PubMed  31 Schwartz J, Riis J, Elbel B, et al.   Inviting consumers to downsize fast-food portions significantly reduces calorie consumption. Health Aff . 2012; 31: 399– 407. http://dx.doi.org/10.1377/hlthaff.2011.0224 Google Scholar CrossRef Search ADS   32 Roy R, Beattie-Bowers J, Ang SM, et al.   The effect of energy labelling on menus and a social marketing campaign on food-purchasing behaviours of university students. BMC Public Health . 2016; 16: 727. doi:10.1186/s12889-016-3426-x Google Scholar CrossRef Search ADS PubMed  33 Temple JL, Johnson KM, Archer K, et al.   Influence of simplified nutrition labeling and taxation on laboratory energy intake in adults. Appetite . 2011; 57: 184– 192. http://dx.doi.org/10.1016/j.appet.2011.04.018 Google Scholar CrossRef Search ADS PubMed  34 Aaron JI, Evans RE, Mela DJ. Paradoxical effect of a nutrition labeling scheme in a student cafeteria. Nutr Res . 1995; 15: 1251– 1261. http://dx.doi.org/10.1016/0271-5317(95)02001-C Google Scholar CrossRef Search ADS   35 Dingman DA, Schulz MR, Wyrick DL, et al.   Does providing nutrition information at vending machines reduce calories per item sold? J Public Health Pol.  2015; 36: 110– 122. Google Scholar CrossRef Search ADS   36 James A, Adams-Huet B, Shah M. Menu labels displaying the kilocalorie content or the exercise equivalent: effects on energy ordered and consumed in young adults. Am J Health Promot.  2015; 29: 294– 302. http://dx.doi.org/10.4278/ajhp.130522-QUAN-267 Google Scholar CrossRef Search ADS PubMed  37 Chu Y, Frongillo E, Jones S, et al.   Improving patrons' meal selections through the use of point-of-selection nutrition labels. Am J Public Health.  2009; 99: 2001– 2005. http://dx.doi.org/10.2105/AJPH.2008.153205 Google Scholar CrossRef Search ADS PubMed  38 Hammond D, Lillico HG, Vanderlee L, et al.   The impact of nutrition labeling on menus: a naturalistic cohort study. Am J Health Behav.  2015; 39: 540– 548. http://dx.doi.org/10.5993/AJHB.39.4.10 Google Scholar CrossRef Search ADS PubMed  39 Hoefkens C, Lachat C, Kolsteren P, et al.   Posting point-of-purchase nutrition information in university canteens does not influence meal choice and nutrient intake. Am J Clin Nutr.  2011; 94: 562– 570. http://dx.doi.org/10.3945/ajcn.111.013417 Google Scholar CrossRef Search ADS PubMed  40 Lillico HG, Hanning R, Findlay S, et al.   The effects of calorie labels on those at high-risk of eating pathologies: a pre-post intervention study in a University cafeteria. Public Health . 2015; 129: 732– 739. http://dx.doi.org/10.1016/j.puhe.2015.03.005 Google Scholar CrossRef Search ADS PubMed  41 Nikolaou CK, Hankey CR, Lean MEJ. Preventing weight gain with calorie-labeling. Obesity (Silver Spring).  2014; 22: 2277– 2283. http://dx.doi.org/10.1002/oby.20885 Google Scholar CrossRef Search ADS PubMed  42 National Heart, Blood, and Lung Institute. Study Quality Assessment Tools. 2014. https://www.nhlbi.nih.gov/health-topics/study-quality-assessment-tools. Accessed December 28, 2017. . 43 Cioffi CE, Levitsky DA, Pacanowski CR, et al.   A nudge in a healthy direction. The effect of nutrition labels on food purchasing behaviors in university dining facilities. Appetite . 2015; 92: 7– 14. http://dx.doi.org/10.1016/j.appet.2015.04.053 Google Scholar CrossRef Search ADS PubMed  44 Temple JL, Johnson K, Recupero K, et al.   Nutrition labels decrease energy intake in adults consuming lunch in the laboratory. J Am Diet Assoc . 2010; 110: 1094– 1097. http://dx.doi.org/10.1016/j.jada.2010.04.006 Google Scholar CrossRef Search ADS PubMed  45 Cinciripini PM. Changing food selections in a public cafeteria: an applied behavior analysis. Behav Modif.  1984; 8: 520– 539. http://dx.doi.org/10.1177/01454455840084004 Google Scholar CrossRef Search ADS   46 Davis-Chervin D, Rogers T, Clark M. Influencing food selection with point-of-choice nutrition information. J Nutr Educ . 1985; 17: 18– 22. http://dx.doi.org/10.1016/S0022-3182(85)80014-5 Google Scholar CrossRef Search ADS   47 Freedman MR. Point-of-selection nutrition information influences choice of portion size in an all-you-can-eat university dining hall. J Foodservice Business Res . 2011; 14: 86– 98. http://dx.doi.org/10.1080/15378020.2011.548228 Google Scholar CrossRef Search ADS   48 Nikolaou CK, Lean MEJ, Hankey CR. Calorie-labelling in catering outlets: acceptability and impacts on food sales. Prev Med.  2014; 67: 160– 165. http://dx.doi.org/10.1016/j.ypmed.2014.07.027 Google Scholar CrossRef Search ADS PubMed  49 Larson-Brown LB. Point-of-purchase information on vended foods. J Nutr Educ . 1978; 10: 116– 118. http://dx.doi.org/10.1016/S0022-3182(78)80053-3 Google Scholar CrossRef Search ADS   50 Freedman MR, Connors R. Point-of-purchase nutrition information influences food-purchasing behaviors of college students: a pilot study. J Am Diet Assoc . 2010; 111: S42– S46. Google Scholar CrossRef Search ADS   51 Stran KA, Knol LL, Turner LW, et al.   College students must overcome barriers to use calorie labels in fast-food restaurants. J Nutr Educ Behav.  2016; 48: 122.e1– 130.e121. Google Scholar CrossRef Search ADS   52 Emrich TE, Qi Y, Lou WY, et al.   Traffic-light labels could reduce population intakes of calories, total fat, saturated fat, and sodium. PLoS One.  2017; 12: e0171188. doi:10.1016/j.jneb.2015.09.009 Google Scholar CrossRef Search ADS PubMed  53 Guenther PM, Casavale KO, Reedy J, et al.   Update of the Healthy Eating Index: HEI-2010. J Acad Nutr Diet.  2013; 113: 569– 580. http://dx.doi.org/10.1016/j.jand.2012.12.016 Google Scholar CrossRef Search ADS PubMed  54 Panagiotakos DB, Pitsavos C, Stefanadis C. Dietary patterns: a Mediterranean diet score and its relation to clinical and biological markers of cardiovascular disease risk. Nutr Metab Cardiovasc Dis . 2006; 16: 559– 568. http://dx.doi.org/10.1016/j.numecd.2005.08.006 Google Scholar CrossRef Search ADS PubMed  55 Cooke R, Papadaki A. Nutrition label use mediates the positive relationship between nutrition knowledge and attitudes towards healthy eating with dietary quality among university students in the UK. Appetite . 2014; 83: 297– 303. http://dx.doi.org/10.1016/j.appet.2014.08.039 Google Scholar CrossRef Search ADS PubMed  56 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. http://dx.doi.org/10.1016/j.jada.2011.08.047 Google Scholar CrossRef Search ADS PubMed  57 Ollberding NJ, Wolf RL, Contento I. Food label use and its relation to dietary intake among US adults. J Am Diet Assoc . 2010; 110: 1233– 1237. http://dx.doi.org/10.1016/j.jada.2010.05.007 Google Scholar CrossRef Search ADS PubMed  58 Menu labeling requirements. Final rule. Fed Regist.  2014; 79 230: 71155– 71259. PubMed  59 Hawkes C, Smith TG, Jewell J, et al.   Smart food policies for obesity prevention. Lancet.  2015; 385: 2410– 2421. http://dx.doi.org/10.1016/S0140-6736(14)61745-1 Google Scholar CrossRef Search ADS PubMed  60 Nikolaou CK, Hankey CR, Lean MEJ. Calorie-labelling: does it impact on calorie purchase in catering outlets and the views of young adults? Int J Obes (Lond).  2015; 39: 542– 545. Google Scholar CrossRef Search ADS PubMed  61 Christoph MJ, Ellison B. A cross-sectional study of the relationship between nutrition label use and food selection, servings, and consumption in a university dining setting. J Acad Nutr Diet . 2017; 117: 1528– 1537. http://dx.doi.org/10.1016/j.jand.2017.01.027 Google Scholar CrossRef Search ADS PubMed  62 Steffen LM, Van Horn L, Daviglus ML, et al.   A modified Mediterranean diet score is associated with a lower risk of incident metabolic syndrome over 25 years among young adults: the CARDIA (Coronary Artery Risk Development in Young Adults) study. Br J Nutr.  2014; 112: 1654– 1661. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press on behalf of the International Life Sciences Institute. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com.

Journal

Nutrition ReviewsOxford University Press

Published: Mar 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 12 million articles from more than
10,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Unlimited reading

Read as many articles as you need. Full articles with original layout, charts and figures. Read online, from anywhere.

Stay up to date

Keep up with your field with Personalized Recommendations and Follow Journals to get automatic updates.

Organize your research

It’s easy to organize your research with our built-in tools.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve Freelancer

DeepDyve Pro

Price
FREE
$49/month

$360/year
Save searches from Google Scholar, PubMed
Create lists to organize your research
Export lists, citations
Read DeepDyve articles
Abstract access only
Unlimited access to over
18 million full-text articles
Print
20 pages/month
PDF Discount
20% off