the healthy eating index - usda

CENTER FOR NUTRITION POLICY AND PROMOTION

The HealthyEating Index

UNITED STATES DEPARTMENT OF AGRICULTURECNPP-1

Acknowledgements

Conceptual and technical assistance were provided by a working group in theUSDA Center for Nutrition Policy and Promotion consisting of Eileen Kennedy, Ex-ecutive Director, and Jay Hirschman and in the Office of Analysis and Evaluation,USDA Food and Consumer Service, consisting of Margaret Andrews, Steve Carlson,Sharon Cristofar, Patricia McKinney and Michelle Lawler. The work undertaken byJim Ohls and colleagues at Mathematica Policy Research was supported by con-tract #53-3198-038 with Food and Consumer Service. External reviewers—JeanneGoldberg, Pamela Haines, Helen Jensen and Suzanne Murphy—provided valuableassistance in the preparation of this report.

The United States Department of Agriculture (USDA) prohibits discrimination in its programs on thebasis of race, color, national origin, sex, religion, age, disability, political beliefs and marital or familialstatus. (Not all prohibited bases apply to all programs.) Persons with disabilities who require alterna-tive means for communication of program information (Braille, large print, audiotape, etc.) shouldcontact the USDA Office of Communications at (202) 720-2791.

To file a complaint, write the Secretary of Agriculture, U.S. Department of Agriculture, Washington,D.C. 20250, or call (202) 720-7327 (voice) or (202) 720-1127 (TDD). USDA is an equal employmentopportunity employer.

October 1995

The HealthyEating Index

UNITED STATES DEPARTMENT OF AGRICULTURECNPP-1

CENTER FOR NUTRITION POLICY AND PROMOTION

Nwho need it, and promoting healthy dietary choices.

Introducing the Healthy Eating Index, USDA is providing an important newtool for meeting our nutritional goals. It makes available for the first time a singlesummary measure to monitor changes in food consumption patterns.

The Healthy Eating Index measures how well the diets of all Americans con-form to the recommendations of the Dietary Guidelines for Americans and theFood Guide Pyramid. It will serve as a report card on the American diet, allowingresearchers to analyze how Americans eat, and aid USDA in developing more effec-tive nutrition promotion messages for all our programs and for the general public.

The Healthy Eating Index was developed by the USDA Center for NutritionPolicy and Promotion (CNPP) under the direction of Dr. Eileen Kennedy, in coop-eration with USDA's Food and Consumer Service and Agricultural Research Service.Within the Department, the Center serves as the focal point for linking scientificresearch to the nutritional needs of the American consumer.

Today, USDA is a recognized leader in nutrition and health. By increasing ourknowledge of what we eat and why, and by better understanding the impact offood choices on our health, we will continue to strengthen the nutritional knowl-edge of our Nation.

utrition is the bridge between agriculture and health. As the lead Federalagency for human nutrition, USDA fulfills its health responsibility throughsupport for a healthful and abundant food supply, getting food to people

The Healthy Eating Index

Foreword

Ellen HaasUndersecretary,Food, Nutrition and Consumer Services

he role of nutrition and diet in reducing the risk of certain chronic diseases,such as cardiovascular disease, diabetes mellitus, and some forms of cancer,has been well documented. The Dietary Guidelines and Food Guide PyramidT

recommend the selection of foods from a variety of food groups, the choice of adiet that is low in fat, saturated fat, and cholesterol, and moderate use of salt andsodium.

The average American diet does not meet these recommendations. TheAmerican diet is high in total fat and saturated fat and low in fiber and complexcarbohydrates. In addition, there has been no regular reporting of how wellAmericans have been doing in improving consumption patterns. Thus, USDAdeveloped the Healthy Eating Index to provide a single summary measure ofdietary quality. Prior to this, some instruments had been developed whichevaluated specific diet components such as fat and cholesterol. Few instrumentshave been developed, however, which assess overall dietary quality.

Executive Summary

The Healthy Eating Index 5

Introduction

n an effort to measure how well American diets conform to recommendedhealthy eating patterns, the U.S. Department of Agriculture (USDA) has devel-oped an index, called the Healthy Eating Index. Based on different aspects of a

healthful diet, the Index is designed to provide a measure of overall dietary quality.The Index provides a picture of foods people are eating, the amount of variety inthe diet and compliance with specific dietary guidelines recommendations. Tendietary components have been identified and are shown below.

The overall Index has a total possible score ranging from zero to 100. Each ofthe 10 dietary components has a scoring range of zero to 10. Individuals with anintake at the recommended level received a maximum score of 10 points. A scoreof zero was assigned when no foods in a particular group were eaten. Intermediatescores were calculated proportionately.

The Healthy Eating Index was applied to the 1989 and 1990 USDA data fromthe Continuing Survey of Food Intake by Individuals. The data are based on repre-sentative samples of 3,997 individuals in 1989 and 3,466 individuals in 1990.

IMethods

Other

Milk Meat

Vegetables

Grains

Fruits

6 The Healthy Eating Index

Components 1-5 measure the degree to which a person’s diet conforms to USDA’s “Food Guide Pyramid” serving recommendations for the Grains, Vegetables, Fruits, Milk, and Meat Food Groups.

Component 6 measures total fat consumption as a percentage of total food energy intake.

Component 7 measures saturated fat consumption as a percentage of total food energy intake.

Component 8 measures total cholesterol intake.

Component 9 measures total sodium intake.

Component 10 examines the amount of variety in a person’s three-day diet.

Tota

l Fat

Satu

rated

Fat

Cholesterol

Sodium

Variety

Food Guide Pyramid

he higher the score on the Healthy Eating Index, the better the diet con-formed to the Dietary Guidelines for Americans and the Food Guide Pyra-mid.

Findings T• The average Healthy Eating Index score was 63.8 and 63.9 out of a possible 100

for 1989 and 1990, respectively.• Approximately 12 percent of people had Healthy Eating Index scores above 80;

14 percent of individuals had scores below 50.• Approximately three-fourths of the people had scores on the Healthy Eating In-

dex between 51 and 80.

Individual Components of the Healthy Eating Index:

The scores in the individual component categories of the Healthy Eating Indexvaried. Out of a possible score of 10, the average scores were as follows:

Item Average Score

Grains ................................................................... 6.1Vegetables............................................................. 6.1Fruits .................................................................... 4.0Milk ...................................................................... 6.7Meat ...................................................................... 7.5Total Fat ................................................................ 6.3Saturated Fat ......................................................... 5.0Cholesterol ........................................................... 7.9Sodium ................................................................. 7.0Variety .................................................................. 7.0

Not only were a number of the average scores on the various components ofthe Healthy Eating Index low, but the percent of people meeting the recom-mended levels in any given category was also low. As shown in the table below:

Food Groups• Less than one-third of the people in the study consumed the recommended serv-

ings of milk and meat.• Fewer than one out of five people consumed the recommended servings of

grains, vegetables and fruits.

Dietary Guidelines• Less than 20 percent of the sample achieved the recommendations of the Di-

etary Guidelines for Americans for total fat and saturated fat.• Only 35 percent of the people met the recommendation for sodium.

Variety• Less than one-third of the people achieved the recommended amount of variety

in dietary patterns.


Table A

Levels of Healthy Eating Index Components1

Percent of People Receiving a Score of 10

0

100

80

60

40

20

Sodium VarietyTotal Fat Saturated Fat CholesterolVegetables Fruits Milk

17.013.8

31.9 31.919.8 19.0

67.9

35.5 32.8

aMeat Grains

9.8


Differences in the Healthy Eating Index by Individual andHousehold Characteristics:

There is an association between the Healthy Eating Index and a number of eco-nomic and demographic parameters:

Education• A person’s Healthy Eating Index improved with increasing education.

Income• Increasing income had a modest effect on improving the Index. However,

people in households below the poverty level had Healthy Eating Index scoreslower than the overall average of the sample.

Gender/age• Females tended to score higher on the Healthy Eating Index than did males. The

Healthy Eating Index scores were higher for children and older people and low-est for persons in the 15 to 39 years of age category.

Relationship of the Healthy Eating Index to Nutrient Intake

• People with higher Healthy Eating Index scores were more likely to have betternutrient intake. For example, only 49.3 percent of persons with an Index scoreof 50 or below consumed more than 75 percent of the RDA for Vitamin C. Incontrast, for persons with an Index score of 80 or above, 98.8 percent con-sumed more than 75 percent of the RDA for Vitamin C. Clearly, nutrient intakeimproved with an increasing Healthy Eating Index score.

Relationship of the Index to a Person’s Self-Rating of His or Her Diet

• A comparison of Healthy Eating Index scores and a person’s self-rating of his orher diet showed that individuals who rated their diet as “Excellent” had higheraverage scores on the Healthy Eating Index than individuals who rated their di-ets as “Good,” “Fair” or “Poor.”

aIncludes eggs, nuts, and some legumes.11989 Weighted data; Individuals Ages 2+; 3-day dataData from Continuing Survey of Food Intake by Individuals 1989 - U.S. Department of Agriculture.

• The diets of most Americans need improvement.

Only 12 percent of Americans had scores on the Healthy Eating Index that were 80or above. The average scores for the 2 years studied were similar at 63.8 and 63.9.One-third or less of the people surveyed consumed the suggested number of serv-ings from the 5 major food groups. People were most likely to underconsume inthe fruit, vegetable and grain groups. In addition, variety in the diet was limitedand intakes of total fat and saturated fat were above recommended levels for morethan 80 percent of the individuals studied.

• Some individuals are more likely to consume a poor diet.

Although the average Healthy Eating Index score for most people needs improve-ment, some individuals are at a higher risk of a low Index. Persons from low-in-come households, individuals with less education, and persons in the 15 to 39years of age category were most likely to have lower average scores on the HealthyEating Index.

• The Index provides a standard for assessing overall dietary quality.

Based on the most current scientific information available, including the DietaryGuidelines for Americans published by USDA and DHHS, and the Food GuidePyramid, the Index was developed to provide a single summary measure of di-etary quality. The Index is based on the five major food groups from the FoodGuide Pyramid and the Dietary Guidelines. The Index is a practical standard forassessing dietary quality.

The Index correlates well with other conventional measures of diet quality.Comparisons with RDA levels confirm a positive correlation between the Indexand individual nutrient intake levels. Higher Index scores are associated with im-proved nutrient intakes.

• The Healthy Eating Index reflects the complexity of dietary patterns.

Ten dietary components comprise the Index. All 10 components contribute evenlyto the overall Index score. Doing well on one component does not ensure a highscore on the overall Index. Overall dietary quality is reflected in the total Indexscore and is not determined based on any individual component score. Using onecomponent score, such as percent of calories from fat, as an indicator of dietaryquality can result in misclassifications.

There is significant variation in average scores among the individual Index com-ponents. Fruits and saturated fat have the lowest average component scores, indi-cating that consumers are doing the poorest in these areas. Of all individuals, lessthan 20 percent achieved recommended levels in the grains, fruits, vegetables, to-tal fat, and saturated fat components. Only in the cholesterol component didgreater than 50 percent of individuals achieve a perfect score.

Conclusions


Practical applications of the Healthy Eating Index results are identified below.

• The results of the Index are useful in targeting nutrition educa-tion and health promotion activities.

Results of the Index provide insights into the types of dietary changes needed toimprove American eating patterns. A two-tiered approach is warranted. First, nutri-tion promotion activities are required to address the nutritional needs of all Ameri-cans. To that end, USDA is proposing a general nutrition education and nutritionpromotion initiative for all Americans as part of the 1995 Farm Bill. In addition, therecently implemented Schools Meals Initiative for Healthy Children ensures thatthe nutrition standards for school meals meet the dietary guidelines. This iscomplemented by Team Nutrition which focuses on empowering children to makefood choices for healthful diets.

Targeted strategies for nutrition promotion are also needed. Results from thisresearch suggest that individuals from low-income households and less-educatedpeople are more likely to score lower on the Healthy Eating Index. Therefore, ef-forts are already underway within USDA to integrate nutrition into all of the foodassistance programs.

• The Healthy Eating Index is a single summary measure of dietquality that can be used to monitor changes in food consumptionpatterns over time.

In this study, the Index was applied to the 1989 and 1990 CSFII data to evaluatethe overall quality of American diets. Average scores for the overall Index for bothyears are approximately 64 percent, a score judged as “Needs Improvement." HEIvalues are similar for both years, indicating that dietary intake does not vary greatlyfrom year to year. USDA intends to use the Index as one method to monitorchanges in dietary patterns in the United States population over time. The Indexwill be periodically published as nationally representative dietary survey data be-come available.

• The Index could provide the basis for development of a variety ofadditional tools.

The Healthy Eating Index provides one instrument that will be useful in monitor-ing trends in U.S. consumption patterns over time. This will provide policymakerswith the capability of revising and fine-tuning specific programs in a more timelymanner to be responsive to the changing nutrition profile of the population.

In addition, a self-assessment instrument would be of use to the American con-sumer in that it would provide a standard against which an individual could evalu-ate the quality of his/her own diet. In addition, a self-assessment instrument couldserve as a useful guide to consumers in helping them select the types and amountsof foods needed to achieve recommended intakes. The Index would provide abasis for the development of such an instrument.

It is the intention of the USDA Center for Nutrition Policy and Promotion tobegin developing a consumer-oriented, self-assessment guide following the publicrelease of the Healthy Eating Index.

USDA Applications


Table ofContents

Development of theHealthy Eating Index

Excutive Summary

Study Results

Conclusions

Technical Issues

Page

Introduction 12

Methods 13

Food Group Components of theHealthy Eating Index 15

Other Components of theHealthy Eating Index 17

Data Used To Calculate theHealthy Eating Index 18

Average Overall Scores 19

Component Scores 21

Differences by Personal andHousehold Characteristics 23

Correlations with Meeting RDAs 25

Other Tabulations 27

Comparison of Index Score andIndividual’s Self-Rating of Diet 28

Implications 29

USDA Applications 30

31

5


Introduction

Development of the Healthy Eating Index

he role of nutrition and diet in re-ducing the risk of chronicdiseases, such as cardiovascular

disease, diabetes mellitus, and certainforms of cancer, has been well docu-mented. Concerns about the influenceof dietary practices on the health statusof the country have been increasinglyemphasized in a number of reports andguidelines, among them the DietaryGuidelines for Americans (U.S. Depart-ment of Agriculture and U.S. Depart-ment of Health and Human Services,1980, 1985, 1990), Healthy People2000: National Health Promotion andDisease Prevention Objectives (U.S.Department of Health and Human Ser-vices, 1990), Surgeon General’s Reporton Nutrition and Health (1988), andNational Research Council’s report onDiet and Health (1989).

T
Recommended dietary practicesgenerally include the selection of foodsfrom a variety of food groups, particu-larly the grain, fruit, and vegetablegroups; choice of a diet that is low infat, saturated fat, and cholesterol; andmoderate use of salt and sodium.
Much of the previous research hasfocused on determining which eatingpatterns should be recommended tothe public for reducing the risk ofchronic disease. Future research needsto be directed at developing measuresto assess overall dietary quality. Someanalytical instruments have been devel-oped which evaluate specific dietarycomponents, such as fat and choles-terol. Few instruments have been de-veloped, however, which assess theoverall quality of a diet.


Methods
n an effort to measure how wellAmerican diets conform to recom-mended healthful eating patterns,I
the U.S. Department of Agriculture hasdeveloped an Index, called the HealthyEating Index. The Index was designedto provide a measure of the overallquality of an individual's diet.

The Index provides a picture offoods people are eating, the amount ofvariety in the diet and compliance withspecific Dietary Guidelines recommen-dations. In designing the Index, it wasnot intended that the dietary compo-nents be considered individually as in-dicators of overall dietary quality. Ascore on the Index reflects a sum totalof the dietary components. The follow-ing 10 dietary components were in-cluded in the Index based on differentaspects of a healthful diet (figure 1):• Components 1 - 5 measure the de-

gree to which a person’s diet con-forms to the Food Guide Pyramid

serving recommendations for theGrain, Vegetable, Fruit, Milk, andMeata Groups.

• Component 6 measures total fat con-sumption as a percentage of totalfood energy intake.

• Component 7 measures saturated fatconsumption as a percentage of totalfood energy intake.

• Component 8 measures total choles-terol intake.

• Component 9 measures total sodiumintake.

• Component 10 examines the amountof variety in a person’s diet over a 3-day period.

The Index is a practical approachto measuring overall dietary quality inthat it examines specific food behaviorswhich comprise dietary intake.

aIncludes eggs, nuts and some legumes.


Figure 1

Components of Healthy Eating Index

Food Group Range of Scores Perfect Score of 101

1. Grains 0 to 10 6 - 11 servings

2. Vegetables 0 to 10 3 - 5 servings

3. Fruits 0 to 10 2 - 4 servings

4. Milk 0 to 10 2 - 3 servings

5. Meat 0 to 10 2 - 3 servings

Dietary Guidelines

6. Total fat 0 to 10 30% or lessenergy from fat

7. Saturated fat 0 to 10 Less than 10%energy fromsaturated fat

8. Cholesterol 0 to 10 300 mg. or less

9. Sodium 0 to 10 2400 mg.or less

10. Variety 0 to 10 16 different fooditems over 3-dayperiod

1Depends on recommended Energy Intake - See Table 1; all amounts listed are based on a per day basis with the exception of food variety.


Food GroupComponents of theHealthy Eating Index

Some people needmore servings fromthe food groupsthan others.

TPyramid. A range of servings is shownfor the Grain Group, Vegetable Group,Fruit Group, Milk Group, and MeatGroup. The number of recommendedservings depends upon an individual’scaloric requirements. Recommendedservings for calorie levels of 1,600,2,200, and 2,800 are presented inTable 1.

For each of the five food groupcomponents of the Index, individualswho consumed the recommendednumber of servings received a maxi-mum score of 10. A score of zero wasassigned to any food group where noitems from that category were con-sumed. Intermediate scores were calcu-lated proportionately to the number ofservings consumed. For example, if therecommended level of servings waseight and an individual consumed fourservings, the component score for theindividual was 5 points. A score of 7.5points was assigned if six servings wereeaten.

In developing the Index, servingrecommendations from the FoodGuide Pyramid were interpolated toindividuals with other food energy re-

he Healthy Eating Index examinesdietary intake in relation to the fivemajor groups in the Food Guide

Number o

Table 1

Recommended Numbers of Servings Perin the Food Guide Pyramid Bulletin (USD

Kilocalories Grains Vegetable

1600 6 3

2200 9 4

2800 11 5

quirements. For example, food energyRDAs for children between 1 and 3years of age are less than 1,600 kilo-calories. The recommended number ofservings was retained at the minimumserving level for these children, but theserving size was scaled down to be pro-portionate with their energy require-ments. This approach was consistentwith the guidance contained in theFood Guide Pyramid. In contrast,adult males between the ages of 15 and50 have food energy RDAs slightlygreater than 2,800 kilocalories. The textof the Food Guide Pyramid providesno guidance as to how to adjust servingsizes to meet calorie levels above 2,800kilocalories. Instead of slightly increas-ing serving sizes, it was decided thatfood portions for these individualswould be truncated at the maximumlevels recommended in the Food GuidePyramid. It should be noted, based onpreliminary analysis, none of the resultsfrom the Index were shown to be sig-nificantly affected even if a slightlylarger serving size were used.

Table 2 shows the serving recom-mendations for various age/gender cat-egories.

Servingsf Servings

Day at Food Energy Levels DiscussedA, 1992)

s Fruits Milk Meat

2 2 2

3 2 2.4

4 2 2.8


Table 2

Recommended Number of Servings Per Day for Age/Gender Categories

Age/GenderCategory Kilocalories Grains Vegetables Fruits Milk Meat

Children 1-3 1300 6a 3a 2a 2a 2a

• 1600 6 3 2 2 2

Children 4-6 1800 7 3.3 2.3 2 2.1

Females 51+ 1900 7.4 3.5 2.5 2 2.2

Children 7-10 2000 7.8 3.7 2.7 2 2.3

Females 11-50 2200 9 4 3 2b 2.4

Males 51+ 2300 9.1 4.2 3.2 2 2.5

Males 11-14 2500 9.9 4.5 3.5 3 2.6

• 2800 11 5 4 2 2.8

Males 19-50 2900 11 5 4 2b 2.8

Males 15-18 3000 11 5 4 2 2.8

aPortion sizes are reduced for children age 1-3.bIs 3 for persons age 11 to 24.•RDA levels included in the Food Guide Pyramid bulletin.


Other Components ofthe Healthy EatingIndex

Fat and Saturated Fat

Index scores for fat and saturated fat in-takes were examined in proportion tototal food energy (or calories).• Fat intakes less than or equal to 30

percent of the total calories were as-signed a score of 10 points. Thescore declined to zero when the pro-portion of fat to total caloriesreached 45 percent. Intakes between30 percent and 45 percent werescored proportionately.

• A score of 10 points was assigned tosaturated fat intakes at less than 10percent of total calories. Zero pointswere assigned when the saturated fatintake reached a level of 15 percentof the total calories. Scores betweenthe two cutoff values were calcu-lated proportionately.

Cholesterol and Sodium

The scores for cholesterol and sodiumwere each based on milligrams con-sumed.• A maximum point value for choles-

terol was assigned when intake wasat a level of 300 milligrams or less.Zero points were assigned when in-take reached a level of 450 milli-grams or more. Values between thetwo cutoff points were scored pro-portionately.

• A maximum score for sodium was as-signed at an intake level of 2,400 mil-ligrams or less. Zero points wereassigned at a level of 4,800 milli-grams or more. Scores between thetwo levels of intake were scored pro-portionately.

Variety

Dietary variety was assessed by totalingthe number of “different” foods eatenby an individual in amounts sufficientto contribute at least one-half of a serv-ing in a food group. Similar foods weregrouped together and counted onlyonce in measuring variety. Food mix-tures were broken down into theircomponent ingredients and assigned tothe relevant food groups. Index scoresfor variety were calculated in a manneranalogous to the method used for theother Index components. Cutoff scoresfor variety were defined based on 3days of recorded data. A maximumscore was given if 16 or more differentfood items were consumed over a 3-dayperiod. A score of zero was given if sixor less different items were eaten.When based on 1 day of reported data,the cutoff scores for variety were re-duced by a factor of two. Intermediateintakes were calculated proportion-ately.


Study Results

Data Used To Calculatethe Healthy EatingIndex

consumption of foods and nutrients,and the dietary status of individuals. Inaddition to dietary intake information,the CSFII contains extensive data aboutpersonal and socio-economic character-istics and knowledge of, and attitudestoward, healthful eating practices. Thesurvey sample included individuals inhouseholds with incomes at all levels.Three days of dietary intake data werecollected. Dietary recall informationwas collected on the first day during anin-person interview. Respondents com-pleted a food record for the second and

he Continuing Survey of Food In-takes by Individuals (CSFII) pro-vides information on the Nation’sT
third days. Although lower income
groups were oversampled in conduct-ing the surveys, weights were availablefor making the sample representative ofthe U.S. population.

The Index was applied to the USDA1989 and 1990 CSFII. Most of the analy-sis conducted focused on individuals 2years of age and older for whom 3 daysof dietary intake data were available.Sample sizes were approximately 4,000for the 1989 survey and 3,400 for the1990 survey.


Average Overall Scores

The Average Scoreon the HealthyEating Index issimilar for 1989and 1990. Mostpeople fall in themiddle rangebetween 51 and80.

he maximum overall score for the10 components of the Index is100 points. Average overall scores

on the Index for the 1989 and 1990CSFII are 63.8 and 63.9, respectively.The values of the Index are quite simi-lar for both years with the majority of

T

Table 3

Distribution of Individuals by Healthy Ea

Level of Index 198

0-30 a

31-40 2

41-50 12

51-60 26

61-70 28

71-80 22

81-90 10

91-100 1

Mean 63

Sample Size 399

a Less than 0.5 percent.1Weighted data, composed of individuals ages 2 and older;2Data from Continuing Survey of Food Intakes by Individua*May not add to 100% due to rounding

scores ranging between 51 and 80. Ap-proximately 12 percent of personshave index scores above 80 while 14 to15 percent have index scores of 50 orbelow. The distribution of individualsby Index scores is presented in Table 3.

ting Index Level, 1989, 19901, 2

9* 1990

a

% 3%

% 12%

% 23%

% 29%

% 21%

% 10%

% 2%

.8 63.9

7 3400

3-day datals, 1989, 1990 U.S. Department of Agriculture


The majority ofpeople fall in the“Needs Improve-ment” category.Smaller percentagesfall in the "Good" or"Poor" categories.

In an effort to provide a “rating” ofthe overall American diet, a gradingscale was developed. The grading scaleused to classify Index scores is pre-sented in Table 4 along with the per-cent of individuals receiving aparticular rating.

Table 4

Healthy Eating Index Grading Scale

HEI Scores Rati

Greater than 80 Goo

51 - 80 Needs Impr

Less than 51 Poo

The majority of individuals havediets rated as “Needs Improvement.”Eleven to 12 percent of individualshave diets rated as "Good" whereas 14to 15 percent have diets rated as "Poor."

ng Percent of Sample

d 11 - 12

ovement 73 - 76

r 14 - 15


Component Scores

Table 5a

Levels of Healthy Eating I

aIncludes eggs, nuts, and some legumes.11989 Weighted data; Individuals Ages 2+Data from Continuing Survey of Food Inta

People do worst onthe consumption ofFruits and SaturatedFat. These are theareas that needmost improvement.

Percent of People Receiving a

0

100

80

60

40

20

Vegetables

17.0

Grains

9.8

Percent of People Receiving a

here is significant variation in aver-age scores among the individualcomponents of the Index. No oneT

category contributes disproportion-ately to the average score. Some keypatterns across Index components be-come apparent in examining the per-cent of observations at maximumscores of 10. Mean component scoresand percent observations at the maxi-mum scores are presented in Tables5a and 5b.

ndex Components1

; 3-day datake by Individuals 1989 - U.S. Department of Agriculture.

Perfect Score of 10

Total Fat SatuFruits Milk

13.8

31.9 31.9

19.8

aMeat

Score of 10

High component scores are indica-tive of intakes within recommendedranges. The lower the componentscore, the poorer the conformance torecommended intake levels. Meanscores for the different componentsrange from 4.0 to 7.9. Fruits (4.0) andsaturated fat (5.0) had the lowest meancomponent scores, indicating they arethe areas needing greater improve-ment. The highest mean componentscores were for cholesterol (7.9) andmeats (7.5). Sodium (7.0) and dietaryvariety (7.0) also had a relatively highproportion of people at the maximum.

Sodium Varietyrated Fat Cholesterol

19.0

67.9

35.5 32.8


Table 5b

Levels of Healthy Eating Index Components1

aIncludes eggs, nuts, and some legumes.11989 Weighted data; Individuals Ages 2+; 3-day dataData from Continuing Survey of Food Intake by Individuals 1989 - U.S. Department of Agriculture.

People's Average ScoreRepresentative Sample of American Population

6.1 6.1

4.0

6.77.5

6.3

7.97.0 7.0

Sodium VarietyTotal Fat CholesterolGrains Vegetables Fruits Milk aMeat Saturated Fat0

10

8

6

4

2

5.0


Differences by Personaland HouseholdCharacteristics

levels.

People's HealthyEating Index scoresimprove with in-creasing education.Increasing incomehas a more modesteffect on the overallscore.

he Index scores for weighted 1989and 1990 CSFII data were strati-fied by selected indicators. Table 6

summarizes the mean Index scores forthe five indicators analyzed (sex, age,head of household, educational level,and income as a percent of poverty).

Index scores by indicator weresimilar for 1989 and 1990. Femaleshave a higher average Index thanmales. Persons in the younger andolder age groupings have a higher In-dex than those in the middle agegroupings. Individuals living in eitherjointly headed or female-headed house-

T
holds have a higher Index than indi-viduals living in male-headed house-holds. Index scores generally increasewith increasing levels of education. Av-erage Index scores were highest for in-dividuals having 4 or more years ofcollege education. The Index generallyresponds more to increases in educa-tion than increases in income. Overall,the average Index was higher for indi-viduals at 201 percent of poverty orover than for those at lower poverty


Table 6

Mean Dietary Index Weighted 1989 and 1990

Eating Index 1989 1990Score Mean Mean

<2 Yrs. High School 60.9 59.4

2-3 Yrs. High School 59.7 60.5

4 yrs. High School 62.3 60.9

Some College 63.9 65.0

4 Yrs. College 65.8 66.3

>4 Yrs. College 68.2 68.0

Indeterminable 67.1 66.7

Income as a % of Poverty

0-50% 59.5 61.7

51-100% 61.2 59.7

101-130% 62.0 63.9

131-200% 62.0 62.4

201-300% 63.3 63.9

301% plus 65.4 65.1

Eating Index 1989 1990Score Mean Mean

Sex

Male 62.0 60.0

Female 65.6 65.0

Age

5-14 66.1 66.4

15-39 60.5 60.2

40-64 63.5 64.2

65+ 69.2 69.0

Head of Household

Male & Female Head 64.1 64.0

Female Head 64.5 64.9

Male Head 58.8 60.5

Education Level


Correlations withMeeting RecommendedDietary Allowances

he Index was compared to nutri-ent intake as a percent of theRDAs to assess the degree toT

which it correlates with other conven-tional measures of diet quality. In thatthe RDAs are set to be higher than thelevel needed by most individuals, thecriterion of meeting 75 percent of anRDA was selected. Table 7 shows the re-lationship between key nutrients andscores on the Index.

A higher Index score is associatedwith an increased likelihood that atleast 75 percent of the RDA for mostnutrients will be met. Nutrient intakeimproves as the Index score improves.For example, only 49 percent of the in-dividuals with Index scores 50 or lesshave Vitamin C intakes greater than 75percent of the RDA. With an Indexscore between 71 and 80, the likeli-hood increases to 90 percent. With In-

dex scores of 80 or above, the likeli-hood increases to nearly 99 percent.

Index tabulations were also per-formed using a criterion of meeting100 percent of the RDA. Results are es-sentially the same as those for meeting75 percent of the RDA.

Correlation coefficients provide astatistical measure of the Index’s abilityto rank individuals along a distributionof high intake to low intake. The corre-lation coefficients between the Indexand nutrient intake levels confirmed apositive relation for each of the nutri-ents analyzed. Correlation coefficientswere in the range of 0.30 to 0.40 for 5of the 17 nutrients analyzed. A modestcorrelation (0.18) exists between foodenergy and the Index, of indicating thatan increase in the calorie level will notby itself dramatically improve theIndex.


People who have abetter HealthyEating Index scoreare more likely tohave a betternutrient intake.

Table 7

Percent of Observations Meeting 75 Percent or More of RDAs by HEI Levels

CorrelationIndex Score Coefficients

of IndexNutrient 0-50 51-60 61-70 71-80 81-100 with Nutrient

Food Energy 35.6 47.7 47.0 55.3 63.1 .18

Protein 86.3 90.3 94.0 98.1 99.7 .18

Vitamin A - IU 41.1 54.7 65.2 83.1 90.3 .22

Vitamin A - RE 37.3 50.0 58.9 74.8 86.1 .21

Vitamin E 34.5 44.2 45.8 51.0 64.5 .12

Vitamin C 49.3 67.7 82.3 90.5 98.8 .39

Thiamin 62.9 81.7 91.1 96.1 96.9 .30

Riboflavin 72.6 83.2 86.0 93.7 98.7 .23

Niacin 74.7 84.9 93.9 97.0 98.4 .29

Vitamin B6 35.9 50.7 67.8 80.4 91.4 .36

Folate 58.1 74.3 87.8 93.0 98.0 .38

Vitamin B12 87.5 93.1 93.8 95.6 98.1 .03

Calcium 35.8 51.4 50.7 65.4 66.0 .12

Phosphorus 75.5 83.7 89.5 95.7 98.2 .10

Magnesium 30.3 45.3 58.7 72.2 90.5 .40

Iron 59.4 67.2 75.2 82.8 87.4 .17

Zinc 43.9 43.5 42.8 51.5 51.4 .04


Other Tabulations

Similar results arefound with 1-dayand 3-day dietaryintake data. This isencouraging sincemost surveys collectonly 1 day’s dietaryinformation.

Comparison of IndexScores for 1 Day Versus3 Days of Dietary Intake

Many dietary surveys include only oneday of food consumption data. To testits use on a single day of data, the In-dex was applied to the first day of the3-day sample. With the exception of va-riety, the parameters used on the 1-daydata set were the same as those usedon the 3-day data set. It is not reason-able to expect that the number offoods eaten in 1 day would equal thenumber of foods eaten over a 3-day pe-riod. In computing the variety compo-nent for the 1-day tabulations, thenumber of foods needed to receive themaximum score was reduced by a fac-tor of two, from 16 foods to 8 foods.

Table 8

One-Day Versus Three-Day Index Score

Index Component Three-Day M

Grains 6.1

Vegetables 6.1

Fruits 4.0

Milk 6.7

Meat 7.5

Total Fat 6.3

Saturated Fat 5.0

Cholesterol 7.9

Sodium 7.0

Variety 7.0

Total 63.8

1Weighted data, individuals ages 2 years and over, CSFII 1989*May not add due to rounding

A comparison of 1-day versus 3-dayIndex scores is presented in Table 8.The pattern of results for 1 day of in-take is similar to that for 3 days of in-take, although most of the componentscores are somewhat lower for a 1-dayintake. Overall, the average value fallsfrom 63.8 to 61.4. Consistent with the 3-day findings, fruits and saturated fat havethe lowest component scores, followedby grains, vegetables, and total fat.

Daily variations in an individual’s di-etary intake provide one possible ex-planation for the higher scores in the3-day data compared to the 1-day data.If an individual has an exceptionallyhigh component score on 1 of the 3days analyzed, this high score is aver-aged with the component scores forthe other 2 days.

s, 19891

ean* One-Day Mean*

6.1

5.9

3.7

6.2

7.1

6.3

5.4

7.5

6.7

6.6

61.4


Individuals who self-rate their diets as"Excellent" have ahigher Index scorethan individuals whorate their diets as"Fair" or "Poor."

Comparison of Index Score andIndividual’s Self-Rating of Diet

In addition to dietary intake infor-mation, the CSFII contains informationon the self-rating of dietary patterns.The Index score was compared to anindividual’s rating of his or her diet toexamine how accurately he or she as-sesses dietary quality. A comparison ofIndex scores and self-ratings of diet forweighted CSFII 1990 data is shown inTable 9. Results were similar for CSFII1989 data.

Individuals who rate their diet as"Excellent" have a higher average Index

Table 9

Healthy Eating Index Score Compared toWeighted 1990 Data

Self-Rating Sample HEI <

% MEAN %

Excellent 16.3 67.6 2

Very Good 29.2 65.4 2

Good 39.0 63.2 3

Fair 11.9 59.9 3

Poor 3.1 55.8 11

score than persons who rate their dietas "Very Good," "Good," "Fair," or"Poor." Those individuals rating their di-ets as "Excellent" are also less likelythan persons in the other categories tohave an Index score below 40. Simi-larly, individuals who rate their diet as"Poor" are more likely to have an Indexscore below 40 and less likely to havean Index score above 80. These resultsindicate that individuals are fairly accu-rate in assessing the quality of their di-ets and that factors other thanknowledge affect good eating habits.

Individual’s Self-Rating of Diet,

HEI SCORE40 40- 60 61-80 81-100

% % %

.2 26.9 52.3 18.7

.2 32.0 50.7 15.1

.1 35.9 51.5 9.5

.7 49.1 42.9 4.3

.6 43.8 43.4 1.2


Conclusions

Implications • The diets of most Americans need improvement.

Only 12 percent of Americans had scores on the Healthy Eating Index that were 80or above. The average scores for the 2 years studied were similar at 63.8 and 63.9.One-third or less of the people surveyed consumed the suggested number of serv-ings from the 5 major food groups. People were most likely to underconsume inthe fruit, vegetable and grain groups. In addition, variety in the diet was limitedand intakes of total fat and saturated fat were above recommended levels for morethan 80 percent of the individuals studied.

• Some individuals are more likely than others to consume a poordiet.

Although the average Healthy Eating Index score for most people needs improve-ment, some individuals are at a higher risk of a low Index. Persons from low-in-come households, individuals with less education, and persons in the 15 to 39 yearage category were most likely to have lower average scores on the Healthy EatingIndex.

• The Index provides a standard for assessing overall dietary quality.

Based on the most current scientific information available, including the DietaryGuidelines for Americans published by USDA and DHHS, and the Food GuidePyramid, the Index was developed to provide a single summary measure of di-etary quality. The Index is based on the five major food groups from the FoodGuide Pyramid and the Dietary Guidelines. The Index is a practical standard forassessing dietary quality.

The Index correlates well with other conventional measures of diet quality.Comparisons with RDA levels confirm a positive correlation between the Indexand individual nutrient intake levels. Higher Index scores are associated with im-proved nutrient intakes.

• The Healthy Eating Index reflects the complexity of dietary patterns.

Ten dietary components comprise the Index. All 10 components contribute evenlyto the overall Index score. Doing well on one component does not ensure a highscore on the overall Index. Overall dietary quality is reflected in the total Indexscore and is not determined based on any individual component score. Using onecomponent score, such as percent of calories from fat, as an indicator of dietaryquality can result in misclassifications.

There is significant variation in average scores among the individual Index com-ponents. Fruits and saturated fat have the lowest average component scores, indi-cating that consumers are doing the poorest in these areas. Of all individuals, lessthan 20 percent achieved recommended levels in the grains, fruits, vegetables, to-tal fat, and saturated fat components. Only in the cholesterol component didgreater than 50 percent of individuals achieve a perfect score.


Practical applications of the Healthy Eating Index results are identified below.

• The results of the Index are useful in targeting nutrition educa-tion and health promotion activities.

Results of the Index provide insights into the types of dietary changes needed toimprove American eating patterns. A two-tiered approach is warranted. First, nutri-tion promotion activities are required to address the nutritional needs of all Ameri-cans. To that end, USDA is proposing a general nutrition education and nutritionpromotion initiative for all Americans as part of the 1995 Farm Bill. In addition, therecently implemented Schools Meals Initiative for Healthy Children ensures thatthe nutrition standards for school meals meet the Dietary Guidelines. This iscomplemented by Team Nutrition which focuses on empowering children to makefood choices for healthful diets.

Targeted strategies for nutrition promotion are also needed. Results from thisresearch suggest that individuals from low-income households and less-educatedpeople are more likely to score lower on the Healthy Eating Index. Therefore, ef-forts are already underway within USDA to integrate nutrition into all of the foodassistance programs.

• The Healthy Eating Index is a single summary measure of dietquality that can be used to monitor changes in food consumptionpatterns over time.

In this study, the Index was applied to the 1989 and 1990 CSFII data to evaluatethe overall quality of American diets. Average scores for the overall Index for bothyears are approximately 64 percent, a score judged as “Needs Improvement." HEIvalues are similar for both years, indicating that dietary intake does not vary greatlyfrom year to year. USDA intends to use the Index as one method to monitorchanges in dietary patterns in the United States population over time. The Indexwill be periodically published as nationally representative dietary survey data be-come available.

• The Index could provide the basis for development of a variety ofadditional tools.

The Healthy Eating Index provides one instrument that will be useful in monitor-ing trends in U.S. consumption patterns over time. This will provide policymakerswith the capability of revising and fine-tuning specific programs in a more timelymanner to be responsive to the changing nutrition profile of the population.

In addition, a self-assessment instrument would be of use to the American con-sumer in that it would provide a standard against which an individual could evalu-ate the quality of his/her own diet. In addition, a self-assessment instrument couldserve as a useful guide to consumers in helping them select the types and amountsof foods needed to achieve recommended intakes. The Index would provide abasis for the development of such an instrument.

It is the intention of the USDA Center for Nutrition Policy and Promotion tobegin developing a consumer-oriented, self-assessment guide following the publicrelease of the Healthy Eating Index.

USDA Applications


Technical Issues

The following technical issues were considered critical to the developmentof the Healthy Eating Index:

• Determination of Portion Sizes• Allocation of “Mixtures” to Individual Food Groups• Estimation of Food Group Serving Requirements by Age and

Gender• Coding Structure Used To Compute the Index Variety Component• Design Alternatives: What To Count• Other Components

Determination of Portion Sizes

Serving sizes used to compute the In-dex scores were intended to be as con-sistent as possible with those presentedin the Food Guide Pyramid. Identifica-tion of the key underlying commoditiescontained in various foods and determi-nation of appropriate serving size con-version factors provided a basis for theapproach used.

Serving amounts for breads andgrains were determined according toan “equivalent flour” approach. For ex-ample, the Food Guide Pyramid desig-nates a slice of bread as one serving. Atypical slice of bread contains 17 gramsof flour. The number of servings for anybread on the Index was calculatedbased on the number of grams of flourit contained divided by the number ofgrams of flour contained in a slice ofbread (17). Similar approaches were ap-plied to pastas and cereal grains.

The Food Guide Pyramid counts1/2 cup of most cooked vegetables, 1cup of most raw leafy vegetables, and1/2 cup of most raw nonleafy choppedvegetables as single servings. Differentvegetables have different densities, re-sulting in different gram weights percup or half-cup measures. Differentgram/serving size factors were used tocalculate the index for most vegetables,based on the weight of a cup or half-

cup of the relevant commodities.Fruits were treated similarly to veg-

etables. Gram/serving size factors weredeveloped for each fruit based on theweights of the various fruit amounts.

Serving amounts for various kindsof milk and milk products were calcu-lated based on the grams of nonfat milksolids contained in a food divided bythe amount of grams of nonfat milk sol-ids contained in 1 cup of milk (theserving size specified for milk in theFood Guide Pyramid). To determine aserving size for different cheeses, theweight of all milk products in a cheesewas totaled and then divided by theconversion factor used in the FoodGuide Pyramid for cheese.

Serving sizes of meats are specifiedin the Food Guide Pyramid in terms of2 - 3 ounces of lean meat. The Indexcalculations assume a serving size of2.5 ounces for meats. Serving size con-version factors for meats were basedon the amount of fat-free meat com-modity included in the database forvarious foods. Conversion factors forconverting grams of nuts and peanutbutter to serving sizes were developedaccording to those specified in theFood Guide Pyramid. The gram conver-sion factors developed were based onthe weight of these quantities.


Allocation of Mixtures toIndividual Food Groups

In calculating the Index, it was neces-sary to assign the foods in “mixtures” totheir constituent food groups in the ap-propriate amounts. Pizza, for instance,may make significant contributions toseveral different food groups, includinggrains, vegetables, milk, and meat.

The approach used was a straight-forward extension of the approach usedto estimate serving sizes. Commoditycompositions of foods were identified.Once identified, commodities were as-signed to appropriate food groups,based on the gram/serving size factorcalculated.

Estimation of Food GroupServing Requirements by Ageand Gender

Prior to scoring the first five compo-nents of the Index, it was necessary todetermine the recommended numbersof servings by food group for each ofthe individuals on the data file. The foodenergy RDAs for some age/gender com-binations were different from the threelevels of energy intake presented in theFood Guide Pyramid. Interpolationtechniques were used to estimate the re-quired number of servings for age/gen-der combinations not addressed in theFood Guide Pyramid. Food servingsspecified in the Food Guide Pyramidfor three food energy RDA levels wereused as a basis for predicting compa-rable food servings at other food energylevels for each food group.

Two issues arose in taking this ap-proach. Children 1 - 3 years of age havea food energy RDA less than the lowestcalorie level in the Food Guide Pyra-mid. Extrapolation of the Food GuidePyramid’s recommended number ofservings to a lower calorie level wouldresult in smaller numbers of servingsthan the minimums shown. The follow-ing statement from the Food GuidePyramid provided a basis for the techni-cal approach taken to address the issue:

Preschool children need the samevariety of foods as older family mem-bers do, but may need less than 1,600calories. For fewer calories, they can eatsmaller servings.

The number of servings for chil-dren 1 - 3 years of age was thereby heldconstant at the minimums shown inthe Pyramid, but the serving sizeswere reduced proportionately.

Similarly, males 15 - 50 years of agehave food energy RDAs slightly higherthan the highest calorie level in theFood Guide Pyramid. Simple extrapola-tion would result in larger numbers ofservings than the maximum numbersshown. The text of the Food GuidePyramid provides no guidance regard-ing the adjustment of numbers of serv-ings or serving sizes to accommodatehigher food energy levels. Rather thanexceeding recommended serving sizes,it was decided that food servingswould be truncated at the maximumsshown in the Food Guide Pyramid.Preliminary analysis indicated that ifserving sizes had been slightly in-creased, the results obtained from theIndex would not have been signifi-cantly different.

Coding Structure Used ToCompute the VarietyComponent of the HEI

The food coding structure used to com-pute the Index was based on USDA’scoding structure for data in the 1989and 1990 CSFII. In an effort to simplifycoding, items which were similar butcoded separately in CSFII weregrouped together for the purposes ofthis study. The following principlesguided the coding decisions made:• Only foods judged nutritionally simi-

lar were grouped together.• Foods made with separate commodi-

ties were generally grouped sepa-rately.


• Foods differing only in fat contentwere generally grouped together.

• Vegetables were each given separatecodes, but different forms of thesame vegetable were coded together.

• Different forms of the same meatwere generally coded the same. Or-gan meats and ham were two excep-tions.

• Each type of fish was given a sepa-rate code, but different forms of thesame fish received the same code.

• Most forms of fluid milk had thesame code. Pudding was coded sepa-rately from milk.

• Most cheeses had the same codewith the exception of cottagecheese.

• All white bread was given the samecode. Sweet rolls and pasta receiveddifferent codes.

• Whole wheat products were codeddifferently from products made withrefined wheat flour.

• Ready-to-eat cereals made principallyfrom the same grain received thesame codes; those made from differ-ent grains received different codes.

In an effort to facilitate coding, foodmixtures were broken down to theirconstituent components. Only compo-nent foods present in substantial quan-tities were included in the varietycalculations. A threshold of 1/2 a FoodGuide Pyramid serving was used. Foodcomponents contributing less than 1/2a Food Guide Pyramid serving werenot computed in the Index varietyscore. It is possible that variety scoresfor some individuals were slightly un-derestimated by this approach. Severalservings in a food group consumed inless than 1/2 serving amounts through-out a day could exceed the 1/2 servingthreshold when added together. The po-tential effects of this underestimationon the final results is believed to be ex-tremely small.

A second simplifying conventionused was to assume that food mixturescontaining two or more components

from the same food group (e.g., mixedvegetables) could be reasonably allo-cated, equally, to two codes represent-ing the components present in thehighest proportions.

Design Alternatives:What To Count

Foods often principally fall within onegroup but contain small amounts ofother groups. Bread, for example, ismainly a grain but also contains smallamounts of milk and egg products. Anapproach considered for this study wasto exclude the “incidental” food groupcontributions from the computations.This approach was rejected based onthe following rationale:1) Even relatively small amounts of inci-

dental foods contribute to anindividual’s overall nutrient intake.

2) Suppressing the incidental foodswould have often involved arbitraryjudgements for establishing mini-mum size cutoffs.

It was decided that all contribu-tions to various food groups would becounted in computing the Index withno minimum size cutoff values im-posed. The following examples illus-trate some of the implications of thisapproach:• The nutrition value from condi-

ments, such as mayonnaise, wascounted in computing the Index.

• The nutrition value of milk used insome sweets, such as a milk choco-late bar, was counted in the milkgroup in computing the Index. If al-located to a single food group, theoverall food would be assigned tothe “Sweets” group and not countedin the Index.

• The fruit juice in a soft drink whichis at least 10 percent fruit juice wascounted in computing the Index. Thewater and sugar contained in the softdrink were not counted.

• The potato content of potato chipswas counted in computing the


Caloric intake doesnot predict obesityvery well.

Index. The fat content was notcounted in computing the vegetableand variety components of the Indexbut was counted in computing thefat component.

Other Components

In developing the Index, consider-ation was given to including a compo-nent to address food energy intake.Obesity is a significant health problemin this country. Nevertheless, it was de-cided that the inclusion of physicalmeasures of appropriate body weight,such as a body mass index (BMI) orconformance to standard weight-for-height tables, would be inappropriatesince they are influenced by factors,

Table 10

Energy Intake as a Percentage of RDA V

% Kilocalories

RDA1 15-20 20-25

<60% 3.17 36.41

60-80% 4.05 40.51

81-100% 4.92 43.21

101-120% 3.91 38.62

>120% 3.10 46.41

Column percent 3.96 40.37

Comments:Population is adult males age 18-74.BMI = weight/height*height.RDA1 = energy intake as a percent of RDA.Correlation coefficient: r = 0.02594 (p<0.38).

such as physical activity, unrelated topeople’s eating patterns. Use of a mea-sure based on food energy in relationto the RDA was also rejected for thisstudy, as preliminary tabulations of thedata indicated that these measureswere not highly correlated with physi-cal measures of obesity. A comparisonof the BMI and caloric intake is pre-sented in Table 10.

BMI values from 24 to 27 in womenor 25 to 27 in men indicate overweight;those over 27 indicate obesity. In Table10, persons with a BMI of greater than30 have a level of energy intake similarto individuals with a BMI of 20 or less.Thus, BMI is not highly correlated withcaloric intake.

ersus Body Mass Index

Row %BMI (n)

25-30 >30 Total

43.17 17.26 24.25(280)

39.89 15.55 33.08(397)

36.02 15.86 24.29(272)

48.70 8.77 11.29(125)

40.81 9.68 7.09(79)

40.80 14.86 100.00

the healthy eating index - usda

Documents