alternative approaches to the calculation of nutrient density

7
Alternative approaches to the calculation of nutrient density Eileen Kennedy, Patrick Racsa, Gerard Dallal, Alice H Lichtenstein, Jeanne Goldberg, Paul Jacques, and Raymond Hyatt A renewed interest in promoting health and wellness has prompted both public- and private-sector organizations to adopt systems for rating the nutritional quality of individual food products. Compared here are three food quality scores for ranking foods. The absolute score varies by the food quality score algorithm used, but the relative ranking of foods within a food group is stable. Fruits and vegetables are substantially more nutrient-dense than items from other food groups. There is an imperative need for a simple, consistent method to guide consumers in making healthier food choices. © 2008 International Life Sciences Institute INTRODUCTION The “Nutrition Facts” panel developed by the US Food and Drug Administration (FDA) in the early 1990s has been a key product-based resource for consumers seeking to choose nutritious foods. 1 In 2007, the FDA unveiled two new tools to enhance the impact of this label on consumers’ food choices and their ability to achieve a healthy weight 2 : Make Your Calories Count (a web-based learning program) and the Nutrition Facts Brochure. Indeed, in the past 10 years there has been a proliferation of systems in both the public and private sectors that are intended to assist consumers in making more nutritious food choices. A major emphasis of these new systems for rating the nutritional quality of food has been on using the information to educate consumers on front-of-pack (FOP) labeling of food. A FOP approach is viewed as an attractive way to complement the information that is con- tained in the “Nutrition Facts” panel on the back of food products. The widespread emphasis on health and well- ness has been spurred on, in part, by the alarming statis- tics on overweight and obesity. 3 Thus, a number of approaches for nutrition profiling – scoring or rating foods based upon individual components present in a food – have become common. 4 The logic behind nutri- tional profiling is that if individual food choices can be more nutritious, the total diet will improve; in turn, this will contribute to improving an individual’s nutritional status. The statement “We are an overfed but under- nourished nation” expresses a sentiment that echoes the concerns of many US health officials in the current environment. 5 In the United States, many people are con- suming diets that are energy dense but nutrient poor. 6 In an effort to reverse this trend, varied approaches for helping consumers make healthier food choices have been developed. A number of these approaches use nutri- ent density as the basis for defining a nutritious food. Nutrient density is not a new concept. Over 30 years ago, researchers developed a variety of different methods for rating or measuring the nutritional quality of foods. 7–9 Nutrient density, as the initial concept emerged, was most commonly defined as the ratio of the amount of nutrients in a food to the energy provided. The nutrient density equations used were the ratio of nutrients contained in a food to the energy provided by the food item on a per serving, a per 100 calorie, or a per 100 g as consumed basis. 7–9 The earliest nutrient density scores were devel- oped prior to the first USDA/HHS Dietary Guidelines for Americans in 1980. 10 Recently, there has been a renewed emphasis on the concept of nutrient density as a result of recommendations made in the 2005 Dietary Guidelines for Americans (DGA). 11 Statements from these most recent guidelines such as “Get the most nutrition out of Affiliations: E Kennedy, P Racsa, G Dallal, AH Lichtenstein, J Goldberg, P Jacques and R Hyatt are with the Friedman School of Nutrition Science and Policy, Tufts University, Boston, Massachusetts, USA. Correspondence: E Kennedy, Friedman School of Nutrition Science and Policy, Tufts University, 150 Harrison Avenue, Boston, MA 02111, USA. E-mail: [email protected], Phone: +1-617-636-3702, Fax: +1-617-636-3794. Key words: food quality score, front-of-pack labeling, nutrient density, nutrition profiling NutritionScience Policy doi:10.1111/j.1753-4887.2008.00124.x Nutrition Reviews® Vol. 66(12):703–709 703

Upload: eileen-kennedy

Post on 21-Jul-2016

212 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Alternative approaches to the calculation of nutrient density

Alternative approaches to the calculation of nutrient density

Eileen Kennedy, Patrick Racsa, Gerard Dallal, Alice H Lichtenstein, Jeanne Goldberg, Paul Jacques, andRaymond Hyatt

A renewed interest in promoting health and wellness has prompted both public- andprivate-sector organizations to adopt systems for rating the nutritional quality ofindividual food products. Compared here are three food quality scores for rankingfoods. The absolute score varies by the food quality score algorithm used, but therelative ranking of foods within a food group is stable. Fruits and vegetables aresubstantially more nutrient-dense than items from other food groups. There is animperative need for a simple, consistent method to guide consumers in makinghealthier food choices.© 2008 International Life Sciences Institute

INTRODUCTION

The “Nutrition Facts” panel developed by the US Foodand Drug Administration (FDA) in the early 1990s hasbeen a key product-based resource for consumers seekingto choose nutritious foods.1 In 2007, the FDA unveiledtwo new tools to enhance the impact of this label onconsumers’ food choices and their ability to achieve ahealthy weight2: Make Your Calories Count (a web-basedlearning program) and the Nutrition Facts Brochure.Indeed, in the past 10 years there has been a proliferationof systems in both the public and private sectors that areintended to assist consumers in making more nutritiousfood choices. A major emphasis of these new systems forrating the nutritional quality of food has been on usingthe information to educate consumers on front-of-pack(FOP) labeling of food. A FOP approach is viewed as anattractive way to complement the information that is con-tained in the “Nutrition Facts” panel on the back of foodproducts. The widespread emphasis on health and well-ness has been spurred on, in part, by the alarming statis-tics on overweight and obesity.3 Thus, a number ofapproaches for nutrition profiling – scoring or ratingfoods based upon individual components present in afood – have become common.4 The logic behind nutri-tional profiling is that if individual food choices can bemore nutritious, the total diet will improve; in turn, this

will contribute to improving an individual’s nutritionalstatus.

The statement “We are an overfed but under-nourished nation” expresses a sentiment that echoes theconcerns of many US health officials in the currentenvironment.5 In the United States, many people are con-suming diets that are energy dense but nutrient poor.6 Inan effort to reverse this trend, varied approaches forhelping consumers make healthier food choices havebeen developed. A number of these approaches use nutri-ent density as the basis for defining a nutritious food.

Nutrient density is not a new concept. Over 30 yearsago, researchers developed a variety of different methodsfor rating or measuring the nutritional quality of foods.7–9

Nutrient density, as the initial concept emerged, was mostcommonly defined as the ratio of the amount of nutrientsin a food to the energy provided. The nutrient densityequations used were the ratio of nutrients contained in afood to the energy provided by the food item on a perserving, a per 100 calorie, or a per 100 g as consumedbasis.7–9 The earliest nutrient density scores were devel-oped prior to the first USDA/HHS Dietary Guidelines forAmericans in 1980.10 Recently, there has been a renewedemphasis on the concept of nutrient density as a result ofrecommendations made in the 2005 Dietary Guidelinesfor Americans (DGA).11 Statements from these mostrecent guidelines such as “Get the most nutrition out of

Affiliations: E Kennedy, P Racsa, G Dallal, AH Lichtenstein, J Goldberg, P Jacques and R Hyatt are with the Friedman School of NutritionScience and Policy, Tufts University, Boston, Massachusetts, USA.

Correspondence: E Kennedy, Friedman School of Nutrition Science and Policy, Tufts University, 150 Harrison Avenue, Boston, MA 02111,USA. E-mail: [email protected], Phone: +1-617-636-3702, Fax: +1-617-636-3794.

Key words: food quality score, front-of-pack labeling, nutrient density, nutrition profiling

Nutrition↔Science Policy

doi:10.1111/j.1753-4887.2008.00124.xNutrition Reviews® Vol. 66(12):703–709 703

Page 2: Alternative approaches to the calculation of nutrient density

your calories” and “Consume a variety of nutrient-densefoods and beverages within and among the basic foodgroups”, focus a clear emphasis on nutrient density as away to guide good choices.11

In addition, the USDA My Pyramid, which was alsoissued in 2005,12 uses nutrient density both within andacross the food groups. Unlike the original Food GuidePyramid issued in 1992,13 the 2005 My Pyramid uses nofood icons. Rather, My Pyramid uses colored bands orarrays to convey the messages of moderation, proportion-ality, and variety. The wider the colored bands, the morefood items need to come from within a specific foodgroup; grains, fruits, and vegetables are the food groupsemphasized in My Pyramid followed by foods from themeat and milk groups. Within the major food groups,some choices are more nutritious than others. The thin-ning colored bands from bottom to top in the Pyramidindicate that some food items in a given food group aremore nutritious than others. While the consensus is thatwithin each food group consumers should choose only alimited number of foods from the tip of the band, this hasbeen a difficult message to communicate. It is here thatmuch of the criticism of My Pyramid has occurred.14

Many professionals believe My Pyramid is an improve-ment over the 1992 version, yet there has been a deluge ofcriticism that the new Pyramid does not, by itself, lay outclearly what people should be eating.15

What is clear is that most Americans do not havedietary patterns consistent with the DGA. Using theUSDA Healthy Eating Index (HEI), a single summarymeasure to evaluate total diet quality, results indicate thatin both 199516 and 200517 the majority of people had dietsthat did not meet the Dietary Guidelines for Americans.Indeed, the diets of individuals improved only slightlysince the 1995 HEI was used to measure dietary quality.This has prompted researchers, as well as public- andprivate-sector organizations, to devote attention to devel-oping approaches, based on nutrient density, that can beused as tools to help consumers improve their dietarypatterns by selecting more nutritious individual fooditems.

NUTRITIONAL SCIENCE LINKED TO ACTION: RATINGINDIVIDUAL FOODS

Various methods have been developed over the past threedecades to provide a nutritional profile of specificfoods.4,5,18 Different approaches have then been used torate specific foods, either qualitatively or quantitavely. Forexample, the American Heart Association’s (AHA) “heartcheckmark” program is given to foods that meet the(FDA) criteria for a healthy food claim; AHA allows theircheckmark if one serving of the product has less than orequal to 3 g of fat, 1 g of saturated fat, 20 mg of choles-

terol, 480 mg of sodium, or 0.5 g of trans fat. The Alliancefor a Healthier Generation collaboration between theClinton Foundation and an initial group of five majorfood companies – Dannon, Kraft, PepsiCo, Mars, andCampbell Soup Company – have agreed to voluntarynutrition standards for foods sold in schools.

In addition, there has recently been a proliferation ofsystems and icons from the private sector aimed athelping consumers make healthier food choices.4 Theemphasis on health and wellness has been an impetus forthe development of healthier versions of popular prod-ucts. PepsiCo launched a “Smart Spot” campaign wherebyfoods that receive a FOP icon – a check mark – must meetnutritional targets. These criteria include reduced calo-ries, fat, sodium, or sugars, and the product must containat least 10% of a specified target nutrient. PepsiCo’s FOPlabeling indicating foods that are“better for you” is part ofa larger campaign on health and wellness emphasizingphysical activity, among other components. Other foodmanufacturers also use FOP symbols to indicate healthierfood choices; these include Kraft’s “Sensible Solutions”,Nestle Company’s “Nutritional Compass”, and Unilever’s“Choices” program.

In addition to individual food companies, privatesector retailers have also used a variety of approachesto promote the purchase of more nutritious foods.Hannaford Brothers, a supermarket chain in NewEngland, launched their “Guiding Stars” program aspart of an emphasis on health and wellness for customers.This proprietary system assigns one to three stars to selectfoods. In essence, a product receiving one star is good, twostars are better, and a three-star rating is the best. A limi-tation of the Hannaford system is that only about 26% ofproducts warrant any stars. In addition, there is variancein the foods getting stars from the Hannaford systems andother public sector rating systems. For example, V8® juicegets an AHA checkmark, yet under Hannaford’s “GuidingStars” system, the product gets no stars. The inconsis-tency in identifying products that are “better for you”across the different approaches is problematic. There isincreasing concern that this proliferation of FOP iconsand rating systems is adding to consumer confusion. Todate, there has been no systematic evaluation of the effectof any of the rating systems or icons on dietary patterns.

The interest in promoting “healthier” food productsis not limited to the United States: in the European Unionrather active discussions on the topic are ongoing. Inaddition, the International Union of Nutritional Sciences(IUNS), which represents scientific bodies from 79 coun-tries around the world, has convened a task force toexamine the issue of science-based measures for evaluat-ing diet quality. The challenge is to develop methods forassessing the impact of improved, individual food choiceson the total diet. Given the emphasis on obesity preven-

Nutrition Reviews® Vol. 66(12):703–709704

Page 3: Alternative approaches to the calculation of nutrient density

tion, many nutrient profiling or scoring methods areusing nutrient density as an approach to develop icons orrating systems to be used on the FOP.

NUTRIENT DENSITY REVISITED

Despite the fact that USDA and HHS have used theconcept of nutrient density throughout the 2005 DietaryGuidelines report,11 there is no agreed upon federal gov-ernment method for calculating nutrient density. Toassess the effect of alternative measures for rating indi-vidual foods based on nutrient density, a series of foodquality scores (FQS), based on the 2005 DGA, were devel-oped and tested. Each FQS is a nutrient-density algorithmthat is used to rate individual foods and is based on opti-mizing the intake of specific nutrients relative to the calo-ries provided. Results for three distinct FQS are comparedand contrasted. Implications for arriving at a harmonizedapproach are then discussed.

The 2005 DGA advise the US population to selectnutrient-dense foods within their diets.11 The FQS werebased on guidance provided in the latest DGA. TheDGA identify what are termed shortfall nutrients;11

these have been defined as essential nutrients consumedin low enough amounts (less than 60% of the RDA) inthe United States to be of concern and even at levels solow as to possibly be detrimental to health. For adults,these nutrients include calcium, potassium, vitamins A(as carotenoids), C and E, fiber, and magnesium. Thesame group of nutrients is classified as shortfall nutri-ents for children and adolescents with the exception ofvitamins A and C. In addition, there are subgroups ofthe population that appear to lack several nutrients intheir diets: for people over the age of 50 years, theseinclude vitamins B12 and D. Consumption of vitamin Dmay also be inadequate for people with dark skin andthose exposed to insufficient sunlight. Women of child-bearing age and those in the first trimester of pregnancyare at a higher risk of insufficient intake of iron andfolic acid.

Not all 54 essential nutrients are included in calcu-lating the FQS since only a subset are classified as shortfallnutrients. Here again, guidance from the 2005 DGA indi-cate that, for example, while protein is important as anessential nutrient, it is also consumed in sufficient quan-tities in the U.S. diet and therefore need not be classifiedas a shortfall nutrient.11 Therefore, while protein is clearlyessential, it is, on average, adequate in the American diet.

The 2005 DGA11 also identify a series of nutrientstermed “avoidance nutrients.” Avoidance nutrients arethose for which overconsumption may result in adversehealth effects. These include calories, saturated fats, cho-lesterol, sodium, and trans fatty acids.

OPERATIONALIZING NUTRIENT DENSITY

Several FQS were developed based on the ratio of short-fall nutrients (in the numerator) to avoidance nutrients(in the denominator) (Table 1). The reference value foreach of the eleven shortfall nutrients was the daily value(DV) used by the FDA for the food label.19 The DV isbased on a 2000-calorie prototype diet, which is also thereference level used by the FDA for labeling. In usingthe FQS to rate the individual products, foods can not beparticularized to a specific calorie level for each con-sumer. Thus, a 2000-calorie reference value was used.Each of the shortfall nutrients has an established DV. Foreach of the shortfall nutrients, a maximum of 100% of theDV is possible. The maximum value for each shortfallnutrient was truncated at 100%, since amounts above thislevel are unlikely to produce a health benefit and in somecases pose a health risk.

Among the five avoidance nutrients, calories andtrans fatty acids do not have an established DV. Therefore,for food energy, a 2000-calorie reference value wasused. The caloric content of the specific food item wasexpressed as a percentage of a 2000-calorie referencevalue. While no DV has been established for trans fattyacids, the 2005 DGA suggest the following: “Keep transfatty acid consumption as low as possible.”11 The Ameri-can Heart Association’s recommendations are to keep

Table 1 Food quality score algorithms.FQS type Numerator componentsFQS1 (Universal) % DV: (fiber, vitamins A, C,

E, D, and B12, folate,calcium, magnesium,iron, potassium)/11

FQS2 (Food-group-specific) Fruits and vegetables: % DV(fiber, vitamins A, C andE, folate, calcium, iron,potassium)/8

Grains: % DV (fiber,vitamins A and E, folate,potassium, calcium, iron,magnesium)/8

Dairy: %DV (calcium,potassium, vitamins Aand D, magnesium)/5

FQS3 (Expanded) In addition to nutrients inFQS1: %DV (protein,phosphorous, zinc,copper, thiamin,riboflavin, niacin,pantothenic acid,vitamin K, manganese,selenium, vitamin B6)/23

Note: the denominator in each FQS is the same and equals (%DV calories, % DV saturated fats, % DV cholesterol, % DVsodium, % calories from trans fats)/5.

Nutrition Reviews® Vol. 66(12):703–709 705

Page 4: Alternative approaches to the calculation of nutrient density

trans fatty acids to less than 1% of total energy consump-tion.20 Therefore, for trans fatty acids, the FQS uses thepercent of calories from trans fatty acids. By using this incomputing the FQS, the score decreases when the transfatty acids are high and increases when the trans fatty acidcontent of the food is low. Unlike the computation forshortfall nutrients, the value for each avoidance nutrientwas not truncated at 100%. Avoidance nutrients are notcapped because additional amounts of these nutrients inexcess of 100% can worsen diet quality and health. Alower limit of 0.05% was set for calculating the values foravoidance nutrients. When trace amounts of particularnutrients are found, exceptionally low avoidance scoresartificially increase the FQS; thus, a floor of 0.05% was setfor each avoidance nutrient.

Each FQS is a nutrition density algorithm based onthe ratio of the average percent of the shortfall nutrientsto the average percent of the avoidance nutrients. Inessence, a high score on the FQS indicates that the foodcontains more nutrients per calorie consumed; a lowerscore indicates the reverse, i.e., energy but fewer nutri-ents. Worth noting is the fact that the denominatorcontains calories but not total percent of fat. Foodenergy and calories from total fat are correlated; thus,only the percent of total calories is included in thecalculation.

Three FQS are compared: universal (FQS1); food-group-specific (FQS2); and expanded (FQS3). FQS1 is auniversal algorithm based on the same formula appliedto all foods in the five major food groups. Its numeratorincludes all of the shortfall nutrients. FQS2 is a special-ized algorithm based on the five major food groups:grains, fruits, vegetables, meat, and dairy. The DietaryGuidelines11 are clear that items from each of the majorfood groups should be consumed, since each food groupis a good source of different nutrients (Table 1). Thus,for FQS2, the shortfall nutrients that are included inthe numerator of the algorithm are food group specific.This prevents foods from being penalized for failing tohave nutrients that are not common to their food group.For example, fruits and vegetables, on average, providefiber, vitamins A (as carotenoids), C and E, folate,calcium, iron, and potassium. Milk, on the other hand, isnot a source of iron, so milk products should not bepenalized for not containing iron, even though iron isan essential shortfall nutrient. The expanded FQS3 is cal-culated in the manner of FQS1 but with additionalnutrients included in the numerator of the algorithm.The additional nutrients include zinc, copper, thiamin,riboflavin, niacin, B6, vitamin K, selenium, pantothenicacid, manganese, and protein. A key reason for includingthis larger group of nutrients was to ascertain if somefoods had a higher FQS with more nutrients in thecalculation.

ALTERNATIVE APPROACHES TO NUTRIENT DENSITY

The three FQS were applied to the USDA Nutrient Data-base for Standard Reference Release 1821 (SR-18). TheSR-18 contains nutrient information on 7146 foods.Complete nutrient profiles were not available for all 7146foods in the database. As a result, missing nutrient valueswere replaced with the mean nutrient values per foodgroup. Data regarding vitamin D and trans fatty acids arenovel in USDA SR-18. Standardization artificially inflatesvalues in certain food groups for vitamin D and/or transfatty acids as only a few foods particularly high in thesevalues have been tested. Each food item was evaluated fora standard serving size and per 100 g. A standard servingsize is necessary to ensure uniform comparisons amongfoods. Results reported in this paper use only 100 g por-tions, which is the standard reference value used in SR-18.Food items like pizza with components falling into morethan one food group category were classified into a sixthfood group of mixed dishes; these dishes were dividedinto their component food-group parts and an FQS basedon the proportionate contribution of each food groupwas calculated. Nutrients added to foods through fortifi-cation were treated in an identical fashion to naturallyoccurring nutrients.

The three FQS systems were applied to each of the7146 foods and a mean FQS was computed for each of thethree algorithms. In addition, foods were divided intotertile rankings within each category, with a ranking ofone reflecting the lowest level of nutrients. Results forboth the mean score and the tertile rankings for illustra-tive foods are shown in Table 2. The absolute values deter-mined with each of the three FQS systems differ forindividual food items. On average, the FQS2, or the food-group-specific score, tends to deviate most from the uni-versal score (FQS1) and the expanded score (FQS3). Thedifference is due to the larger number of shortfall nutri-ents that are included in the numerators of FQS1 and theeven greater number of nutrients included in FQS3. Itemsfrom specific food groups will not be good sources of allnutrients. For example, many fruits and vegetables scorelow on protein in the expanded FQS, yet are good sourcesof vitamins A and C and minerals. The average FQS2

scores tend to be higher for most food items since they arebased on the shortfall nutrients that are found in plentifulsources in each particular food category. For example,note in Table 2 that the FQS2 of 7.62 for nonfat fluid milkis almost twice as high as the FQS1 and the FQS3, whichare identical at 4.30.

Differences in the ratings of individual food itemswithin a food group category are apparent. For example,oranges are more nutrient dense when compared toapples regardless of the FQS employed (Table 2). Simi-larly, legumes are significantly more nutrient dense than

Nutrition Reviews® Vol. 66(12):703–709706

Page 5: Alternative approaches to the calculation of nutrient density

frankfurters or beef. Despite the differences among theabsolute values for the three FQS, the relative ranking offoods into tertile categories is similar among the threealgorithms. This indicates that the relative ranking offoods is invariant to the type of FQS used.

Table 3 presents the mean FQS based on FQS2 for thefive major food groups. Not surprisingly, fruits and veg-etables are more nutrient dense, on average, than itemsfrom the other food groups.

Table 4 presents data for the grains stratified forwhole grains compared to all grains. Average FQS valuesfor each of the three algorithms are 60% to 88% higher forwhole grains compared to all grains.

The issue of whether all foods can be part of a health-ful diet continues to be controversial.22 A final analysiswas done by establishing a “floor” level for each FQS.Foods that did not provide at least 5% of the DV for atleast one shortfall nutrient included in the numerator ofthe specific FQS were eliminated. Based on this criterion,232 foods were removed. Table 5 provides example of thetypes of foods that did not meet this minimum nutri-tional standard. Most items not achieving this minimalnutritional value are either alcoholic or carbonated bev-erages or certain snack-type foods.

Table 2 Comparison of three food quality scores and tertile rankings for selected foods.Food group Food quality score Rating (tertile)

FQS1 FQS2 FQS3 FQS1 FQS2 FQS3

DairyMilk: dry, whole, 3.25% milk fat 0.96 1.6 0.94 ++ ++ ++Milk: nonfat, fluid, with vitamin A added (fat free or skim) 4.3 7.62 4.3 +++ +++ +++Ice creams: vanilla 0.39 0.61 0.38 + + +

VegetablesTomatoes: red, ripe, raw, year round average 11.11 14.59 8.76 +++ +++ +++Catsup 0.68 0.88 0.66 + + +Carrots: raw 13 17.51 8.96 +++ +++ +++

FruitsApples: raw, with skin 4.18 5.46 3.19 ++ ++ +Oranges: raw, all commercial varieties 22.44 30.22 13.31 +++ +++ +++

GrainsBread: white, commercially prepared 0.98 1.11 1.34 ++ ++ ++Cake: angel food, commercially prepared 0.62 0.59 0.71 + + +Cereals: ready-to-eat, GENERAL MILLS, TOTAL Corn flakes 6.02 5.48 5.95 +++ +++ +++

Meats, fish, grains, legumes, eggsBeans: pinto, mature seeds, cooked, boiled, without salt 7.2 6.65 6.56 +++ +++ +++Frankfurter: beef 0.19 0.24 0.27 + + +Beef: ground, 70% lean/30% fat 0.32 0.39 0.45 + + +Egg: whole, raw, fresh 0.21 0.27 0.32 + + +Fish: cod, Atlantic, raw 1.15 1.3 1.99 +++ +++ +++

Note: +, lowest tertile; ++, middle tertile; +++, highest tertile.

Table 3 Food quality score averages by food group.Food group FQS score averageFruits 8.09Vegetables 8.02Grains 1.89Dairy 1.24Meats 1.03Other 1.21Note: based on food-group-specific (FQS2) score.

Table 4 Whole grains and all grains – average FQS.Grain type FQS1 FQS2 FQS3

All grains 1.89 1.96 2.45Whole grains 3.06 3.69 4.51Note: based on food-group-specific (FQS2) algorithm.

Table 5 Foods eliminated by criteria.FQS Food item1.17 Alcoholic beverage, beer, light1.00 Alcoholic beverage, wine, table all0.47 Cake, snack cakes, crème filled, sponge

(Twinkies)0.87 Carbonated beverage, cola, contains caffeine0.97 Potatoes, French fried, shoestring, salt added

in processing, frozen, oven-heated0.34 Doughnuts, cake-type, plain,

chocolate-coated or frosted0.33 Pie, coconut crème, commercially prepared0.25 Bologna, beef and pork, low fat0.24 Beef, cured, sausage, cooked, smoked0.33 Croissants, butter0.51 Turkey and gravy, frozen0.15 Candies, butterscotch0.52 Candies, caramels, chocolate-flavor roll

Nutrition Reviews® Vol. 66(12):703–709 707

Page 6: Alternative approaches to the calculation of nutrient density

DISCUSSION

The emphasis in the latest Dietary Guidelines for Ameri-cans11 on nutrient density as a way to guide food choiceshas focused renewed attention on the rigor of differentrating systems and icons for evaluating individual foods.The keen interest in nutritional profiling of foods trans-cends the United States. The UK government, forexample, recently announced a £372 million program tocombat obesity.23 A key part of this program involvespartnering with the food industry to develop a HealthyFood Code of Practice. A core component of the code ofpractice will be the adoption of one “simple and consis-tent nutritional labeling scheme for all foods to be usedon FOP labeling.”23 Similarly, in the United States, theCenter for Science in the Public Interest submitted a peti-tion to the Food and Drug Administration to urge thegovernment to consider adopting one food rating system(Jacobson MF, unpublished data). One key argument inthe petition is that the proliferation of icons and scoringsystems is creating confusion for the consumer. Thus, thepresent comparison of various methods of rating foods istimely. However, there is currently no agreed-uponapproach for FOP labeling that would be able to reach theconsumer; given the emphasis in the 2005 Dietary Guide-lines, the use of nutrient density as the basis for rating orscoring food products holds appeal.

Past work on nutrient density has tended to polarizeone approach for measuring nutrient density versusanother.24 In essence, organizations are arguing for onesystem compared to an alternative. The present report isone of the few places in which different algorithms havebeen used to systematically compare the results for dif-ferent nutrient density rating systems applied to the samefoods. Some clear findings emerge. First, the relativeranking of foods within tertile categories did not vary bytype of FQS. This is encouraging and indicates the ratingof individual foods is robust to the actual specification ofalgorithms. This is not to imply there are no differences.There were variations in the absolute scores for a singlefood among the three approaches tested. On average, thehighest mean scores for an individual food were shownwhen applying the food-group-specific algorithm FQS2.This is not surprising since FQS2 is based on the nutrientspresent in significant amounts in a particular food group– for example, calcium in milk products. However, thehigher average scores that were achieved using FQS2 wereconsistently elevated for each food item and within eachfood group; thus, the relative ranking within and acrossfood groups did not vary with the FQS used. Given this,among the three FQS types tested, FQS2 is preferred. Thisis consistent with comments in the 2005 Dietary Guide-lines Advisory Committee Report that a nutrient densityrating system is needed, which would “allow consumers

to choose more easily among similar foods within a foodgroup”.11 The use of FQS2 is also consistent with the 2005DGA11 and the 2005 My Pyramid,12 which suggest thatindividuals should consume foods from each of the majorfood groups, given the different nutrient profiles of foodsfrom different groups.

Whole grains score substantially higher on each ofthe three FQS when compared to grains as a whole. Thekey factor accounting for this difference is the fibercontent of whole grains. This reinforces the message thatconsumption of whole grain is important from a nutrientdensity perspective.

There continues to be controversy in the nutritioncommunity about whether or not all foods fit into ahealthful diet. Since the first edition of the Dietary Guide-lines was published in 1980, they have been based on theassumption that there are no good or bad foods; rather,there are good or bad dietary patterns. Eliminating foodsbased on a floor requirement of 5% DV for at least oneshortfall nutrient affected only 232 individual foods.Some might argue that more stringent criteria should beestablished to provide a higher standard that a foodshould meet in order to apply a FQS. The reasoningbehind this approach is that this would lead to improve-ments in dietary patterns. However, there is little, if any,research to indicate that a more stringent ranking ofindividual foods improves food purchases and/or foodconsumption, in part, because consumers don’t respondwell to negative messages.25

The private sector has increasingly emphasizedhealth and wellness in product development and mar-keting. However, many of the nutritional rating systemsdeveloped by the private sector have been difficult toevaluate since the processes and exact specificationsfor particular approaches are not transparent. Unlessalgorithms for nutritional profiling are available in thepeer-reviewed, scientific literature, it will be impossibleto critically evaluate the utility of one approach com-pared to another. Opinions or best guesses about whatworks most effectively are a poor metric for judgingrigor.

Even more challenging is the issue of reaching theconsumer with rating systems to improve food purchasesand diet quality. The ultimate test of any nutritionalscoring approach is how effectively these ranking systemshelp consumers purchase and consume more nutritiousproducts. This obviously implies comparing each nutri-tional rating system to an objective measure of dietquality and health. More research needs to be focused inthis area. It is likely that attention to developing a simple,consistent means of alerting the consumer to healthierfoods choices will continue and will likely be the focus ofincreased attention in the upcoming 2010 Dietary Guide-lines process.

Nutrition Reviews® Vol. 66(12):703–709708

Page 7: Alternative approaches to the calculation of nutrient density

CONCLUSION

There is a renewed emphasis on health and wellness inboth the public and private sectors. The overweight andobesity epidemic is a key factor influencing the develop-ment of new paradigms for promoting health and well-ness. A variety of systems have emerged to aid individualsin making healthier food choices. However, the prolifera-tion of different approaches has raised the concern thatthis may lead to consumer confusion and informationoverload.

A series of three FQS systems were developed basedon the 2005 Dietary Guidelines for Americans.11 Resultsfrom comparison of the three algorithms indicate thatwhile the absolute scores vary, the relative ranking of fooditems within a food group is invariant to the specificationof the FQS. These results are encouraging and indicatethat the rating of individual food items as being “betterfor you” is not particularly sensitive to the method ofclassification.

There is an imperative need for more research on theinfluence of food rating systems on the overall diet. Theultimate test of any approach aimed at improving foodchoices is the net effect on diet quality of individual con-sumers. The underlying science of nutrition profilingbased on nutrient density is sound. Less clear is whatinterventions are needed to effectively link the science ofrating foods to consumer food choices. This is a research-able area that requires more attention.

REFERENCES

1. Food and Drug Administration, Center for Food Safety andApplied Nutrition. How to Understand and Use the Nutri-tion Facts Label. Washington, D.C., 2000. Available at: http://www.cfsan.fda.gov/dms/foodlab. Accessed 1 September2008.

2. Food and Drug Administration. HHS, FDA Announces NewTools for the Nutrition Facts Label. FDA Consumer Maga-zine. Washington, D.C., 2007. Available at: http://www.fda.gov/fdac/features. Accessed 1 September 2008.

3. Center for Disease Control and Health Promotion, Depart-ment of Health and Human Services. Overweight andObesity. Washington, D.C., 2007. Available at: http://www.cde.gov/nccdphp/dnpc/obesity. Accessed 1 Septem-ber 2008.

4. Drewnowski A, Fulgoni V. Nutrient profiling of foods: creatinga nutrient-rich food index. Nutr Rev. 2008;66:23–39.

5. Zelman K, Kennedy E. Naturally nutrient rich . . . putting morepower on Americans’ plates. Nutr Today. 2005;40:60–67.

6. Drewnowski A. Concept of a nutritious food: toward a nutri-ent density score. Am J Clin Nutr. 2005;82:721–732.

7. Hansen RG, Wyse BW, Sorenson AW. Nutrition Quality Index ofFood. Westport, CT: AVI Publishing Co.; 1979.

8. Guthrie H. Concept of a nutritious food. J Am Diet Assoc.1977;71:14–19.

9. Sorenson AW, Hansen RG. An index of food quality. J NutrEduc. 1975;7:53–57.

10. US Department of Agriculture and Human Services. Nutritionand Your Health: Dietary Guidelines for Americans. Washing-ton, D.C.: Home and Garden Bulletin No. 232; 1980.

11. US Department of Agriculture and Human Services. DietaryGuidelines Advisory Committee Report. 2005. Available at:http://www.health.gov/DietaryGuidelines/dga2005/report/.Accessed 10 May 2007.

12. US Department of Agriculture, Center on Nutrition Policy andPromotion. My Pyramid. 2005. Available at: http://www.cnpp.usda.gov/MyPyramid. Accessed 10 May 2008.

13. Welsh S, Davis C, Shaw A. Development of the food guidepyramid. Nutr Today. 1992;27:12–23.

14. Semuels A. Nations New Food Pyramid Baffles Critics. Pitts-burgh: Post-Gazette; 2005.

15. Golberg J, Folta S. Dietary guidelines 2005 and My Pyramid.In: Kennedy E, Deckelbaum R, eds. The Nations Nutrition.Washington, D.C.: ILSI Press; 2007:221–229.

16. Kennedy E, Ohls J, Carlson S, Fleming K. The healthy eatingindex: design and application. JADA. 1995;95:1103–1108.

17. Center for Nutrition Policy and Promotion. Diet quality ofAmericans in 1994–1995 and 2001–2002 as measured by thehealthy eating index – 2005. Washington, D.C.: USDA/CNPP;2005.

18. Scheidt DM, Daniel E. Composite index for aggregating nutri-ent density using food labels: ratio of recommended torestricted food components. J Nutr Ed Behavior. 2004;36:35–39.

19. Kurtzweil P. Daily Values Encourage Healthy Eating. Avail-able at: http://www.fda.gov/fdac/special/foodlabel/dvs.Accessed 10 May 2008.

20. Eckel RH, Borra S, Lichtenstein AH, Yin-Piazza SY. Understand-ing the complexity of trans fatty acid reduction in the Ameri-can diet. American Heart Association Trans Fat Conference2006: Report of the Trans Fat Conference Planning Group.Circulation. 2007;115:2231–2246.

21. US Department of Agriculture. National Nutrient Databasefor Standard Reference, Release 18. USDA, 2005. Availableat: http://www.nal.usda.gove/fnic/foodcomp/Data/SR18/sr18.html. Accessed 10 May 2008.

22. Guthrie H. There’s no such thing as “junk food”, but there arejunk diets. Healthline. 1986;5:11–12.

23. Food Standards Agency. Taking Forward the Healthy FoodCode of Good Practice. 2008; Available at: http://www.food.gov.uk/news/newsarchive. Accessed 1 September2008.

24. Kennedy E. Food rating systems, diet quality, and health. NutrRev. 2008;66:21–22.

25. Contento I. Innovative approaches for behavior change innutrition. In: Kennedy E, Deckelbaum R, eds. The NationsNutrition. Washington D.C.: ILSI Press; 2007:187–204.

Nutrition Reviews® Vol. 66(12):703–709 709