Framework for Dietary Risk Assessment in the WIC Program: Interim Report (2000)

Workshop on Dietary Risk Assessment in the WIC Program

Thursday, June 1, 2000

National Academy of Sciences

Lecture Room

2101 Constitution Avenue, N.W., Washington D.C.

Presented by Jean Anliker, Ph.D., R.D.

Adjunct Associate Professor

University of Maryland, Baltimore

Dietary assessment is an effective tool for identifying risk, tailoringinterventions, and testing the effects of nutrition education targetedto WIC participants. The Maryland WIC 5-A-Day Promotion Program andthe Maryland Food for Life Program, both funded by the National CancerInstitute, are two programs in which dietary assessment instrumentswere developed and utilized to evaluate program interventions. Bothused trained peer educators to recruit program participants and brieflyexplain questionnaire instructions. However, in order to avoid anyadministrator bias, neither were permitted to help the participantsin any capacity besides reading difficulties. While the instrumentswere not validated in terms of comparing survey results to actualfood consumed, the experiences of both programs provide valuableinsight into the feasibility and challenges of conducting detaileddietary assessment in the WIC program.

The Maryland WIC Five-A-Day promotion program was a randomized cross-overdesign with a 6-month intervention administered in 16 sites throughoutMaryland. Eight sights were selected to be intervention sites thefirst year; the other eight served as control sites. Their statuswas reversed in the second year so that each site served as theirown control. Subjects included 3,102 English-speaking women at least18 years of age who were either pregnant, breast feeding, or weremothers of children enrolled in WIC. Sites were chosen with differentdemographic characteristics including rural and urban populations,as well as different racial and ethnic profiles. The program had5 objectives: The intervention group would show significantly greaterincreases than the control group in (1) their knowledge about fruitand vegetable consumption; (2) their attitudes about fruits and vegetablesconsumption; (3) their self efficacy for fruit and vegetable consumption;(4) the stage of change for fruit and vegetable consumption; and(5) their actual fruit and vegetable consumption.

Survey development incorporated focus groups, pilot testing, andcognitive testing prior to initiation of its use in the study. Thecognitive testing involved participants reading out loud and verbalizingtheir thought processes as they answered questions. It was helpfulin ensuring that participants understood what was really being asked,and therefore, fine-tuning questions. The standard National CancerInstitute seven-item fruit and vegetable frequency consumption assessmentwas used along with a 35-item, 24-hour checklist for fruits and vegetables.The survey also considered psychosocial variables, including self-efficacy,perceived barriers, attitudes, social support, responsibility forfood preparation and purchasing, and knowledge relating to fruitsand vegetables. The final survey was self-administered and took participants15 to 30 minutes to complete. A $10 incentive was given to participantsafter the post-survey (8 months later) was completed.

Results of the study showed a significant difference, with a directlinear relationship between fruit and vegetable consumption and programattendance. An increase in 0.13 servings per day among the controlgroup and increase of 0.56 servings per day in the intervention group(p<0.002) were observed. Those who attended all sessions increasedconsumption by 1.25 servings per day. Other significant differenceswere seen in participant's knowledge regarding

the number of servings recommended for fruits and vegetables, attitudesabout fruits and vegetables, self-efficacy for fruit and vegetableconsumption, and perceived social support.

The Maryland Food for Life Program was similar in design to the Five-A-Dayprogram, however only 10 sites were selected and the program interventionwas 6 months rather than 8. While final data analysis has not yetoccurred, the food assessment methodology can be described.

The food survey development for the Food for Life program includedsimilar formative research (cognitive and pilot testing) to thatof the Five-A-Day program. In addition, Gladys Block, Ph.D. servedas a consultant in developing a food frequency instrument that wouldbe culturally appropriate for the program. Objectives for this programfocused on lowering the percentage of calories from fat, increasingdietary fiber, and increasing the number of servings of fruits andvegetables. Hence, the survey instrument needed to evaluate dietarypatterns rather than just servings of fruits and vegetables. An 83-itemfood frequency instrument, capable of generating a complete nutrientprofile, was developed for this program. In addition, a questionnairewas used to assess the fat levels of commonly eaten foods, stagesof change for fruit and vegetable consumption, as well as fat andfiber consumption.

The 83-item food frequency questionnaire included estimations ofportion size. Serving sizes were shown on the questionnaire and participantswere asked to estimate their usual servings as small, medium, orlarge. This self-administered questionnaire utilized computer-scannedbubble sheets and took participants an average of 30 to 45 minutesto complete. A $20 incentive was given to participants after thefinal survey was completed.

Data generated by the food frequency assessment instrument was vast.Profiles of nutrient data included calories, protein, fat, carbohydrate,cholesterol, fiber, vitamins, and minerals. Servings of food groupsfrom the food guide pyramid were also generated. The data generatedcould be used not only for evaluation, but was enormously valuablefor identifying target behaviors for interventions, both at the individualand population level.

Experience with both of these programs lead to the following conclusionsregarding dietary assessment in WIC programs: (1) WIC participantswill complete food surveys, even surveys that take 15 to 45 minutesto complete; (2) literacy was not a common problem; 80 percent ofthe participants had high school education and in only a few instanceswere peer educators asked to help with reading; (3) food survey questionswere well understood; (4) dietary data can be a useful tool in tailoringnutrition education; (5) trained peer educators can collect dietarydata; and (6) self-administered forms which can be computer scannedwork very well. The forms were scanned by the computer at the university,but having scanners on site in the WIC clinics would likely be feasible.

Presented by Valerie Tarasuk, Ph.D.

Associate Professor, Department of Nutritional Sciences

Faculty of Medicine, University of Toronto

Applying dietary assessment techniques to appraise the adequacy ofan individual's intake requires: (1) an estimate of the individual's usual or habitual dietary intake over the time period of interestand (2) an estimate of the individual's dietary requirements. Neitherestimate is straightforward.

The estimation of usual intake is complicated by the fact that individuals' intakes vary markedly from one day to the next. This variation isa function of environmental and biological pressures on intake andmeasurement errors. Further, the level of day-to-day variation differsacross nutrients, depending on their concentrations in the foodsthat comprise the individual's diet. Classically, the highest variationis seen for vitamin A and the lowest is seen for energy.

An equation developed by Basiotis et al. (1987) can be used to determinethe confidence interval around an estimate of usual intake basedon the assessment of actual intake on one or more days. Using thisequation, it can be demonstrated that the error term around an estimateof usual intake for an individual decreases as the number of daysof intake data increases. However, the error term for estimates basedon 1, 3, or even 7 days is very large, indicating that using a limitednumber of days to assess dietary risk for an individual is hugelyproblematic.

The difficulty in estimating individuals' usual intakes is compounded by the fact that the precise magnitudeand nature of day-to-day variation differs substantially betweenpeople. Further, data from low-income women seeking food assistancein Toronto suggest that issues of day-to-day variation are probablyeven more pronounced in the target group for WIC. Intakes becomeeven more erratic when economic constraints are added to everydayenvironmental and biologic influences. Because day-to-day variationcomes from multiple sources that vary among individuals, it is likelyimpossible to strategize data collection in such a way as to eliminatethis source of error in the estimation of individuals' usual intakes.Although there are statistical techniques to estimate within-personvariation and adjust for its effects on intake estimates in largepopulation studies, these methods are not appropriate for use atthe level of individuals.

Other potential sources of error in the estimation of individuals' usual intakes include systematic under- or overreporting, proxyreporting, social desirability, and errors in the assumptions maderegarding food composition and nutrient bioavailability. Underreportingappears to be pervasive in dietary intake surveys, affecting an estimated10 to 45 percent of samples. It has been associated with a numberof factors, but appears to be a particular problem among women withhigh body weight. There is also some evidence that children or youngadults are less prone to underreporting, and that individuals withlower socioeconomic status, education, and literacy levels are morelikely to underreport. It is unclear exactly what is being underreportedalthough there is some data to suggest that the reporting of fat,sugar, and alcohol may be particularly problematic. In contrast tothis, some studies have suggested that individuals may be likelyto overreport on the consumption of foods perceived to be healthful,particularly if they had been engaged in interventions designed toimprove healthful intake.

Usual intake can be estimated from the measurement of actual intakeover a limited number of days (using dietary recalls or records).Alternatively, a food frequency questionnaire

can be used. However, neither method yields an estimate of usualintake without error. This is well illustrated by a study of Crawfordet al. (1994) in which a small sample (n = 58) of 9-10 year-old girlshad their lunch time intake observed and were asked to report theirintake by one of three methods: 24-hour recall, 3-day food record,or 5-day food frequency. Median percent error between observed andreported intake by the different assessment methods (see Table 1) indicates that while the particular nature of the errors differed,all three methods measured intake with error.

Lastly, the problem of measurement is compounded by the difficultyin interpreting comparisons of intake estimates with dietary standardsor nutrient requirement estimates. Individuals differ from one anotherin their dietary requirements. Thus, if an individual's usual intakeof some nutrient falls below a particular reference value, for example,it cannot be assumed that the intake is inadequate to meet her requirementfor that nutrient. Nonetheless, the lower an individual's usual intakeis, the greater the probability that the intake is inadequate tomeet her needs. Both the between-person variation in nutritionalrequirements and the error inherent in our estimation of individuals' usual intakes need to be considered when applying dietary assessmentmethods to determine nutritional risk.

Basiotis PP, Welsh SO, Cronin FJ, Kelsay JL, Mertz W. 1987. Number of days of food intake records required to estimate individualand group nutrient intakes with defined confidence. J Nutr 117: 1638–1641.

Crawford PB, Obarzanek E, Morrison J, Sabry ZI. 1994. Comparative advantage of 3-day food records over 24-hour recall and5-day food frequency validated by observation of 9- and 10-year oldgirls. J Am Diet Assoc 94: 626–630.

Presented by Cutberto Garza, M.D., Ph.D.

Professor, Division of Nutritional Sciences

Cornell University

By law, the dietary guidelines represent the basis for federal policyand are used to guide nutrition information, education, and interventions.While the food guide pyramid is one of the major tools used for consumereducation, the dietary guidelines, which incorporate the food guidepyramid, are developed to be used as a policy instrument. The guidelinesare quite relevant to this committee's task, in that the guidelinesform the basis of federal food and nutrition education programs.

The development of the new Dietary Guidelines for Americans was avery complex exercise which relied heavily on available consensusdocuments. Contributing to the complexity was incorporating all agegroups from age 2 and older, all socioeconomic groups, both genders,and multiple physiological states. In addition, the information neededto be reduced in such a way as to be understandable to the averageconsumer, and yet be scientifically accurate.

In developing the dietary guidelines, it was necessary to come upwith generic information when, in fact, the basis for it was questionablein terms of the available science. Most difficult has been the nearlycomplete lack of usable data on the nutrient needs of infants, children,adolescents, pregnant and lactating women—the groups that the WICprogram deals with. While the DRIs are currently being revised, theinformation gaps in the estimation of average requirements by lifestage and gender have been very difficult to come by. For this reason,it is important to understand that there is an enormous amount ofscientific judgment that goes into the development of these guidelines.But yet, the best scientific information for developing the dietaryguidelines was available to this last guideline committee—more sothan to other committees who worked on previous versions of the guidelines.

The first “dietary goals” were issued by the McGovern report and were released in 1977. Thework of the Senate select committee was very controversial. Nutritionscientists were, quite literally, almost killing each other overwhether the science was adequate enough for these dietary goals tobe set. History would suggest that, in fact, the Senate committeewas quite wise in their recommendations on the whole. Most of themare quite relevant today and have since been borne out by the additionalscience.

In 1979, the American Society for Clinical Nutrition brought togethera panel that was to review the relationship between dietary practicesand health outcomes. Their findings were incorporated in the SurgeonGeneral's report on health promotion and disease prevention. Thesewere then the basis of the 1980 guidelines issued jointly by theDepartment of Health and Human Services (DHHS) and the United StatesDepartment of Agriculture (USDA). While it was primarily a federaleffort, some external individuals were involved as well. In 1983,nongovernment scientists were selected to review and make the recommendationsto USDA and DHHS about the first edition. Today, it is an externalgroup which reviews the wide breadth of information for all agesand income groups and then advises the government. Ultimately, theguidelines are issued jointly by USDA and DHHS based on the recommendationsof the external group.

The dietary guidelines are to be based on current knowledge abouthow dietary intake may reduce the risk of major chronic diseasesand how a healthful diet may improve nutrition. The basis is notjust deficiency diseases, but in fact, the emphasis is on how dietarypatterns can be used to both avoid deficiency and also achieve otherhealth outcomes that are believed to be related to diet and dietarypatterns. There is now a public law that requires publication andrevision of the guidelines at least every 5 years.

The guidelines serve five very broad objectives. The first is toassist consumers in making dietary choices—choices which are mostlikely to promote well-being and avoid or postpone the onset of diet-relatedchronic diseases. This presents some challenges given the paucityof data in terms of nutrient requirements of children and pregnantor lactating women. When one adds to this the relationship betweendiet, nutrient-specific recommendations, and chronic disease, onehas to rely almost exclusively on epidemiological data. This typeof data is great for hypotheses and associations, but in terms ofcausality, it presents some challenges.

The remaining objectives of the guidelines are: to assist federal,state, and local agencies in the development, implementation, andformulation of regulatory policies and programs; to assist healthcare providers in primary disease prevention efforts; and lastly,to guide other domestic and international for-profit and not-for-profitorganizations in the implementation of nutrition and health goals.

In developing the new dietary guidelines, the number of guidelinesincreased from seven to ten and are now grouped under the categoriesof Aim for Fitness, Build a Healthy Base, and Choose Sensibly (see Table 1).

Aim for Fitness. Under this heading are the guidelines Aim for a healthy weight and Be physically active each day. These guidelines are just as relevant to pregnant or lactating womenand young children over the age of 2 as they are for any other agegroup. While there are no specific recommendations as to what a healthyweight should be for a pregnant woman, weight is discussed for ageneral adult population. The advice regarding physical activityis relevant to pregnant and lactating women.

Build a Healthy Base-Let the Pyramid guide your food choices is the second group of guidelines and represents somewhat of a tautology.The pyramid is supposed to reflect the guidelines and the guidelinesare saying to follow the pyramid. Under this guideline is a sectionof particular relevance to pregnant women—Use of dietary supplements. The guideline committee felt that it was very difficult for pregnantwomen to meet the iron requirements for pregnancy and for this reasona statement was added to this effect. The other point that this guidelinemakes is for folate. Women who are pregnant or who are at risk forbecoming pregnant should choose foods that are fortified in folateor take a folate supplement. It was felt that, given present levelsof supplementation, if women were judicious in their food choices,a supplement would not be necessary.

Choose a variety of grains daily, especially whole grains. This guideline is of relevance to the WIC population because of theconcern regarding folate intake for pregnant women. When the U.S.fortification policy and consumption patterns were looked at, theguideline committee came to the conclusion that it was necessaryto put the emphasis on whole grains because many grains are not fortifiedand without the emphasis, women could be put at risk of folate deficiencyand birth defects.

Another reason whole grains were emphasized was recent epidemiologicaldata suggesting that whole grains can be, for reasons we do not yetunderstand, associated with significant reductions in cardiovasculardisease risk. The reason for using the word especially was that the grains group was the food group for which Americanshave the largest gap between consumption patterns and recommendations.By using the word especially, the development committee hoped to emphasize grains and increasepublic attention to this area.

Whole grains were separated from Choose a variety of fruits and vegetables because dietary patterns based on grains, fruits, and vegetablesappear to be associated with the highest reductions in disease risk.Consumers have treated these groups as interchangeable when, in fact,benefits from these nutrients are significantly distinct. By separatingthe groups, it was intended to point out that they are not interchangeable.

Keep foods safe to eat is the other new guideline and is the second guideline where pregnancyis specifically mentioned. The committee felt very strongly thatif a recommendation is made for a healthy diet, it had to be safeas well. One could not divorce microbiological safety or other contaminantsin the food supply from a healthful diet. So these were not nutrientguidelines but food-based dietary guidelines. Obviously, women maybe at risk if they are not following food safety guidelines, andthey may benefit from WIC, to the extent that WIC incorporates thisinto their educational materials. Pregnant women, along with theelderly, young children, and immune-compromised groups are at thehighest risk for certain food-borne pathogens.

The third group, Choose Sensibly, is the most complicated of all the guidelines. The difficulty stemsfrom the guideline Choose a diet that is low in saturated fat and cholesterol and moderatein total fat. Becausethe emphasis has been placed on low-fat dietswhen most of the concern was being driven by saturated fat, the messageregarding dietary fats and cholesterol was not being adequately translatedby either industry or consumers. This guideline is complicated becauseconsumers are now being asked to differentiate between total fat,saturated fat, polyunsaturated fats, monounsaturated fats, and trans-fats.While there is science to back this up, trying to educate the publicon these various types of fats will be an enormous challenge. However,it is one that the committee felt was significant enough because most of the risk is

related to saturated fats and trans-fats. The guideline committee5 years ago did not have the science that is available today on theissue of trans-fatty acids.

The most controversial guideline, certainly in terms of public comment,was the guideline Choose beverages and foods to moderate your intake of sugars. There was quite a bit of debate among the development group abouthow the guideline should be worded. The controversy was whether theword moderate or limit should be used. In addition, dietary intake trends indicated thatthe highest level of increasing sugar consumption came from addedsugars in beverages. There was significant concern that many consumersdidn't recognize beverages as a source of sugars—they were not foods in their minds. For this reason, the committeeneeded to also include the word beverages in the guideline.

The development committee needed to be very careful whenever theterm moderation was being used in terms of quantitating the advice. One of the majorchanges they tried to make with the new guidelines was to make themmore actionable. For example, the word moderate in the total fat guideline—the wording states to aim for a total fat intake of no more than30 percent of calories, but one does not need to go much beyond that.The committee felt that telling consumers to moderate their intakewas very difficult to put into action. On the other hand, there weremany others who felt that by using the term limit, the committee was going beyond what the science provided. The committee's rationale was that given the increasing rates of obesity in thiscountry, it is obvious that there is an energy balance problem. Ifconsumers are asked to eat more of certain products, they neededto be asked to eat less of other products. Consumers were asked tolimit their consumption of alcohol, fat, and sugar.

The next guideline in this grouping, Choose and prepare foods with less salt has been reworded but the message is essentially the same. The onlydifference is that there is now a quantitative recommendation forsalt intake.

The third guideline in which pregnancy is specifically mentionedis If you drink alcoholic beverages, do so in moderation. Throughout committee deliberations, it was clear that there was noway to distinguish at what level of drinking would birth defectsbecome a problem or have other adverse effects in women. For thisreason, pregnant women or those at risk of becoming pregnant wereasked not to drink alcohol at all.

The other reason for the way the guideline is stated is because thereare health benefits associated with alcohol. It does not matter whatform the alcohol is in (wine, beer, distilled), it is the ethanolthat appears to be the active agent. There is a significant reductionin risk to cardiovascular disease that appears to operate via a reductionin platelet aggregation. Because it is an acute effect, one doesnot obtain any benefits from a lifetime of drinking. Rather, if oneis interested in drinking for health benefits, there is no reasonthat drinking should begin at age 21 because the benefit does notbecome available until one is at risk for cardiovascular disease—45 years of age for men and 55 years of age for women. It is alsoclear that, at least for women, more than one drink a day increasesrisk for breast cancer. However, women should not be asked to refrainfrom drinking alcohol because there are benefits as well as risks.For this reason, postponing drinking until after the age of 55 ifconsuming it for health reasons seems the most rational and prudentthing to do.

Overall, the major changes in the guidelines are mainly the groupingof the recommendations and the addition of guidelines on physicalactivity and food safety. Because pregnancy is such a teachable moment,it is hoped that WIC provides an opportunity to impress the pointthat physical activity is important for everyone. It is not justfor women in other age groups, men, or for those who need to loseweight. There are significant health benefits that

physical activity provides. Physical activity permits individualsto increase food intake and therefore increases the likelihood ofgetting all the nutrients needed from food. Other changes in the2000 guidelines include a separation of guidelines on grains fromfruits and vegetables, a shift from an emphasis on total fat to thereduction of specific types of fats, a recognition of the trendsin sugar intake, and greater specificity in the alcohol guidelines.

The guidelines are quite applicable to the WIC population, but onlyafter the age of 2. The likely reason that everybody has shied awayfrom the under 2 population is that trying to look at nutrient-specificrequirements is very difficult based on the data available for thefirst year of life. However, focusing on women's health in the WICsetting can be very beneficial and advantage should be taken of thisopportunity. In fact, it is scientifically appropriate to do so.Pregnancy represents an ideal time to get women to focus on theirhealth and enhance their knowledge, behavior, and practices. In manyhouseholds, women are the gatekeepers for health. In addition, formany women, problems with overweight and obesity appear to beginduring pregnancy and the postpartum period.

While the guidelines are very applicable to the WIC population interms of defining an inadequate diet, it is not likely possible tocome up with a simple paradigm or algorithm for compliance by lookingat only a few of the guidelines. The grouping of guidelines werenot intended to be prioritized in any way; they all are relevantto good health. It is important to remember that the risk is to long-termoutcomes. For the individual, we have gone to some lengths to askpeople to look at their own family histories and their own risks.For example, if one knows that there is no hypertension in theirfamily history and no one has ever died of stroke, then for thatindividual, perhaps the salt guideline under those conditions wouldnot be as important. Perhaps in this particular family history, everyoneseems to be dying at the age of 50 from atherosclerosis. For thisindividual, paying attention to fat and physical activity or havingone drink a day once over the age of 40 may be more important tolowering this individual's risk. The guidelines could perhaps beprioritized in some way for each individual. However, from a publichealth perspective, it is not likely possible.

Presented by Kristin Marcoe, M.B.A., R.D.

Nutritionist, Center for Nutrition Policy and Promotion

U.S. Department of Agriculture

The Food Guide Pyramid and the Food Guide Pyramid for Young Childrenare sciencebased food guidance. A food guide translates recommendationson nutrient intake into recommendations on food intake. The FoodGuide Pyramid and, subsequently, the Food Guide Pyramid for YoungChildren were developed using a sound research process and were basedon a set of established philosophical goals. These philosophicalgoals included: (1) a food guide should promote overall health; (2)it should be based on current nutrition research; (3) it should bebased on the total diet instead of a core or foundation diet; (4)it should be useful to a target audience; (5) it should meet nutritionalgoals in a realistic manner; (6) it should allow flexibility in howconsumers meet nutritional goals; (7) it should be practical, hencevarying numbers of servings to meet different calorie and nutrientneeds; and (8) it should be evolutionary.

Nutrition goals for the food guide were originally based on the 1980Recommended Dietary Allowances (RDAs) for nutrients for which thereare adequacy concerns such as calories, protein, vitamins, and minerals.Also considered were the 1980 Dietary Guidelines for Americans forfat and sugar—for which there are moderation concerns. Subsequentreleases of the RDAs and Dietary Guidelines have been used to updatethe food guide.

The food groups were defined for the food guide by primarily consideringthe nutrient content of a food, but also the usual use of a foodin meals and how a food was grouped in earlier food guides. In themost recent food guide, foods which were high in fat and added sugarsand low in nutrient density were put in a separate group, the tipof the pyramid, in order to highlight the need to moderate intakeof these food components. In addition, vegetables were separatedfrom fruits, and subgroups within the vegetables were used to emphasizenutrients and specific food components.

Serving sizes for the food groups were based on typical serving sizesreported in the U.S. Department of Agriculture's (USDA) food consumptionsurveys. Consideration was also given to using common measures easilyutilized by consumers, such as 1 cup. Nutrient content was used toestablish serving sizes in the milk group and for meat alternatesin the meat group. Each food in the milk group was assigned a servingsize that would provide 300 mg of calcium, the amount in 1 cup ofskim milk. Amounts of meat alternates would provide approximatelythe same amount of protein and minerals found in 1 oz of meat.

Tradition dictated that serving sizes used in previous food guideswill continue to be used for some foods. For example, two slicesof bread is the typical amount eaten at a single occasion, yet oneslice of bread is considered a serving. There was concern that ifthe serving size was changed to two slices, it would reduce the minimumnumber of servings for grains from six to three. This would havecontradicted the philosophy of the Dietary Guidelines, which encouragethe consumption of more grains.

In order to determine the nutrient profiles of each food group andsubgroups, composites were created using USDA food consumption surveydata. Each profile represented the amounts of nutrients one wouldexpect to obtain from eating a serving of a food group or subgroup. The

original Pyramid work used 1977–1978 Nationwide Food Consumption Survey data. The 1989– 1991 Continuing Survey of Food Intake by Individuals (CSFII) datawas used to update the original data. It was based on foods reportedlyconsumed by 11,488 individuals, 2 years and older over 3 days. Sampleweights were applied to provide estimates representative of the population.Work is currently in progress to update the composites using 2-dayweighted data from the 1994–1996 CSFII and the Agriculture ResearchService's Pyramid Servings Data in conjunction with consumption datafrom this survey.

For the initial update, composites for each food group and subgroup,such as dark-green vegetables, were developed based on reported consumptionof food items (i.e., cooked broccoli) in the CSFII 1989–1991. USDA's Center for Nutrition Policy and Promotion developed a Food GuideServings Database for the CSFII 1989–1991. This database was usedto convert grams of foods consumed into numbers of food guide servingsconsumed. A composite of item groups was then constructed, weightedby the numbers of servings of each reported by all individuals. Onefood code was selected to represent each food item group in eachof the composites.

Nonfortified ready-to-eat and cooked breakfast cereals were usedin both the whole grain and enriched grain composites. In this way,the nutrient profiles of these composites do not overestimate thenutrients for those individuals who do not eat fortified breakfastcereals. Nutrients added at standard enrichment levels, such as inenriched bread, were used in the composites. Folate fortificationwas not mandated by the government at the time the 1989–1991 surveywas conducted, and was therefore not reflected in the nutrient profilesfor grain products. However, in the 1994–1996 work, nutrient valuesfor folate were modified to bring them up to current levels.

Nutrient profiles for each composite were calculated using formsof food items lowest in fat and added sugars. This reflected thephilosophical goals that the food guide should be realistic and allowmaximum flexibility for users to select specific sources of fatsand sugars within their diets.

To minimize sodium, the form of the food item with the lowest amountwas used in the composite. For most of the vegetable and cooked grainitem groups, food codes were used which specified “no salt addedin preparation.” However, in a few cases, the salted form of thefood was used for vegetables based on 1989–1991 food supply data.

Once a food code was selected to represent each of the food itemgroups in each of the composites, grams and corresponding nutrientvalues of each were calculated for its portion of the composite serving.Nutrient values were then summed across all item groups in the foodgroup or subgroup to determine the composite's nutrient profile perserving. The nutrient values were derived from USDA's Survey NutrientDatabase. Thus, each composite had a nutrient profile which representedthe most frequently consumed foods in that food group.

A fat composite was developed to represent “discretionary” fat added to the diet above the fat found in lean meats and theother forms of composite foods that were lowest in fat. For example,the fat in cakes and margarine spread on bread are discretionaryfats. The nutrient profile for the fat composite was updated usingproportions of animal and vegetable fats in the food supply during1989–1991.

The numbers of servings for each food group and subgroup in the foodguide were based on nutrient adequacy and moderation. The numbersof servings required to meet the 1989 RDAs for protein, vitamins,and minerals were determined. Since RDAs vary based on age, sex,and pregnancy status, ranges in the numbers of servings of the foodgroups were set to cover the full

range of nutrient needs. The higher number of servings is for individualswith greater caloric and nutrient needs.

Sample Food Guide Pyramid diet patterns for several caloric levels,including 1,600, 2,200, and 2,800 calories, were developed to reflectthe range of food group servings. For example, the minimum numberof servings for each food group is based on a 1,600-calorie pattern,while the maximum number of servings reflects a 2,800-calorie pattern.The greater numbers of servings in the higher calorie diet patternsare for individuals with higher nutrient and calorie needs, suchas teenage boys, active men, and very active women. The lowest numberof servings is for sedentary women and older adults.

Choosing a variety of foods from within each food group is importantin obtaining the expected levels of nutrients. This is particularlytrue for the vegetable group, where certain subgroups (i.e., dark-greenand deep-yellow vegetables, legumes) have been targeted for increasedconsumption.

Three whole grain servings in the patterns were based on the FoodGuide Pyramid recommendation to choose “several” servings a day of foodsmade from whole grains. There has always been an emphasis on wholegrains in Pyramid materials, and this continues to be consistentwith the Dietary Guidelines. For the meat alternates, the numberof egg servings was based on three eggs per week. This ensured thatthe cholesterol value across the patterns would average 300 mg orless per day, based on the National Research Council's 1989 Diet and Health report.

Three servings from the milk group had been recommended in the pastto meet the higher calcium needs of pregnant or lactating women,teenagers, and young adults to age 24. In light of the change incalcium requirements in the new DRIs, three servings from the milkgroup are now recommended for older children and teenagers (childrenages 9 to 18) and adults over age 50. During pregnancy and lactation,the recommended number of milk servings is the same as for nonpregnantwomen. This information appears in the newly-released Dietary Guidelines2000.

As indicated previously, sample Food Guide Pyramid diet patternswere developed and composites and their nutrient profiles used inthese at three calorie levels that would meet established goals forprotein, vitamins, and minerals. Ranges in numbers of servings wereused to cover the varying needs of males and females 2 years of ageand older. These patterns included added or discretionary fat andadded sugars. Discretionary fat was added to bring the level of totalfat in the three patterns to 30 percent of calories. Added sugars,represented as teaspoons of sugar, were included to bring caloriesto the targeted levels. It should be emphasized that the differentamounts of sugar in the three patterns are not recommendations, butmerely indicate the calories, in teaspoons of sugar, needed to meetthe targeted calorie levels for each pattern.

Analysis of the Food Guide Pyramid diet patterns have shown thatthe RDAs are generally met. Nutrients that failed to meet the RDAswere iron and zinc. Iron is the main nutrient for which adequacyis a concern, but only for individuals who have both high RDAs foriron and who choose the minimum number of servings from the foodgroups. For example, a female whose energy requirement is 2,200 caloriesper day may be choosing the minimum amount of servings from eachfood group, and therefore, only consuming about 1,600 calories perday. At this reduced calorie level, nutrient needs, especially ironand zinc, may not be met. Therefore, these females would need toconsume foods rich in iron or eat an iron-fortified breakfast cerealin order to meet their nutrient needs.

The diet patterns also meet the saturated fat goal of less than 10percent of calories. Cholesterol values in the 1989–1991 composites,being used in the patterns, are now lower than what appear from the1977–1978 composites. Sodium levels continue to fall below the 2,400mg recommendation. Total dietary fiber continues to range from 17g in the 1,600-calorie pattern to about 27 g in the 2,800-caloriepattern.

The Food Guide Pyramid for Young Children was developed using thesame methodology as was used for the original pyramid. The quantitativeand qualitative research for this project will be published in thenext issue (this summer) of Family Economics and Nutrition Review, a peer-reviewed journal of the USDA Center for Nutrition Policy andPromotion.

In order to meet the objective of adapting the original Food GuidePyramid to a younger audience, a literature search was first conductedto help decide on the target audience of 2–6-year-olds. The philosophicaland nutritional goals, developed for the original Pyramid, were reexaminedand adapted for young children. Once the goals were set, discussionswere held with nutrition professionals who educate young children,parents and care-givers.

Just as composites were developed for the original Pyramid, foodgroup and subgroup composites were developed based on reported foodintakes of 1,053 children ages 2 to 6 years over 3 days in the CSFII1989–1991.

The whole grains composite is an example of what was done for eachof the food groups and subgroups. The children obtained the bulkof their whole-grain servings from cereals, both hot and ready-to-eat.Once the make-up of each of the food groups and subgroups was determinedvia composites, the average quantity of food a child would eat ata single occasion was calculated. The amounts reported for youngchildren were 60 to 80 percent of those for all individuals. Thissuggests that it is appropriate to continue to use the serving sizeestimate of one-third smaller serving that was used previously inassessing nutrient levels in Food Guide Pyramid patterns for youngchildren. Two- to 3-year-old children eat two-thirds of a serving,whereas the older or 4- to 6-year-old children have regular-sizedservings.

While a 1,300-calorie pattern was developed for 2 to 3 year olds,a 1,600-calorie pattern was developed for 4 to 6 year olds. Althoughthe REI for 4 to 6 year olds is 1,800 calories, food consumptiondata reported for this age group in the CSFII 1989–1991 averaged1,533 calories.

Focus groups with parents indicated that parents were concerned thattheir children might not be getting adequate nutrients from the amountof food that they were eating. It was important to determine whethera diet pattern set closer to the level of calories that childrenreportedly eat, based on their reported food choices, could meettheir nutrient requirements. The 1,300-calorie pattern representedthe minimum number of servings per Food Guide Pyramid food groupand was calculated by reducing the 1,600-calorie pattern by one-thirdfor all the food groups and subgroups except milk, which remained2 cups. This was to represent the smaller serving size estimatedfor 2- to 3-year-old children.

The nutrient levels in these patterns were compared to the RDA for2 to 3 and 4 to 6 year olds, as well as to actual nutrient intakesof these children from survey data. The goal was to meet or exceedthe RDA for nutrients, or to at least improve levels relative toactual consumption. The 1,600-calorie pattern met all nutrient requirementsfor children age 4 to 6 years, except for vitamin E. The 1,300-caloriepattern provided the RDA for most nutrients for 2 to 3 year olds.The major exceptions were iron and zinc. When breakfast cereals fortifiedwith

iron and zinc were included in the grain composites, the patternsdid provide recommended levels of these nutrients. In looking atthe average actual consumption of servings in each of the food groupsand subgroups, it was evident that children were not consuming therecommended numbers of servings from most food groups. Their dietscould be improved by eating more dark-green and deep-yellow vegetables,legumes, whole grains, and lean meat, poultry, and fish.

Beyond the composites for children, analysis for other subpopulationgroups, like Hispanics and African Americans, have not yet been done,although this is possible if the sample size is large enough in USDA's food consumption surveys.

The food guide is based on nutrient recommendations such as the DRIsand Dietary Guidelines. For this reason, as these are revised andreleased, the food guide recommendations will be reassessed to makesure they meet the nutritional objectives. An article entitled Reassessingthe Food Guide Pyramid: Decision-Making Framework, written by AnneShaw and others at the Center for Nutrition Policy and Promotionwill appear in the March/April 2000 issue of the Journal of Nutritional Education and serves as a good reference of the reassessment process.

Updating the nutrient profiles of the food group composites can alsobe done when there are major changes in nutrient composition, likethe fortification of certain grain products with folate. By usingupdated composites in the Food Guide Pyramid diet patterns, and comparingexpected nutrient levels to nutritional objectives, it can be determinedwhether or not the patterns meet objectives. If not, there are severaloptions. Guidance could be individualized for selected sex/age groupson choosing certain foods in food groups to help them meet highernutrient standards. The numbers of servings for a food group or subgroupcould be modified. Another possibility would be to create a new foodgroup or subgroup to emphasize sources of a target nutrient or foodcomponent, such as a separate tomato group if lycopene intakes continueto generate interest, assuming serving recommendations were developed.Each proposed change would need to be evaluated in terms of its effectson calories and other nutrients in the diet patterns. In addition,consumer understanding of the Food Guide Pyramid and barriers toits use must be monitored. Such research is currently being plannedand through it, the USDA will continue to have a tool that consumerscan successfully use to meet their nutrient needs.

Presented by Amy Subar, Ph.D., M.P.H., R.D.

Research Nutritionist, Division of Cancer Control

National Cancer Institute

From a cognitive point of view, food frequency questionnaires (FFQs)can be difficult for respondents to understand and complete. Theyrequire complex knowledge, skills, and attentiveness with respectto food intake. Cognitive techniques can be used on all types ofinstruments, whether they are self- or interviewer-administered,and would be useful in the WIC population or any specific population.

This research describes the methods and results of three rounds ofcognitive interviewing, the purpose of which was to improve self-administeredFFQs. Cognitive aspects having to do with the wording, ordering,or anything that would ease the task and help respondents betterunderstand the questions were evaluated. The cognitive research wasconducted with the help of two cognitive psychologists, Jared Jobeand Albert Frederick Smith. The goal was to learn about the cognitivestrategies individuals use in responding to FFQs, identify the problems,modify and test new questions, and design a cognitively improvedquestionnaire. The instrument that was developed based on this cognitivetesting was then to be evaluated with respect to response rates,validation, calibration, and measurement error.

Using a Block questionnaire as the baseline instrument, 24 respondentsin each of three phases were enlisted. While an effort was made torecruit a diverse group of respondents in terms of socioeconomiclevel and ethnic diversity, individuals with major literacy issueswere not included. The respondents were asked to think aloud as theycompleted the instrument. In this way, the researcher could hearwhat the respondents were thinking as they formulated their answersto each question. Mistakes and misunderstandings could be heard asthe instrument was completed. Doing pilot testing in this way versusdiscussing questions, thoughts, or problems after completion of theinstrument has advantages in that interviewers hear and see problemswith question, wording, layout, and understanding for each individualquestion as it is answered. In some cases, such problems might notbe remembered at the end of a long task. Each interview took, onaverage, 2 hours per respondent.

Following the interviewing, the researchers pooled findings and discussedcommon problem areas. The interviewing uncovered a series of bothsubtle and generic issues in responding to FFQs having to do withwording, layout, design, and order. Specific examples of problemareas had to do with respondents having difficulty with not havinga response category of “never” separated from “a few times per year,” responding to portion size questions referredto as “small,” “medium,” and “large,” responding to frequency of intake for seasonal foods, andresponding to line items which included one or more foods not typicallyeaten as substitutes for one another (for example, tomatoes and tomatojuice). Although averaging intake over 1 year is not particularlyeasy, it does provide the best picture of usual intake. Asking aboutshorter time frames such as the past month or week may be easierbut may not best represent long-term usual intakes. Many of the stumblingblocks for respondents are simple things that can be easily changedhelping respondents to easily get through the process of completingthe instrument. Although most of the innovations incorporated intothe new instrument based on the cognitive testing would likely makecompleting the questionnaires easier for respondents, the next

important area of research was whether or not such innovations wouldlead to better food group and nutrient intake estimates.

Findings from the cognitive testing led to the development of a newFFQ at the National Cancer Institute, the Diet History Questionnaire(DHQ), which was validated using a checklist approach, to assessfrequency of intake in the past month. Three studies were conductedor are in the process of being conducted to validate it to see if,in fact, the cognitive improvements would prove to be valuable invalidation studies. Findings from the Checklist Validation Studyshow that most, but not all of what were considered cognitive improvementsincorporated in the DHQ were superior. Many of the changes made tothe instrument based on the cognitive testing showed significantimprovements over the Block instrument as compared to reference datafrom the past month. In a few areas, the data showed that the changeswere not improvements. In some cases, even if accuracy was no different,the cognitively improved questions were retained if the investigatorsfelt strongly that they were easiest for the respondents to answer.Based on the findings of the Checklist Validation Study, the DHQwas further modified.

Further piloting compared response rates and data completeness betweenthe DHQ and a standard FFQ in a sample of participants from a clinicalscreening trial. Response rates were identical for both FFQs. A significantlylower proportion of respondents skipped or missed portion size andsupplement questions on the DHQ versus the standard FFQ. These findingssuggest that intensive cognitive interviewing is beneficial in thedevelopment of dietary assessment instruments and would be beneficialin developing a dietary assessment tool to be used in the WIC setting.

Presented by Gladys Block, Ph.D.

Professor, Department of Public Health and Health Policy Management

University of California, Berkley

This study was designed as a validation study for the Block FoodFrequency Questionnaire (Block FFQ) and the Harvard Food FrequencyQuestionnaire (Harvard FFQ). The questionnaires were self-administeredand had both geographic and ethnic variation. Full length questionnaireswith manual scoring systems were tested. Subjects consisted of WICparticipants of African American, Caucasian, and Hispanic origin.Seven hundred fifty-seven pregnant, lactating, and nonpregnant women,and children ages 1 to 4 completed the study. There was equal distributionacross the three ethnic groups and throughout WIC clinics of California,Ohio, Texas, and New York. Intake was examined for protein, iron,calcium, vitamins A and C, and energy.

Participants in WIC clinics were randomized to receive either theHarvard or Block FFQ (see Table 1). The questionnaires, which were self administered at baseline,were offered in English or Spanish. Then, over the next month, threetelephone interviews took place using the Minnesota NDS system. Atthe conclusion of the 1-month time period, a self-administered questionnairewas repeated at the WIC clinic.

Results among the women indicated identical caloric intake and nosignificant differences in the mean nutrient intakes between theHarvard and Block FFQ. Among the children, there were significantdifferences in the mean nutrient intakes by recall data in protein(47.5 g [Harvard] vs. 52.0 g [Block]; p = 0.04) and calcium (876mg [Harvard] vs. 893 mg [Block]; p = 0.06). Data were recorded forthe length of time for completion of the FFQ, the amount of assistancerequired, and the time for the WIC staff member to perform manualscoring.

While these tools were originally developed for epidemiological study,in this study they were used to determine eligibility for the WICprogram. Both are long instruments designed to calculate an extensivelist of nutrients. For this reason, the time for completion and evaluationis

overestimated since the instruments are more extensive than whatwould be needed in a WIC clinic.

Nutrient intake recalls were examined and compared to both the Harvardand Block FFQs. Both the Harvard and Block FFQ were found to significantlyoverestimate intakes of protein, calcium, vitamin A, and vitaminC. The Harvard FFQ also overestimated energy intake and the BlockFFQ overestimated iron intake. While both instruments were foundto overestimate various nutrient intakes, the Block FFQ tended tooverestimate less often than the Harvard FFQ.

Pearson correlations were performed on the Harvard and Block FFQbetween the FFQ and three 24-hour recalls (see Table 2). Correlations below 0.40 were considered inconclusive. The BlockFFQ tended to have a higher correlations for African American andCaucasian WIC participants. Both instruments had low correlationswith three 24-hour recalls in children and Hispanic women, and thereforeseemed to be no more reliable than chance.

One possible explanation for the poor outcomes seen in the Hispanicpopulation could be attributed to the level of education. While allof the African American respondents had greater than a ninth gradeeducation, 35 percent of the Hispanic respondents and 1.4 percentof Caucasian respondents had less education than the ninth grade.In addition, the FFQs were designed for African American and Whitewomen's cognitive capabilities, not Hispanics. This is an importantarea to pursue, given that various cultures answer questions in differentways.

Cross-classification is a better measure than correlation becausecorrelations are influenced by a number of factors, including therange of distribution. The questionnaires and the diet recall weredivided into quartiles by 3-day averages, then cross-tabulated. Forthe African American and Caucasian participants, the agreement was28 percent. The cross-tabulation approach worked better for the BlockFFQ and did not work at all for the Hispanic participants

Manual scores were examined for the Block FFQ to determine if thismethod was able to identify a group with low intake (or high intake).Based on the average of three 24-hour recalls, low true intake levelswere observed in the low quartile of the manual score method in theWIC nutrients. With the manual scoring system it is possible to movethe cut point. For example, if individuals who are getting less proteinthan the recommended level need to be identified, the cut point couldbe moved down. Another example would be to move the calcium cut-offpoint in order to identify individuals with intake under 800 mg,or less than the RDA. To accomplish this, the cut point could bemoved. Unfortunately, this approach would not be valid for the HispanicWIC population or in children.

Use of a FFQ is difficult in WIC children. Women who had 24-hourresponsibility for the nutrition of their children were asked toreport the food intake for their children. No significant correlationsin FFQ and diet recall were found. Using short simple screening questionsmay be a more effective method for evaluating dietary risk in children.For example, questions such as “Did you give your child any fruitlast week?” or “How many days a week did you give your child any fruit?” There will be problems in justifying the scientific validity, butin terms of common sense, one could say if it is less than sevenfor any of these, then there should be a concern for that child.

In terms of Hispanic participants, similar problems exist. Correlationswere found to be poor for both instruments. Since the Block FFQ wasnot designed to contain Hispanic food choices, it is not currentlya good choice for this population. The instrument would need to bemodified for a Hispanic population. Because of problems with educationlevel, interviewing respondents rather than providing participantswith questionnaires that need to be self administered may help. Overall,the Block questionnaire appears to be an instrument that works inthe African American and Caucasian population when using the manualscore methodology.

Presented by Graham Colditz, M.D., Dr.P.H.

Professor of Medicine

Harvard School of Public Health

The Harvard Food Frequency Questionnaire (HFFQ) is a self-administeredtool in which a client makes reference to a 4-week time period incontrast to a 1-year reference period that many other FFQs use. Itwas originally developed as a 60 item questionnaire for nurses inresponse to a request from the National Cancer Institute. It hassince been expanded to include approximately 120 items. In the mid-1980s,the questionnaire was modified and implemented for use in pregnantand lactating women. Later, it was also modified for use in children1 to 5 years of age. Most of the development for use in childrenincluded input from local WIC participants in Boston.

The HFFQ is currently available in both English and Spanish and canbe administered in a paper format as well as a computerized directentry format. It is comparable to other FFQs, except for the lengthof the reference period. It is simple and tends to have less itemsthan other FFQs. Responses can be manually entered by either theparticipant or staff in a WIC setting and can provide instant computer-generatedprintouts. Computerized scoring systems are optimal to manual scoringmethods because they can provide tailored feedback.

The development of the list of foods and portion sizes for the HFFQis based on data from the Continuing Survey of Food Intake by Individuals.It has been designed with substantial input from clinics in Massachusettsas well as U.S. Department of Agriculture (USDA)-funded focus groupsand clinics in North Dakota. The Harvard nutrient database is used.Although this database is primarily based on USDA data, it is additionallysupplemented for some nutrients. While it contains data on more than80 nutrients, the WIC Clinic computerized version focuses on nutrientsspecifically focused on in the WIC program. The database is continuallyupdated. For example, when the food supply was fortified with folate,the values for folate were updated for all foods in the HFFQ. Onelimitation of the instrument is the focus on nutrients rather thanwhole foods and food patterns. This is an area that is currentlybeing reexamined.

The computerized direct entry version was developed and pilot testedin Massachusetts with feedback from WIC providers and clients. Theinstrument is in a menu-driven format with on-screen directions.This set-up allows for branching. For example, if a client fillsin that she does not drink milk, she does not have to get the nextset of questions regarding milk consumption. This makes the overallprocess more efficient. Another advantage of direct entry is thatthere is not the opportunity for clients to skip lines. In contrast,in the old grid system, one could go across the rows when markinganswers and put two marks on the same line. Another advantage isa significantly lower error rate on the computerized entry versioncompared to the pencil and paper version. Hence, there is less likelyto be gross overestimation. It is also efficient in terms of theprovider's time. The participant can complete the HFFQ prior to meetingwith the WIC counselor and the allotted time can be used for nutritioncounseling, rather than completing the assessment. The disadvantagesto this process include the equipment cost and space requirementsin addition to a paper copy needed for backup in the client record.

Several validation studies using the HFFQ have been performed inpregnant women and children. In 1999, Blum and colleagues conducteda validation study using the HFFQ in White and Native American children.Suitor and colleagues (1989) conducted a validation study on Caucasian,African American, and Hispanic women. The analysis consisted of WICnutrients only. In 1999, Wei and colleagues used the same data, expandedthe range of nutrients in the analysis, and used an alternative approachto error correction for intra-individual variation in day-to-daydiet, based on three 24-hour recalls. A summary of the correlationcoefficients found in these studies can be found in Table 1.

Recent data indicate that the HFFQ can be used for determinationof WIC eligibility based on dietary risk. In a completed, but notyet published, study by Rodan and colleagues, an 80-item versionof the HFFQ is compared to a 31-item version of a Massachusetts FFQ.The HFFQ uses a reference period of 4 weeks and the Massachusettsquestionnaire a time period of 1 week. Ninety-six percent of womenfound the HFFQ easy to use and can complete it independently within9–12 minutes. The HFFQ identified 95 percent of these women as beingat nutritional risk compared to 85 percent based on the Massachusettsquestionnaire. Only 6 percent, however, were found to be eligiblefor WIC based solely on dietary risk. In children, the HFFQ identified98 percent to be at nutrition risk compared to 94 percent based onthe Massachusetts questionnaire; 49 percent were eligible for WICbased solely on diet.

In summary, WIC programs in several states have implemented the HFFQand there is growing evidence from validation studies that indicateits performance in WIC settings is comparable to that of a researchtool. It is conceptually superior to a 24-hour recall because WIC's focus is to measure long-term intake. The ease of use has alsobeen documented by studies in Massachusetts. While measurement ofbiochemical markers would be the best validation of the HFFQ, unfortunatelythis has not been done and is not practical at this time. The printoutthat the client receives from the HFFQ includes both nutrients andfood groups and this

output can easily be used for nutrition counseling. In addition,if the HFFQ is administered repeatedly, research has indicated thatit could be used to evaluate change in diet over time. In the future,there is also the potential to use the data for surveillance andprogram planning.

Blum RE, Wei EK, Rockett HR, Langeliers JD, Leppert J, Gardner JD, Colditz GA. 1999. Validation of a food frequency questionnaire in Native American andCaucasian children 1 to 5 years of age. Matern Child Health J 3: 167–172.

Suitor CJ, Gardner J, Willett WC. 1989. A comparison of food frequency and diet recall methods in studiesof nutrient intake of low-income pregnant women. J Am Diet Assoc 89: 1786–1794.

Wei EK, Gardner J, Field AE, Rosner BA, Colditz GA, Suitor CW. 1999. Validity of a food frequency questionnaire in assessing nutrientintakes of low-income pregnant women. Matern Child Health J 3: 241–246.

Presented by Anna Maria Siega-Riz, Ph.D.

Assistant Professor of Nutrition and Maternal and Child Health

University of North Carolina School of Public Health

Over the last 10 years, the Hispanic population has increased morethan 50 percent. The majority of the increase has occurred in individualsof a Mexican, Central South, and South American origin. Hispanicchildren are the second largest group of children in the country.The Mexican-American subgroup is the largest (60 percent) of theHispanic population; however, the second largest Hispanic subpopulation(23 percent) cannot be classified as Mexican, Puerto-Rican, or Cuban.This reinforces the concept that one dietary assessment tool cannotserve everyone within a state or across a nation. There are fourmain considerations when examining a dietary assessment tool. Theseare language, educational level, cultural practices, and dietaryculture.

It is usually not appropriate to use a translator as an interpreterin the public health setting. It is more useful to utilize a personwho is linguistically and culturally competent in the area of Hispanicattitudes and practices when administering a dietary assessment andproviding other WIC services. Although there are no studies availablethat validate the accuracy of the information reported by an interpreterat a WIC clinic, it would not be uncommon that the viewpoint of theinterpreter may be expressed or the participant's responses may befiltered. This is an especially important consideration when translatingparticipants' views of food voodoos or taboos or breast feeding attitudesand practices, which are very much an everyday occurrence in theWIC program.

Based on national data, Hispanics have the lowest rates of high schooldegrees. The majority of native-born Hispanics have less than 8 yearsof formal education. Many of the available food frequency questionnaires(FFQs) require thinking on an abstract level. It is difficult forindividuals with minimal education to understand an FFQ, especiallywhen asked about the frequency of consumption of a particular foodwhen it is not in the context of a meal. This is not only true forHispanics, but for anyone with a low level of education.

When providing health services to the Hispanic population, thereare many cultural issues that present challenges. Often, a Latinawoman has strong viewpoints that have been passed down from generationto generation, such as “the family is always first.” The woman may provide her family with theadvice and services provided to her by WIC, while neglecting herself.Another strong cultural issue is conflict with authority figuresand social desirability. For example, a Hispanic woman may not disclosein-depth answers to an interviewer for fear that the interviewerwould think less of her because of her inability to provide for herfamily.

Different subgroups of Hispanics have different food consumptionpatterns. One dietary assessment tool can not adequately assess allHispanics living in America today. None of the existing dietary assessmenttools have been designed to capture the dietary habits of differentsubgroups of Hispanics. In an ethnically diverse population, the24-hour recall or a food diary would be the preferred methods ofassessing dietary intake because they are open ended and allow theindividual to report the foods they are consuming as opposed to selectingfrom a restricted food list. These methods may be limiting as thedatabase of foods in computerized programs may not contain the ethnicfoods consumed or the WIC provider may not be

knowledgeable with regard to the ethnic foods and their composition.

Among some Hispanic populations, particularly migrant workers, environmentalfactors such as nonexistent or inadequate housing can cause highintra-individual variability of dietary intake. Limited income alsohas the potential to decrease the variety of the foods. Income isvariable from month to month and can affect dietary intake. In summary,in order to conduct dietary assessments in the Hispanic population,it is recommended that the WIC program promote and support the attitudes,behaviors, knowledge, and skills necessary for staff to work respectfullyand effectively with clients.

The National Coalition of Hispanic Health and Human Services Organizations. 1998. Delivering Preventive Health Care to Hispanics: A Manual for ProvidersWashington D.C.: The National Coalition of Hispanic Health and Human Services Organizations.

Sanjur D. 1995. Hispanic Foodways: Nutrition and Health. Needham Heights, Mass.: Allyn & Bacon.

Presented by Anna Maria Siega-Riz, Ph.D.

Assistant Professor of Nutrition and Maternal and Child Health

University of North Carolina School of Public Health

A Diet Quality Index for Pregnancy (DQI-P) was developed using datafrom the Pregnancy, Infection, and Nutrition (PIN) cohort (n > 3000) in central North Carolina. The goal was to develop a compositethat reflected current nutritional recommendations for pregnancyas well as the Food Guide Pyramid (Shaw et al., 1999). Dietary intakewas assessed during the second trimester using a Food Frequency Questionnaire.The DQI-P includes eight components: percent of recommended servingsof grains, vegetables, and fruits based on the Food Guide Pyramid,percent of dietary reference intakes (DRI) for folate, calcium, andiron (NRC, 1989; Yates et al., 1998), percent of energy from fatbased on the Dietary Guidelines, and dietary variety score basedon the Food Guide Pyramid (Shaw et al., 1999; USDA/DHHS, 2000). Scoresrange from 0 to 80 with each component contributing 10 points.

The DQI-P differentiates diets both quantitatively and qualitatively.The mean score for the population was 51. Higher amounts of grains,vegetables, and fruits and other DQI-P components were associatedwith an increasing DQI-P score. High intakes of nutrients not measuredon the DQI-P, such as vitamin A and vitamin C, were also associatedwith high DQI-P scores. One drawback of the study was that the foodfrequency questionnaire underestimated intake of grains.

Sociodemographic characteristics of the women were also examined.There was no difference in mean score based on ethnicity, howeverseveral significant differences were found based on sociodemographicfactors. Women who were nulliparous, over 20 years of age, over 350percent of the poverty level, and had at least a high school educationhad significantly higher overall scores. Non-Hispanic black womenconsumed more grains compared to non-Hispanic white women. Vegetablesconsumption was higher among non-Hispanic blacks, women over 20 yearsof age, over 350 percent poverty, and women with some college education.Fruit consumption was higher in nulliparous and non-Hispanic blackwomen. Women who had incomes less than 185 percent of the povertylevel, were nulliparous, younger than 30 years of age, non-Hispanicblack, and had less than a college education had greater intakesof iron. These results may reflect nutrition counseling that thesewomen received. Non-Hispanic white women who had higher incomes,were greater than 30 years of age, and were better educated tendedto consume diets lower in fat. Compared to their counterparts, non-Hispanicblack and nulliparous women had higher dietary variety scores. Therewere no differences found in sociodemographic characteristics forcalcium, folate, or diet variety.

Establishing textiles as cut points for the DQI-P may be useful.Women who were in the lowest tertile had a mean diet quality scoreof 37. This equated to meeting 60.2 percent of the Recommended DietaryAllowance (RDA) for folate, 40.7 percent the RDA for iron, consuming36.5 percent of energy from fat, as well as not meeting the recommendedintakes for fruits, vegetables, and grains. Another approach to theDQI-P would be to examine women who do not achieve two-thirds ofthe recommendations on any six of the components. This would be

equivalent to a mean diet quality score of 35.

In the population of pregnant women examined, the DQI-P was qualitativelyand quantitatively differentiated. It has also been reproduced inlater research. An advantage of using the DQI-P is that it assessesoverall variation in diet as opposed to assessing one nutrient. Italso can be used in a computerized assessment. The DQI-P may be auseful public health tool for evaluating the overall diet qualityof pregnant women.

NRC (National Research Council). 1989. Recommended Dietary Allowances, 10th ed. Washington, D.C.: National Academy Press.

Shaw A, Fulton L, Savis C, Hogbin M. 1999. Using the Food Guide Pyramid: A resource for nutrition educators. Available at: http://www.nal.usda.gov/fnic/Fpyr/guide.pdf. Accessed November 30, 1999.

USDA (United States Department of Agriculture)/DHHS (Department ofHealth and Human Services). 2000. Nutrition and Your Health: Dietary Guidelines for Americans, 5th ed. USDA Home and Garden Bulletin No. 232.

Yates AA, Schlicker SA, Suitor CW. 1998. Dietary Reference Intakes: The new basis for recommendations forcalcium and related nutrients, B vitamins, and choline. J Am Diet Assoc 98: 699–706.

Presented by Anna Maria Siega-Riz, Ph.D.

Assistant Professor of Nutrition and Maternal and Child Health

University of North Carolina School of Public Health

The 1990 Institute of Medicine report, Nutrition During Pregnancy, recommended that pregnant women eat nutritious snacks and small tomoderate-sized meals at regular intervals in order to meet the increasednutritional needs during pregnancy (IOM, 1992). This has been translatedinto a recommendation of three meals and at least two snacks perday. Several studies have examined eating frequency and results indicatethat eating three to six meals per day improves glucose toleranceand lipid and lipoproteins profiles in pregnant women.

In order to identify meal patterns of pregnant women and investigatethe relationship between these meal patterns and preterm delivery,an analysis was performed using data from pregnant women (n = 2,065)in the Pregnancy, Infection, and Nutrition Study (Siega-Riz et al.,in press). Meal patterns were created from the reported number ofmeals (breakfast, lunch, and dinner) and snacks consumed per dayduring the second trimester, as well as having the women characterizethe times of meals and snacks on a 24-hour time clock. In accordancewith the IOM's recommendations, an optimal meal pattern was definedas three meals and two or more snacks.

Seventy-two percent of women were found to meet this recommendation.These women had the lowest rates of preterm births. The highest ratesof preterm birth were associated with women who had erratic eatingpatterns. Women who did not meet the IOM recommendation had a 30percent higher risk of preterm delivery (adjusted odds ratio [AOR]=1.30,95 percent confidence interval [CI]=0.96, 1.76). Women who consumedmeals/snacks less frequently than recommended were found to be slightlyheavier prior to conception, were older, and had a lower total energyintake. While there was no difference in risk found to be associatedwith gestational age, women delivering after premature rupture ofthe membranes (AOR=1.87, 95 percent CI=1.02, 3.43) had a higher riskthan those who delivered following preterm labor (AOR=1.11, 95 percentCI=0.65, 1.89).

The above mentioned study supports previous animal model work, whichhas shown an association between frequency of eating and pretermdelivery (Binienda et al., 1989; Fowden et al., 1994). Experimentallyinduced fasts during late gestation have been found to stimulateintrauterine prostaglandin production, uterine contractions, andpreterm delivery. In one study, pregnant sheep who were fasted for12 to 48 hours in the last half of pregnancy were found to experiencehypoglycemia and an increase in prostaglandins leading to uterinecontractions and subsequent preterm delivery. In another study, effectsof a 16-hour extended overnight fast in pregnant women and nonpregnantwomen found that only the pregnant women experienced hypoglycemia,hypoinsulinemia, and ketonemia (Metzger et al., 1982).

Given the results of these studies, the eating frequency of pregnantwomen (n=1,494) in the Pregnancy, Infection, and Nutrition Studyusing reported meal and snack times was evaluated (Herrmann et al.,in press a). The usual time period without food (including an overnightfast) was found to range from 2 to 24 hours. Thirteen percent ofthe women fasted 3 to 8 hours per day, 58 percent fasted 9 to 12hours per day, and 29 percent fasted 13 to 24 hours per day. Whilecontrolling for age, income, race, pregravid body mass index, andcaloric intake, multivariate

logistic regression analysis indicated that the risk for pretermbirth was highest among women who fasted 13 hours per day comparedto women who fasted 8 hours per day (AOR = 3.2, 95 percent CI = 1.1–9.7). These results suggest an association between fasting duringthe second trimester of pregnancy and preterm delivery.

In order to better understand the relationship between fasting andpreterm labor, levels of corticotropin-releasing hormone (CRH), whichcan serve as an indicator of stress during pregnancy and have beenassociated with preterm delivery were studied (Herrmann et al., inpress b). To investigate the relationship between fasting, CRH andgestational age in pregnant women, 24-hour food recalls and CRH weremeasured in 688 pregnant women at 18 to 20, 28 to 30, and 35 to 36weeks. Women who fasted for more than 13 hours the day prior to plasmaCRH analysis were found to have an elevated level of CRH as comparedto women who did not fast. Lastly, women who skipped meals and snackswere found to be at a higher risk for preterm birth even after totalcaloric intake and other factors were considered. Measuring mealand snack intake, as well as the timing of meals and snacks, maybe a very simple public health tool that can be used to evaluatethe prenatal diet and could easily be incorporated into assessmentsof dietary risk for WIC eligibility.

Binienda Z, Massmann A, Mitchell MD, Gleed RD, Figueroa JP, Nathaielsz PW. 1989. Effect of food withdrawl on arterial blood glucose and plasma 12,14-dihydro-15-keto-prostaglandin F2α concentrations and nocturnalmyometrial electromyographic activity in the pregnant rhesus monkeyin the last third of gestation: A model for preterm labor? Am J Obstet Gynecol 160: 746–750.

Fowden AL, Ralph MM, Silver M. 1994. Nutritional regulation of uteroplacental prostaglandin productionand metabolism in pregnant ewes and mares during late gestation.Exp Clin Endocrinol 102: 212–221.

Herrmann T, Siega-Riz AM, Hobel CJ, Aurora C, Dunkel-Schetter C. In press a. Fasting during pregnancy increases risk for elevated maternal corticotropinreleasing hormone concentrations.

Herrmann T, Siega-Riz AM, Savitz DA, Zhou H, Thorp J. In press b. Fasting during pregnancy and preterm birth.

IOM (Institute of Medicine). 1992. Nutrition During Pregnancy and Lactation: An Implementation Guide. Washington D.C.: National Academy Press.

Metzger BE, Vileisis RA, Ravnikar V, Freinkel N. 1982. Accelerated Starvation: and the skipped breakfast in late normalpregnancy. Lancet 2: 588–592.

Siega-Riz AM, Herrmann T, Savitz DA, Thorp J. In press. The frequency of eating during pregnancy and its effect on pretermdelivery. Am J Epidemiol.

Presented by Jill Leppert

Nutritionist, North Dakota WIC Program

North Dakota Department of Health

The state of North Dakota has used the Harvard Food Frequency toolsince 1993 to assess dietary risk in all participants. It is usedfor certification and recertification of women and children. Priorto the adoption of this tool, the state used both the 24-hour dietrecall and a food frequency method. Concerns over time constraints,repetition, and lack of consistent results prompted a search fora better tool. The goal was to streamline the process, have a bettertool that was easy for paraprofessionals to use, and could be usedin a variety of settings. In addition, the tool needed to provideenough information to be used to assess both eligibility and serveas a basis for nutrition education.

The Harvard tool provided all of the information needed. The printoutprovides the number of servings eaten each day in all food groups,the number of servings of vitamin A foods, vitamin C foods, simplesugar foods, and fats. In addition, it provides the percent RecommendedDietary Allowance, calories, and a breakdown of the percentage offat, carbohydrate, and protein. The information was all computerized.Given that the staff was comfortable with computers and they wereavailable in all the agencies, it was a good match.

The tool was piloted in three sites. After 3 months, results werevery positive. The tool takes the client approximately 5–7 minutesto complete and it takes staff approximately 2–3 minutes to enterthe data in the computer. This then provides an immediate 3-pageprintout with the information mentioned above. Paraprofessionalswere easy to train in terms of administering the tool and interpretingthe results. Clients liked the new tool and found it easy to fillout. Approximately 90 percent of clients do a very good job in completingthe form. Approximately 3 percent of clients were not able to fillout the previous food frequency tool and this same group has similarlimitations with this tool. The other 7 percent do not take the timeto fill out the questionnaire because they are either rushed or uninterested.However, this would be the case with any tool that was used.

With the time saved in using the Harvard tool, more time was availablefor nutrition education. Clients liked the printouts they receivedand seemed to be more engaged in the nutrition education. They feltconfident in the results received because they had been computergenerated. Using the client education page has worked out well. Clientsfind the food groups to be helpful and appreciate receiving the information.They seem to like computer-generated information about themselves.North Dakota is currently in the process of updating its computersystem and the new technology will work well. In a new program, nutritionrisk codes could be recognized by the computer and automaticallyidentify individuals at risk, which will further decrease errors.

In North Dakota, the average WIC family has 3.9 people. Sixty-sevenpercent of families have both the mother and father present. Sixty-ninepercent have either one or both parents employed full-time. Sixty-sixpercent of the participants are Caucasian, 15 percent Native American,5 percent Hispanic, 3 percent black, and less than 1 percent Asian.Fifty-seven percent of the families have incomes below 100 percent,20 percent have incomes between 100 and 133, and 23 percent haveincomes between 133 and 135. Forty-seven percent of WIC mothers have a

high school education. Twenty-four percent have less than a highschool education, and 29 percent have more than a high school education.

It is difficult to estimate the number of WIC participants who areat dietary risk for a number of reasons. Primarily, it is an issueof coding and priority. Only the five highest priority codes arelisted in the computer. Since dietary risk codes are not high priority(level 4, 5, or 6) they often do not make it into the computer. Eighty-threepercent of pregnant women are priority level one. Typical priorityone clients would be those who are underweight, overweight, anemic,or had a previous poor pregnancy outcome. Only 17 percent are priorityfour; however not all priority four clients are eligible based ondietary risk. Priority four also includes conditions such as migrancy,homelessness, transfer of certification, and inadequate vitamin intake.For children, approximately 36 percent are coded as priority three(overweight, underweight, anemic, or other specific health problem).Dietary risk is priority five. While 64 percent of children are codedas priority five, this includes other conditions such as inappropriateuse of bottle or other inappropriate feeding practice.

Overall, there is an exceedingly small number of income eligibleindividuals who are not certified for the program because they lacknutrition eligibility—perhaps 1 percent. While Congress did not setup the WIC program this way, it would best if the risk assessmentwere divorced from the eligibility criteria. While assessment isimportant for nutrition education, it would be best if eligibilitywere based on income alone. Lastly, when it does come to selectingand approving a tool for dietary assessment, it would be best iftwo or three different tools were approved that the states couldchoose between.

Presented by Amanda Watkins, M.S., R.D.

Nutritionist, Arizona WIC Program

Arizona Department of Health Services

To date, Arizona has used a food frequency questionnaire for dietaryassessment. The state is currently in the process of automating theWIC program and in an effort to incorporate diet assessment intothe automation process, Arizona reviewed numerous tools and decidedto have a new food frequency questionnaire created. The Arizona Departmentof Health Services has contracted with Dr. Douglas Terrin of theUniversity of Arizona to create a short food frequency tool thatcould be validated. This tool unfortunately did not prove to be validand it was recommended that it not be used for certification purposes.For this reason, the state of Arizona is back at square one.

After reviewing many options, the current tool decided on is a 24-hourrecall, adapted from a tool used by a local agency. It is a papertool that utilizes the food guide pyramid. All age groups and typesof clients can be assessed using one form. The community nutritionworker (CNW) shades in each serving of a food group eaten. Thus,the tool may be used for nutrition education as well. While the toolcan be self administered (the participant fills out what they eaton top and the CNW does the evaluation on the bottom), the most commonmethod in Arizona is for the tool to be completed through an interviewprocess, as there is a high percentage of Hispanics and literacydoes tend to be a problem.

Dietary assessment is not mandatory in Arizona. If another risk isdetermined which qualifies an individual for the program, a dietaryassessment does not need to be performed. Only when the certifieris unable to find a valid risk do they proceed with dietary assessment.Thus, Arizona has a low rate of certification for dietary risk. Thedecision for not making dietary assessments mandatory is not completelyclear; it may be because of the controversy over which tools aretruly valid and it is very seldom that an individual qualifies ondietary risk alone. Generally, if an individual is not eating properly,chances are that they will qualify for another reason such as anemia,underweight, overweight, or other risks. The feeling among many clinicsis that given the validity of the assessments, do not go there unlessnecessary.

Sixty-two percent of Arizona's WIC population is Hispanic. Approximately 30 percent are non-Hispanicwhites, 5 percent are African American, 2 percent are Native American,and 1 percent are Asian. The Native American population appears lowbecause there are actually three separate programs in Arizona: ArizonaWIC program, the Navajo Nation WIC program, and the Intertribal Councilof Arizona WIC program. The Navajo Nation program and the IntertribalCouncil of Arizona WIC programs are the two programs that certifyand service the majority of the Native American WIC population.

To address the needs of the Hispanic population, the dietary assessmenttool is in English on one side and in Spanish on the other. In addition,the combination foods list contains popular Mexican fare. The mostimportant way that the Arizona program meets the needs of the Hispanicpopulation however is that the majority of the staff are not onlybilingual, they themselves are Mexican-American, live in the samecommunity, and many of them have been WIC participants themselves.The CNWs are truly their peers. They not only perform the certifications,they also perform the dietary assessments, and in most cases alsoprovide the nutrition education. The

CNWs are required to have a GED or high school diploma. The majorityhave had no other nutrition education other than what they have receivedthrough WIC training. For this reason, the staff drives the requirementswe have for a dietary assessment tool.

Based on different focus and working groups of state and local WICagency staff, several things need to be considered in designing adietary assessment tool. In order to be useful to the state of Arizona,it needs to be fast and easy to administer—3 minutes would be ideal.A new enrollment appointment is allotted approximately 20 minutes.In this time period quite a bit needs to get done (blood work, incomedocumentation, height, weight, health history questions, etc.). Inaddition, the clinic flow needs to be kept in mind. Diet assessmentsare only performed if no other risks are found. Results of the assessmentmust be able to be converted into portion sizes according to theU.S. Department of Agriculture Food Guide Pyramid, be used for dietarycounseling, measure changes in dietary patterns, be easily incorporatedinto Arizona's automated system and, preferably, be a food frequencyquestionnaire. Ideally, the tool would be one which would be intervieweradministered and information obtained would be input into the computerby the CNW. The ideal tool has yet to be created.

Presented by Ann Barone, LDN

Nutritionist, Rhode Island WIC Program

Rhode Island Department of Health

In the state of Rhode Island, the dietary risk assessment processhas been in a state of revision. While a 24-hour recall and foodfrequency tool continue to be used, the food frequency tool is nowbased on the food guide pyramid rather than being nutrient based.In addition, some of the required servings have been adjusted (e.g.,calcium serving size and protein requirements for children). Otherchanges were made in terms of combining groups such as inadequateintake of fruits and vegetables.

In determining the tool to be used for Rhode Island, assessment toolswere collected and examined from many states. A committee of WICagency nutritionists from around the state examined the tools anddetermined what was needed for the Rhode Island population. The foodguide pyramid is the base of the tool. An assessment tool was modifiedfor each group (e.g., pregnant, breastfeeding, and children), including1–2 year olds.

In Rhode Island, every person who applies for WIC has a dietary assessmentperformed. An average WIC appointment takes approximately 30 minutesper participant. The tools are administered by a nutritionist, whohas a minimum of a bachelor's degree. Completing both a 24-hour recalland a food frequency does not leave much time for education. Together,the assessment and evaluation take approximately 15 minutes. It isthe same individual who performs the assessment that does the education.

Approximately 20 to 30 percent of the population in Rhode Islandis Hispanic and less than 5 percent is African American. There isa fairly large Portuguese population in one part of the state. Thirty-sevenpercent of participants qualify for WIC based on dietary risk alone.Only a very small percentage of individuals who apply for the WICprogram are not deemed eligible. Staff will generally find some reasonwhy an individual qualifies. In terms of dietary risk, if an individualis short a serving or over by a serving, they are eligible basedon dietary risk.

Presented by Carol Rankin, M.S., R.D., L.D.

Nutritionist, Mississippi WIC Program

Mississippi State Department of Health

In Mississippi, the tools used to assess dietary risk are basic,but provide the information needed to determine whether an applicantis at nutritional risk. Three separate forms are used for dietaryassessment of women, infants, and children. A food frequency toolis on all forms and a 24-hour recall is on all except for the formfor the infant. There is a section for inappropriate feeding practicesas well, which are used as talking points for the counselor and canbe checked off to indicate risk. The dietary assessment is used forboth determination of eligibility and nutrition education.

For infants, food frequencies are checked against the recommendations.If any group is missing or there are two inappropriate food groups,then the infant is certified based on inappropriate diet. Inappropriatefeeding practices are broken down for the different ages: 0 to 3months, 4 to 5 months, 6 to 9 months, and lastly, 10 to 12 months.The dietary assessment forms for women and children include boththe food frequency and the 24-hour diet recall. Children are dividedinto age groups of 1 to 3 years of age and 3 to 5 years of age.

The tools used for nutrition assessment were chosen in order to fulfillU.S. Department of Agriculture (USDA) and Mississippi WIC requirementsfor the specific populations served. Accepted dietary guidelines,food guide pyramid recommendations, and other established feedingpractices such as American Academy of Pediatrics recommendationswere considered. In addition, two other factors contributed to theselection of the nutrition assessment tools: ease of obtaining usableinformation in a format for busy clinics and ease of use by nonnutritionprofessionals.

Additional nutrition assessments are used for pregnant women andinfants who are enrolled in the perinatal high-risk and infant servicesprogram. This is a program where a team of health professionals casemanage high-risk, Medicaid-eligible patients. The team consists ofa nutritionist, nurse, and social worker. The pregnant woman is followedthrough 60 days postpartum and the infant until 1 year of age. Allof these patients also qualify for the WIC program. The nutritionassessment for these special populations is more in-depth, requiringadditional calculations of calorie levels, growth parameters, andquestions about food security and preparation. The WIC nutritionstaff use the information in determining WIC eligibility and in orderto provide nutrition education targeted to the needs of the patient.

The forms may appear complicated, but once training is received theforms are simple to use. They are used by paraprofessionals, nurses,and lactation specialists. It takes about 10 minutes to completeone of the dietary assessment forms. There is then an additional5 to 10 minutes for nutrition education. Mississippi WIC sites donot currently have computers but are expected to within the next1 to 2 years.

High risk applicants and those who are certified based on inappropriatedietary intake receive a diet history during every clinic visit.All other clients should get a nutrition assessment at least onceper year. If an individual is certified based on another type ofrisk, they may not receive a dietary assessment. However, if an individualmeets no other risk and does have a

dietary assessment, it is rare that they would not be certified.According to USDA figures, 22 percent of WIC clients in Mississippiqualify for the program based on dietary risk.

Presented by Robert Greenstein

Executive Director, Center on Budget and Policy Priorities

Washington D.C.

It has been suggested by many over the years that additional nutritioncriteria on which eligibility is based be done away with. However,for several reasons, the WIC program has continued to operate, thrive,and produce significant positive health outcomes with program eligibilitybased on both income and nutrition criteria.

Overall, research on the impact of the WIC program on diet and healthover the last quarter century has shown striking results in termsof its impact on birth outcomes and lowering the incidence of lowbirth weight. In 1992, the General Accounting Office (GAO) did anevaluation synthesis in which it concluded that the WIC program reducedthe incidence of low birth weight by 25 percent and very low birthweight by 44 percent (USGAO, 1992). While these figures are not precisedue to questions on selection bias and other matters, when lookingat the body of research as a whole, it is pretty clear that the findingsare robust and the that the direction is strong in terms of the significanceof effects on reducing the incidence of low birth weight. Overall,the research has led to findings that low-income women who receiveWIC during pregnancy have better birth outcomes than low-income womenwho do not.

The WIC program also appears to reduce the incidence of anemia inchildren. Researchers at the Centers for Disease Control (CDC) comparedthe anemia levels of infants and children at the time of initialenrollment in WIC to their anemia level at the time of WIC follow-upvisits. Significant reductions in anemia rates were found for mostage groups of infants and young children in most years for whichdata were examined. Researchers reported that the prevalence of anemiawas consistently higher for children seen at initial visits thanfor those seen at WIC follow-up visits (Yip et al., 1987). BarbaraDevaney and others who have looked at the data from CDC have concludedthat the evidence is pretty strong in terms of the WIC program'sanemia-reducing effects (Devaney, 1998).

An area that until recently was the most controversial was whetheror not WIC improved the diets of children. Findings from the nationalWIC evaluation of the mid-1980s found very strong effects in reducingthe frequency of low intakes of certain nutrients among infants andchildren (Rush et al., 1988). While no one was quite sure what tomake of the findings because the study had been marred by significantselection bias problems, Devaney and others noted that the selectionbias problems were more likely to understate, rather than overstate,WIC's effects because the control group was better off than the treatmentgroup (Devaney, 1998). Nevertheless, the problems were significantenough to cast doubt on the findings that WIC improved the dietsof children. In the last 3 years, however, this has changed in partbecause of a better controlled study by Rose, Habicht, and Devaney(Rose et al., 1998). These researchers found that WIC had significanteffects in increasing preschoolers intakes of ten nutrients, includingiron, zinc, and vitamin E. These three nutrients are among the fourmost frequently deficient in the diets of low-income preschoolers.The increases in WIC participants in iron and zinc were particularlylarge.

Within the past month, a new study was released by Oliveira and Gundersonof the Economic Research Service (Oliveira and Gunderson, 2000).This study, which uses data from the 1994 to 1996 Continuing Surveyof Food Intake by Individuals (CSFII), found that

participation in WIC significantly increases children's intake of iron, folate and vitamin B₆. With the addition of this study, there is now a body of evidencethat, when looked at together, suggests that WIC both reduces anemiaand improves diets among children. In particular, WIC apparentlyreduces the incidence of low intakes of particularly important nutrientssuch as iron among low income children. In addition, a review ofthe literature conducted and published by Leighton Ku of the UrbanInstitute, prior to the study by Oliveira and Gunderson, found similarresults (Ku, 1999). While noting that there are holes in the research,Ku concluded that taken as a whole, it is hard to think of any publicprogram with so consistent a body of positive research findings.

At the Center for Budget and Policy Research, the gamut is coveredwhen it comes to means tested benefit programs at both federal andlocal levels (e.g., health, income, supports, housing programs).Research is constantly being reviewed and evaluated. Some programsare found to be less effective, others are found to be more effective,and some are found to need various changes. However, there is notanother means-tested program for which the literature of efficacyapproaches that of the WIC program. WIC is an important remedialand preventive program.

Inadequate dietary pattern as a nutrition risk criteria to determine eligibility for WIC hasbeen described as being applied in an imprecise and sometimes loosemanner. However, to some degree, policy makers have known that inadequatedietary pattern was imprecise and loosely applied and, to some degree,that is the kind of criteria they have wanted. At two points, inthe late 1970s and again in the mid-1980s, policy makers gave seriousconsideration to eliminating nutrition risk as an eligibility criterionfor pregnant women or for pregnant women and infants. Ultimately,this was not done for a few reasons, which could largely be put underthe heading of optics. There was very much a sense on the part of both policy makers andstate WIC directors that if income was the only eligibility criterion,it would lead to the perception of WIC as merely another welfare program as opposed to a health program. It was felt that this would be damaging to WIC's political viability, its image in the community, its effectivenessin reaching working families who do not want to be on welfare, andlastly, that it might result in some agencies paying less attentionto providing the dietary, nutrition, and other health informationand counseling needed.

The decision not to drop the nutrition risk criteria was aided bythe fact that in exploring the elimination of the nutrition riskcriteria, policy makers found that hardly any pregnant women whomet the income criteria and applied for the program were turned awaybecause of not meeting the nutritional risk criteria. This is whatmany policy makers wanted. A nutrition assessment could be performedand the image of a health program rather than a welfare program wouldbe maintained, and yet low income individuals for whom there wouldbe a preventive value would not be denied entry into the program.This raises an important question for this committee: How can thenecessary information be collected in an effective manner while avoidinga significant narrowing of the criteria that could possibly makea few million individuals who are currently eligible for WIC, ineligible?

Shrinking the eligible pool of individuals for WIC by narrowing theeligibility criteria would seem reasonable if this meant openingslots for needier individuals. However, while this may have beenthe case some years ago, it is not the case now. WIC actually hasas much money right now as it needs to serve virtually everybodywho walks in the door and applies. The program has substantial amountsof carry-over funding. It has been funded for the last 3 or 4 yearsby the Congress to serve between 7.4 and 7.5 million women, infants,and children and each year it has served 7.3 million individuals.Last year, there was close to $200 million of

available WIC funds that were unspent and carried forward to thefollowing year. While there sometimes are occasional problems ofa few months when there is too much money in a given state and notenough money in another, it can not be said that substantial numbersof people are trying to get into WIC and are turned away.

Another argument for narrowing the eligibility criteria would beif one thought that the eligibility criteria was too loose and thatmoney could be better spent on other important functions. There area few problems with this argument. The first is that narrowing theeligibility criteria could weaken WIC's effectiveness as a preventiveprogram. People who meet the income test but do not immediately meetone of the more rigorous nutritional risk screens could be deniedWIC, only to meet the nutritional screen subsequently if their nutritionalstatus deteriorates. In addition, the current budgetary situationis not one of deficits that threaten the economy. Rather, the pictureis one of growing surpluses. In the next 4 weeks, both OMB and theCongressional Budget Office (CBO) will release new budget forecaststhat place the new surplus estimate, outside Social Security, atmore than $1.5 trillion over the next 10 years. In other words, OMBand CBO are about to double the size of the surpluses they projectedonly 4 months ago. Congress already is preparing to start expendingmore money on both the tax and the spending side. A plethora of taxbills that cost significant sums is starting to move through theHouse and Senate, the latest of which would ultimately reduce revenue$50 billion a year by eliminating the estate tax, even though thattax applies to the estates only of the wealthiest 2 percent of peoplewho die in the United States. The point is that if one were to narrowsignificantly the eligibility criteria for WIC, the result wouldsimply be fewer people served and a reduction in WIC funding levels.Where would the money saved go? These days, a bigger tax cut wouldbe the likely outcome.

The idea of WIC becoming an entitlement program is politically anonstarter. As previously mentioned, there are currently WIC fundsavailable that are not being spent, and participation in the programhas dropped by 3 percent since its peak in 1996. In 1996–1997, WICparticipation reached its peak of approximately 7.4 million participants.Now participation is 7.2 million. (In comparison, food stamp participationhas dropped 40 percent since 1994.) The drop in WIC participationis likely due primarily to the economy; there are fewer numbers oflow-income individuals now. In addition, in all assistance programs,the working poor have a lower participation rate than the welfarepoor.

Making WIC an entitlement was considered by policy makers in thelate 1970s and mid-1980s. It was not something that proved to bea viable option politically then. It is even less likely to be considereda viable option now, given the more conservative Congress today.The chance that it would be seriously considered is near zero.

It is important to look at what improvements can made in the nutritionalrisk criteria, particularly if such improvements can improve informationfor WIC clinics and provide participants with more effective counselingand other services. However, one would hope that the Committee onDietary Risk Assessment in the WIC Program keeps the Hippocraticoath in mind if it considers options that could significantly narrowthe pool of low-income women, infants, and children who are eligiblefor WIC. The principal effect of such an approach would be fewerlow-income individuals being provided with WIC benefits and servicesand more money made available for tax cuts likely to accrue primarilyto relatively high-income individuals.

Devaney B. 1998. Commentary. In Rossi PL, ed, Feeding the Poor. Washington, D.C.: AEI Press.

Ku L. 1999. Debating WIC. The Public Interest Spring: 108–112.

Oliveira V, Gunderson, C. 2000. WIC and the Nutrient Intake of Children. Food and Rural Economics Division Research Report No. 5. Washington, D.C.: Economic Research Service, U.S. Department of Agriculture.

Rose D, Habicht JP, Devaney B. 1998. Household participation in the Food Stamp and WIC programs increases the nutritional intakes of preschool children. J Nutr 128: 548–555.

Rush D, Sloan NL, Leighton J, Alvir JM, Horvitz DG, Seaver WB, Garbowski GC, Johnson SS, Kulka RA, Holt M. 1988. The National WIC Evaluation: Evaluation of the Special SupplementalFood Program for Women, Infants, and Children. V. Longitudinal study of pregnant women. Am J Clin Nutr 48: 439S–483S.

USGAO (United States General Accounting Office). 1992. Early Intervention: Federal Investments Can Produce Savings. Washington, D.C.: U.S. Government Printing Office.

Yip R, Binkin NJ, Fleshood L, Trowbridge FL. 1987. Declining prevalence of anemia among low-income children in the UnitedStates. J Am Med Assoc 258: 1619–1623.

Presented by Lynn Parker, M.S.

Director of Child Nutrition Programs and Nutrition Policy

Food Research and Action Center, Washington D.C.

In considering how food insecurity applies to WIC, it is importantto understand the history of where this concept came from. The ideaof measuring food insecurity started in the United States in the1980s during the period of recession when there were major cut backsin federal nutrition programs. Many more people were coming in foremergency food assistance and service providers began to documentenormous jumps in need. Concerns, however, were discounted by policymakers and there was a strong desire around the country to convincethem that a hunger problem not only existed, but was increasing,and had negative consequences for the country.

The Food Research and Action Center, working in partnership withthe Connecticut Association for Human Services, developed a systematicapproach to studying the problem of hunger among families with children,which was called the Community Childhood Hunger Identification Project(CCHIP). It was the first time a group attempted to develop an objectivemeasurement of hunger that could be used in a national survey (Wehleret al., 1995). Hunger was defined as food insufficiency due to constrainedresources, not voluntary hunger, but involuntary hunger because ofnot enough money to buy food. A survey was developed for familieswith children under the age of 12. It consisted of a series of eightquestions and based on the results, families were categorized ashungry, not hungry, or at risk of hunger. This type of survey was needed as it would not have been feasibleto measure blood levels of nutrients or to administer 24-hour recalls.These measures would have been incredibly time consuming, extremelyexpensive, and would have required a level of training on the partof interviewers that would not have been feasible with the type ofcommunity groups involved.

Results from the surveys showed that a highly significant numberof children were hungry and that families were food insecure. Inaddition, a relationship was found between hunger and reported infections,fatigue, irritability, headaches, and ear infections and colds amongkids. We also found that parents are the first to be hungry and childrenare the last. This has come through again and again in national surveys.When children are found to be food insecure, it is a very severeproblem in the family.

We also learned that who the interviewer is and who the respondentis can affect the response. There appeared to be a difference inresponse because our interviewers were paraprofessionals who weretrained to carry out the survey. Parents may not be as truthful ifthey are afraid their child will be taken away if they admit to agovernment interviewer that they cannot feed them. In addition, individualsliving in rural areas were less likely to admit that they were hungrythan those who lived in urban areas.

At the same time that the CCHIP survey was being used, work was alsobeing done by researchers at Cornell University, and the Life SciencesResearch Office (LSRO) came out with definitions on how to definehunger, food security, and food insecurity (see Box 1). Also during this time, the Nutrition Monitoring Act was signedinto law. The Act had a provision in it that required the federalgovernment to develop some kind of measure of food insufficiency. In 1994,

the Department of Health and Human Services, United States Departmentof Agriculture, and the Census Bureau got together and developedwhat is now the food security module, which is an annual part ofthe Current Population Survey. The population survey is given outto approximately 58,000 households monthly (Hamilton et al., 1997).The food security module has been included in the population surveyfor one month of the year, every year since 1995. (In 1998, resultsshowed that 19 million adults and 12 million children were in householdsclassified as food insecure.) There are several categories of food insecurity based on severity. Even individuals who are classifiedas “food secure” may still have indicated on the survey that they worry aboutfood running out or that the food they bought did not last. The questionson the survey are listed in the typical order in which they may beanswered affirmatively (Table 1). The conservative nature of this measure is evident here.

Results indicated that approximately 10 percent of the householdssurveyed were considered food insecure—two-thirds of them withouthunger and one-third with hunger. Households that were more likelyto be found food insecure were those with children, households headedby single females, and black or Hispanic households. Inner city andrural families were also more likely to be food insecure than suburbanhouseholds and food insecurity ran higher in the southern and westernparts of the United States than it was in any other regions of thecountry.

Questions from the food insecurity module or various combinationsof them are now being included or are in the process of being includedin several national surveys such as the National Health and NutritionExamination Survey, the Continuing Survey of Food Intake by Individuals(CSFII), and the National Behavior and Knowledge Survey. Shortly,there will likely be good research data to answer questions regardingfood insecurity and how it relates to dietary behavior and nutritionalstatus.

In 1996, the report from the IOM Committee on the Scientific Evaluationof WIC Nutrition Risk Criteria found limited evidence to evaluatethe causal links between food insecurity and nutrition and healthrisk. They also found insufficient scientific evidence to selecta cut-off point for WIC eligibility based on food insecurity. However,they did recommend use of the food insecurity measure at some point.

There have been a number of studies showing an association betweenfood insecurity and dietary risk in terms of nutritional inadequacyin a number of nutrients. Recent research by Dixon et al. (in press)showed low fasting levels of vitamin E and C in relationship to foodinsecurity. This was the first time a biochemical measure has beenrelated to food insecurity. It has also been interesting to notethat the research has shown a deeper and broader impact of food insecurityon nutrient levels in women than was originally thought. Cristofarand Basiotis (1992) looked at data from the CSFII for women aged19 to 50 and children 1 to 5. They were able to demonstrate a relationshipbetween food security status and nutrient intakes for women. Whilethey did not see a strong relationship with children, it was notsurprising given the demonstrated protective factor placed on children(i.e., adults apprear to sacrifice eating first before allowing theirchildren to eat less or go hungry.) Kendall, Olson, and Frongillo(1996) used the RadimerCornell Hunger and Food Insecurity measureand found a relationship between food insecurity and a lower consumptionof fruits and vegetables and lower levels of vitamin C, potassium,and fiber in the diet. They also saw much more disordered eatingpatterns in food insecure households. Rose and Olivera (1997) alsodemonstrated that in all nutrients looked at, adult women who werefood insufficient did worse in terms of nutritional adequacy. Tarasukand Beaton (1999) have also shown similar results using a slightlyaltered survey in Canada.

Dr. Christine Olson (2000) recently reported on a relationship betweenfood insecurity and body mass index (BMI). She found that women whowere food secure and women who were food insecure with hunger had,on average, very similar BMIs that were not indicative of obesity.The women who were food insecure without hunger were much more likelyto be obese. She theorizes that food deprivation may lead to over-eatingat times when food insecure people do have enough food or have enoughmoney to buy food. If an individual is very poor, chronically food-deprived,and lacking the opportunity to overeat, the individual will havea lower BMI. However, if the individual is going through the constantups and downs of having enough and not having enough to eat, it maylead them to adopt unhealthy eating habits that can lead to obesity.Dr. Olson's conclusion was that ending health disparities relatedto chronic disease will require ending food insecurity.

There have also been a number of studies looking at the impact offood insecurity on psychosocial indicators in school-aged children.Associations have been found with increased depression, anxiety,and inattentiveness in class. There is also research showing an impacton children's ability to learn, explore, and interact with theirenvironment.

Overall, the research is important to consider in terms of the WICprogram. WIC has been very successful at being a preventive program.Many of the people who come to the WIC clinic may be food insecureone day and not the next. They tend to go in and out of poverty.It may be as little as one child needing a new coat that could leadthe family to go for a few weeks with less food than needed. It isimportant to remember what food insecurity looks like in these families.There has been work done on adapting the food insecurity survey tomake it shorter. It may be possible to build only one or two questionsinto the dietary assessments in WIC settings. A few food insecurityquestions would be helpful. Just as dietary assessment questionshelp to guide nutrition counseling, food insecurity questions couldlead a nutritionist to recognize when a family may need other assistanceprograms such as food stamps, Temporary Assistance to Needy Families,Medicaid, or emergency food. Using these questions, the nutritionistcould assist in increasing the food security in a family as wellas helping them improve their diet.

Cristofar SP, Basiotis PP. 1992. Dietary intakes and selected characteristics of women age 19–50 and their children ages 1–5 years by reported perception of food sufficiency. Nutr Ed 24: 53–58.

Dixon LB, Radimer K, Winkeeby M, Krebs-Smith S. In press. Differences in foods, dietary nutrients, and serum nutrients betweenfood insufficient and food sufficient low income adults, NHANES III,1988–1994.

Hamilton WL, Cook JT, Thompson WW, Burron LF, Frongillo EA, Olson CM, Wehler CA. 1997. Household Food Insecurity in the United States in 1995: Summary Reportof the Food Security Measurement Project. Washington, D.C.: Food and Consumer Service, U.S. Department of Agriculture.

Kendall A, Olson CM, Frongillo EA. 1996. Relationship of hunger and food insecurity to food availability andconsumption. J Am Diet Assoc 96: 1019–1024.

Rose D, Oliveira V. 1997. Nutrient intakes of individuals from food insufficient householdsin the United States. Am J Public Health 87: 1956–1961.

Tarasuk V, Beaton GH. 1999. Women's dietary intakes in the context of household food insecurity. J Nutr 129: 672–679.

Olson CM. 2000. Presentation at the National Nutrition Summit, Food Insecurity andthe Health of Working Americans, May 30, 2000, Washington, D.C.

Wehler CA, Scott IS, Anderson JJ, Summer L, Parker L. 1995. Community Childhood Hunger Identification Project: A Survey of Childhood Hunger in the U.S.Washington, D.C.: Food Research and Action Center.