Click for next page ( 84


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 83
7 Estimating Eligibility Based on Meeting Nutritional Risk Criteria To be fully eligible for WIC benefits, applicants who meet categorical and income eligibility requirements also must be deemed nutritionally at risk by meeting at least one nutritional risk criterion. Five types of nutri- tional risk criteria are considered in determining whether a person is nutri- tionally at risk: anthropometric, biochemical, clinical/health/medical, di- etary, and other. Examples of each type of risk appear in Table 7-1, as does the number of criteria considered for each type. Each nutritional risk crite- rion includes an indicator of nutritional risk and a cutoff point. For ex- ample, for young children, a blood lead value equal to or greater than 10 micrograms per deciliter is an approved criterion for nutritional risk, so that a child with a blood lead level above the 10 microgram level would qualify as nutritionally at risk. To determine whether an applicant meets at least one of the nutrition risk criteria, a competent professional authority at the local WIC office administers a nutritional risk screen to the applicant. For example, an applicant's height, weight, and hemoglobin values are measured and com- pared with the cutoff values for the respective nutritional risk criteria. Checks for health conditions that confer eligibility are also made. In most cases, the staff member asks the applicant or caregiver for information about the applicant's food intake. Generally, this involves either a 24-hour diet recall (asking what foods and beverages were consumed the previous day, and in what amounts) or a food frequency questionnaire (obtaining infor- mation on the frequency with which the applicant consumed specified foods and the portion sizes usually consumed). 83

OCR for page 83
84 ESTI~TING ELIGIBILI>~D PAR TICIPATION FOR THE ~CPROG~ TABLE 7-1 Types of Nutritional Risk Criteria Used in WIC, Numbers of Criteria, and Examples Type of Risk Criteria Number of Criteria by Typea Anthropometric Biochemical Cl ini cal/h ealth /medical Dietary Other 18 43 19 14 Underweight, overweight Low hematocrit Diagnosed diabetes mellitus Food intake that does not meet food guide pyramid specifications, improper dilution of formula Regression, migrancy, homelessness aNumbers are based on WIC Policy Memorar~d?vm 98-9 (U.S. Department of Agricul- ture, 1998~. Some criteria have subcriteria, such as specific kinds of gastrointestinal disorders. Some but not all criteria apply to every categorical group (women, infants, and children). For example, many of the criteria applicable to infants do not apply to any other category. To account for the nutritional risk requirement in estimating WIC eligibility, the current USDA method adjusts the estimated number of in- come-eligible persons in each categorical group downward using the ad- justment factors listed in Table 7-2. The results are estimates of fully eli- gible individuals in each category. The adjustment factors for all categories TABLE 7-2 Adjustment Factors Currently Used to Estimate the Number of Income-Eligible People Who Also Meet Nutritional Risk Eligibility Criteria Category Adjustment Factor Infants Children Pregnant women NonbreastSeeding postpartum women BreastSeeding postpartum women 0.950 0.752 0.913 0.933 0.889

OCR for page 83
ESTIMATING ELIGIBILITY BASED ON NUTRITION RISK CRITERIA 85 except infants were based on estimates of nutritional risk for income-eligible individuals obtained from the first WIC Eligibility Study (WES I), which used data collected in the early 1980s (U.S. Department of Agriculture, 19871. The procedure for determining these adjustment factors was to de- velop a list of the nutritional risk criteria most commonly used by the states (modal nutritional risk criteria) and to use nationally representative data sets to estimate the proportion of income-eligible women, infants, and young children who met one or more of these criteria. Modal nutritional risk criteria were used because, until 1998, regulations allowed each state to establish its own nutritional risk criteria. Prior to 1998, the states used different numbers and kinds of indicators of nutritional risk and different cutoff points. To produce the adjustment factors, the study combined data from two surveys the 1980 National Natality Survey and the 1978-1980 National Health and Nutrition Examination Survey. In 1991, USDA increased the adjustment factor for infants from the WES I value of 0.752 to 0.950. The higher value was adopted to account for the high percentage of infants who met a "predisposing" nutritional risk criterion (and thus were WIC eligible) based on "other" risk specifically, their mother's participation or eligibility for participation in WIC (see the discussion of criterion 701 in the section "Method for Infants to Age 1 Year". WES II proposed higher adjustment factors for the nutritional risk of women and children, but USDA has not adopted them (U.S. Depart- ment of Agriculture, 1999a). This chapter critiques the current method used to make national esti- mates of the proportions of income-eligible persons who meet at least one nutritional risk criterion (and thus are fully eligible) and discusses alterna- tive methods for estimating those who meet a criterion. In discussing alter- natives, the difficulties of assessing nutritional risk in the field and of esti- mating the prevalence of nutritional risk with survey data are considered. To give a conservative estimate of the level of nutritional risk in the in- come-eligible population, lower bound estimates of the prevalence of nu- tritional risk are presented. We find that for all groups for which quality data are available, even the lower bound estimates of the prevalence of nutritional risk are very close to 100 percent. For one group, children ages 1 to 2, data limitations prevent us from presenting lower bound estimates. The chapter also contains a discussion of the costs and benefits of using a dietary risk screen to determine eligibility. Finally, it provides recommen- dations regarding methods to estimate the percentages of categorically eli-

OCR for page 83
86 ESTI~TING ELIGIBILI>~D PAR TICIPATION FOR THE ~CPROG~ gible and income-eligible individuals who meet at least one nutritional . . . . risk crlterlon. CRITIQUE OF CURRENT METHOD The panel's Phase I report concludes that "the estimates of nutritional risk currently used may not accurately reflect the actual number at nutri- tional risk" (National Research Council, 2001:61. That report identifies a number of concerns with the current USDA nutritional risk adjustment factors and with the adjustment factors estimated in WES II (U.S. Depart- ment of Agriculture, 1999b). These concerns include the use of old data, the method used to account for variation in nutritional risk criteria across states, the use of data on only one day of diet recall, and the method used to combine separate estimates of risk from different data sources. The adjustment factors for the categorical groups other than infants need to be reconsidered for three major reasons: (1) they are based on sur- vey data that are more than 20 years old; (2) states have adopted a relatively standardized set of anthropometric, biochemical, clinical/health/medical, predisposing, and certain dietary risk criteria from an approved list (U.S. Department of Agriculture, 19981; and (3) a recent Institute of Medicine (IOM) report recommends presuming that all income-eligible women and children ages 2 years and older are at dietary risk (Institute of Medicine, 20021.1 As shown in Table 7-1, the term dlietary risk refers to a type of risk that encompasses many specific criteria. All the dietary criteria relate to some aspect of dietary intake. The recommendation of the IOM is made in a report that does not address infants or children under age 2 years, but the presumption of dietary risk for women and children at least 2 years of age also would be a presumption of nutritional risk. USDA has not yet taken an official position on the IOM recommendation concerning presumption of dietary risk. This recommendation is based on the IOM report's two major findings: (1) studies suggest that nearly all children ages 2 years and older and all women in the childbearing years are at dietary risk because they fail to meet the dietary guidelines as translated by recommen- dations of the food guide pyramid and (2) no known assessment methods can identify or hold promise of accurately identifying the small percentages of women and children who do meet the proposed criterion "failure to meet dietary guideline" with the limited amount of on-site information about food intake that is available to WIC field staffs.

OCR for page 83
ESTIMATING ELIGIBILITY BASED ON NUTRITION RISK CRITERIA 87 POSSIBLE METHODS TO ESTIMATE NUTRITIONAL RISK The standard method of estimating the prevalence of a risk is to operationalize the definition of risk in quantitative terms (by specifying an indicator and a cutoff value) and use survey data to determine the percent- age of individuals who fall above or below the specified cutoff value. An example of a nutritional risk prevalence is the percentage of children ages 1 to 5 years who have been diagnosed with diabetes mellitus. Since nutri- tional risk may take many forms, however, there are many approved nutri- tional risk criteria for each categorical group served by WIC. This means that the method used to estimate the prevalence of nutritional risk within a categorical group must consider the risk of failing to meet at least one of the many criteria applicable to that group. The panel considered new approaches to estimate the risk of meeting at least one nutritional risk criterion in the income-eligible population. Different data sources were considered. As we discuss in this chapter, the lack of relevant national data about dietary risk of children ages 1 to 2 years limits our ability to estimate the percentage of these children who meet income eligibility requirements but not nutritional risk criteria. For the other groups, the panel made what we consider to be conservative, lower bound estimates of the prevalence of nutritional risk. The following section discusses how these estimates were made and presents our lower bound estimates. National Data Sets for Estimating Risk Prevalence A big obstacle to estimating the proportion of the income-eligible population that meets at least one criterion for nutritional risk is the lack of a single data source that contains information regarding all the risk criteria for the relevant population groups. Two nationally representative surveys that measure many nutritional risks the Continuing Survey of Food In- take by Individuals (CSFII) and the National Health and Nutrition Exami- nation Survey (NHANES) provide data related to the nutritional risk criteria. Neither survey, however, covers all of the nearly 100 approved nu- tritional risk criteria. For example, neither CSFII nor NHANES provides data to estimate the percentage of income-eligible people with food aller- gies, infectious disorders, pica, or severe nausea and vomiting, which are risk criteria for one or more categorical groups. Table 7-3 lists the indicators for approved nutritional risks, and identifies which survey, if any, provides

OCR for page 83
88 ESTIMATING ELIGIBILI~YANDPARTICIPATIONFOR THE WICPROGRAM TABLE 7-3 Available Data Related to Estimating Nutritional Risk, by Survey Categorical Groups to Nutritional Risk Indicatora Which Related Data Available, by Survey (Code, Description) for Criterion Is Criterion Applicable NHANES CSFII 101-103,Lowweightforheight Each Measured Self-reported 111-114, High weight for Each Measured Self-reported height 121, Short stature 134, Failure to thrive 135, Inadequate growth 141, Low birthweight 142, Prematurity 151, Small for gestational age 152, Low head circumference 153, Large for gestational age 201, Low hematocrit/ Infants, children Infants Infants, children Infants Infants Infants Infants Infants Each low hemoglobin 211, Elevated blood lead Each 311, History of preterm delivery Pregnant women 312, History of low birthweight Pregnant women 321, History of spontaneous Pregnant abortion, fetal or neonatal women loss 331, Pregnancy at a young age Pregnant women 332, Closely spaced pregnancies Pregnant women 333, High parity and young age Pregnant women 334, Lack of adequate prenatal Pregnant care women 335, Multifetal gestation Pregnant women 336, Fetal growth restriction Pregnant women Measured Self-reported

OCR for page 83
ESTIMATING ELIGIBILITY BASED ON NUTRITION RISK CRITERIA 89 TABLE 7-3 Continue] Categorical Groups to Nutritional Risk Indicatora Which Related Data Available, by Survey (Code, Description) for Criterion Is Criterion Applicable NHANES CSFII 337, History of birth of an infant who is large for women gestational age 338, Pregnant woman currently Pregnant breastSeeding 339, History of birth with Pregnant Pregnant women nutrition-related congenital women or birth defect 341, Nutrient deficiency diseases Each 342, Gastrointestinal disorders Each 343, Diabetes mellitus Pregnant Yes women 344, Thyroid disorders Each 345, Hypertension, chronic or Each Yes pregnancy induced 346, Renal disease Each 347, Cancer Each Yes 348, Central nervous system Each disorders 349, Genetic and congenital Each disorders 350, Pyloric stenosis 351, Inborn errors of metabolism 352, Infectious diseases 353, Food allergies 354, Celiac disease 355, Lactose intolerance 356, Hypoglycemia 357, Drug-nutrient interactions 358, Eating disorders Infants Each Each Each Each Each Each Each Pregnant women YYes :es 359, Recent major surgery, Each trauma, burns 360, Other medical conditions Each Yes Yes 361, Depression Each corltirlued

OCR for page 83
90 ESTIMATING ELIGIBILI~YANDPARTICIPATIONFOR THE WICPROGRAM TABLE 7-3 Continue] Nutritional Risk Indicatora (Code, Description) for , - . . Criterion Categorical Groups to Which Criterion Is Applicable NHANES CSFII Related Data Available, by Survey Each 362, Developmental, sensory, or motor disabilities interfering with the ability to eat 371, Maternal smoking Pregnant women 372, Alcohol and illegal drug Pregnant use 381, Dental problems 401, Failure to meet USDA/DHHS Dietary Guidelines for Americans 402, Vegan diets 403, Highly restrictive diets Each Each Each Each women 411, Inappropriate infant Infants feeding practices 412, Early introduction of solid foods 413, Feeding cow milk during the first 12 months 414, No dependable source of iron for infants at 6 months of age or later 415, Improper dilution of formula 416, Feeding other foods low in . , . essential nutrients 417, Lack of sanitation in Infants Infants Infants Infants Infants Infants preparation, handling, and storage of formula or expressed breastmilk 418, Infrequent breastSeeding as Infants sole source of nutrients Yes Yes. Asks about amount of foods eaten. Yes. Asks when an infant is fed breastmilk, formula, milk, and solid foods. Yes. See above. Yes. See above. Yes. Asks about 2 days' food consumption.

OCR for page 83
ESTIMATING ELIGIBILITY BASED ON NUTRITION RISK CRITERIA 91 TABLE 7-3 Continue] Categorical Groups to Nutritional Risk Indicatora Which Related Data Available, by Survey (Code, Description) for Criterion Is Criterion Applicable NHANES CSFII 419, Inappropriate use of nursing bottles 420, Excessive caffeine intake Infants Pregnant women 421, Pica Each 422, Inadequate diet Each 423, Inappropriate or excessive Each Yes. Asks about all Yes. Asks intake of dietary prescription and about supplements nonprescription intake. . . . . . 1ncluc .lng vltamlns, mlnera .s, vitamins, and herbal remedies minerals, dietary 424, Inadequate vitamin/ mineral supplementation 425, Inappropriate feeding practices for children 426, Inadequate folic acid supplements. Each Yes. See above. Yes Children Pregnant intake to prevent neural tube women defects 501, Possibility of regression 502, Transfer of certification 503, Presumptive eligibility for Each Each Pregnant pregnant women women 601, BreastSeeding mother of Pregnant infant at nutritional risk 602, BreastSeeding complications or potential . . comp. .lcatlons 603, BreastSeeding complications or potential . comp. .lcatlons 701, Infant up to 6 months old Infants of WIC mother or of a woman who would have been eligible during pregnancy 702, BreastSeeding infant of woman at nutritional risk women Pregnant women Infants Infants Yes. See above. Yes corltirlued

OCR for page 83
92 ESTIMATING ELIGIBILII~YANDPARTICIPATIONFOR THE WICPROGRAM TABLE 7-3 Continued Nutritional Risk Indicatora (Code, Description) for , - . . Criterion Categorical Groups to Which Criterion Is Applicable NHANES CSFII Related Data Available, by Survey 703, Infant born of woman with Infants mental retardation or alcohol or drug abuse during most recent pregnancy 801, Homelessness 802, Migrancy 901, Recipient of abuse 902, Woman or infant/child of . . . . - . primary giver wltn 1lmltea ability to make feeding decisions and/or prepare food 903, Foster care Each Each Each Each Each aThe code numbers and brief descriptions are from U.S. Department of Agriculture (1998~. This memorandum provides detailed information about each criterion for nu- tritional risk. CSFII = Continuing Survey of Food Intake by Individuals; NHANES = National Health and Nutrition Examination Survey NOTE: Neither survey provides data related to more than 50 of the approved nutri- . tlona . rls. ~ criteria. data related to the indicator. The panel used both data sets when consider- ing lower bound estimates of the proportion of individuals meeting at least one criterion. The next section describes these two data sources. We note that in the future the CSFII will be discontinued and incorporated into NHANES. Continuing Survey of Food Intake Icy Individuals The CSFII surveys fielded in 1994-1996 and 1998 provide the most recent dietary intake data available from a nationwide food consumption survey. CSFII data have a large sample size for children categorically eli- gible for WIC (those less than 5 years of age) and include an oversample of low-income persons. However, the survey includes only a small number of

OCR for page 83
ESTIMATING ELIGIBILITY BASED ON NUTRITION RISK CRITERIA 93 pregnant and breastleeding women and does not identify other postpartum women. The 1998 supplementary CSFII survey was conducted to increase the sample size for children from birth through age 9 years and was de- signed so that the combined 1994-1998 sample of children constitutes a nationally representative probability sample. The combined 1994-1998 CSFII includes over 2,500 children ages 2 to 5 years who live in households with incomes at or below 185 percent of federal poverty guidelines. The 1994-1998 CSFII is a reliable nationwide data source for estimat- ing the proportion of income-eligible individuals for WIC who meet the dietary risk criterion failure to meet dietary guidlelines as specified in the report Dietary Risk Assessment in the WIC Program (Institute of Medicine, 2002~. The 1994-1998 CSFII collects two nonconsecutive 24-hour recalls of dietary intake for each individual in the sample (this replicate diet recall is missing for a negligibly small proportion of individuals in the sample). Replicate diet recalls offer an advantage when the quantity of interest is the usual dietary intake of a food or food group. Because food intake is variable from day to day, a single day's food intake provides a very unreliable esti- mate of the usual or habitual intake of an individual. The CSFII sample includes respondents from every state except Alaska and Hawaii. The sur- vey collects data during all months and seasons of the year in urban, subur- ban, and rural areas. CSFII collects data on respondents' participation in food assistance programs (including WIC), on income, and on other sociodemographic variables. Nonetheless, the CSFII is limited for estimating the proportion of in- dividuals who meet at least one of the many nutritional risk criteria. The survey does not contain information on most of the nondietary measures of nutritional risk, such as biochemical and clinical/health/medical status. Furthermore, the anthropometric data it includes are self-reported rather than standardized measurements, and the survey lacks information on the consumption of dietary supplements. National Health and/Nutrition Examination Survey NHANES provides nationally representative data relevant to many of the nutritional risk criteria in four of the five risk categories. Using highly standardized methods, NHANES obtains anthropometric and biochemi- cal measurements and a broad range of data related to health and medical problems. In addition, it collects one-day dietary intake data through the

OCR for page 83
102 ESTIMATING ELIGIBILII~YANDPARTICIPATIONFOR THE WICPROGRAM Lower BoundI Estimates for Infants If a postpartum woman participates in WIC or would have been eli- gible for WIC during her pregnancy, her infant is automatically considered to be at risk and fully eligible (criterion number 701 for "other" nutritional risk U.S. Department of Agriculture, 19981. This is a widely used crite- rion: according to 1998 WIC Participant and Program Characteristics (U.S. Department of Agriculture, 2000b), 74 percent of all infants who partici- pate in WIC had mothers who were eligible or participating in WIC dur- ing pregnancy. However, this percentage is likely to be an underestimate for two major reasons: Only about 64 percent of states record all the nutritional risk crite- ria under which a person is found eligible (U.S. Department of Agriculture, 2000b). Risk under this criterion might not be re- ported, for example, for infants at high risk because of a medical cone ition. If, as discussed above, more than 97 percent of income-eligible preg- nant women are at dietary risk, at least 97 percent of infants would have mothers who were at dietary risk during pregnancy. . Since infants ordinarily are certified for a one-year period, the above information implies that an adjustment factor of 0.97 is a reasonable lower bound for obtaining estimates of income-eligible infants who also meet a nutritional risk criterion. This is slightly higher than the value of 0.95, which is currently used by USDA. Lower Boundl Estimates for Childiren Ages 1 to 2 Years In 1998, 65 percent of children ages 1-2 years have an identified di- etary risk, a majority of them because of inadequate or inappropriate nutri- ent intake (U.S. Department of Agriculture, 2000b). The next most com- mon category of nutritional risk is anthropometric: 38 percent of children of this age meet at least one of the relevant anthropometric criteria (e.g., low or high weight for height or inappropriate growth or weight gain pat- tern). As stated previously, these percentages are likely to be underestimates of the WIC participants meeting these criteria, since not all state WIC agencies report all applicable nutritional risks. Criterion 425, "inappropriate feeding practices for children," actually

OCR for page 83
ESTIMATING ELIGIBILITY BASED ON NUTRITION RISK CRITERIA 103 includes nine subcriteria, any one of which could be used to establish di- etary risk of children in this age group. For many of these subcriteria, sur- vey data are not available to estimate the prevalence of young children who meet one or more of them. The identification of some of the risks would rely on information that is not collected by either CSFII nor NHANES. One such subcriterion is "Routine consumption or feeding of foods low in essential nutrients and high in calories that replace age-appropriate nutri- ent dense foods needed for growth and development between 12 and 24 months of age." We do not have data to estimate the lower bound of the prevalence of nutritional risk among children ages 1 to 2 years. However, considering the very large variation in day-to-day intake by children of these ages, the many subcriteria that could be used to confer dietary risk, the relatively high percentage of children with an anthropometric risk, and the array of other nutritional risk criteria, it is reasonable to expect that a very high percent- age of these children would have at least one nutritional risk. Furthermore, for previously certified children without other nutritional risks, criterion 501, "possibility of regression," may be used in certain circumstances. Such children are considered at nutritional risk when the competent professional authority at the WIC site determines a possibility of regression of nutri- tional status if the applicant does not continue to receive WIC benefits. This criterion reflects the preventive nature of WIC. Assessing Nutritional Risk in the Field Compared with estimating the percentage of individuals in a popula- tion who meet at least one nutritional risk criterion, screening for nutri- tional risk, especially for dietary risk, is an even more daunting task in the WIC service site. Since WIC field staff are required to screen for nutri- tional risk to determine full eligibility for WIC, and since dietary risk is the most common risk reported for women and children, effective screening for nutritional risk requires an accurate screening method for dietary risk. This section briefly describes the inherent limitations of the methods avail- able to WIC staff for screening for dietary risk. To assess dietary risk, WIC field staff generally obtain a single 24-hour diet recall or administer a food frequency questionnaire. Regardless of the skill of the staff member, both of these instruments have serious shortcom- ings if the goal is to determine whether or not the individual's usual intake of the food groups meets the criterion for dietary risk. Significant measure-

OCR for page 83
104 ESTIMATING ELIGIBILITYAND PAR TICIPATION FOR THE ~CPROGR~ ment error is associated with these instruments, albeit of a different nature for each. In the case of food frequency questionnaires, the measurement error is due to the failure of the instrument to accurately capture the usual or long- run average intake of foods (e.g., Kipnis et al., 19991. Moreover, studies have shown that food frequency questionnaires do a poor job of measuring "true" energy intake (Kipnis et al., 19991. A single 24-hour recall, designed to capture daily food intake, provides limited information about the individual's usual intake of food. Two recent IOM reports (Institute of Medicine, 2000, 2002) have documented that it is very difficult to assess the usual dietary intake of an individual accurately when only one or a few days of dietary intake data are available. In fact, information on daily dietary intake is subject to so much error that one could conclude that a person does not meet the criterion for dietary risk (that is, her habitual intake of a food group is at least equal to the cutoff point) only if the person's mean intake of that food group were consider- ably higher than the cutoff point (Institute of Medicine, 20001. The following example illustrates the problem with the 24-hour recall. If the applicant is a child age 2 to 5 years, then he or she would need to have a usual intake of two or more servings of fruit and six or more servings of grains (and also satisfy other dietary criteria) to be considered ineligible for WIC. However, the WIC staff member has only one day's intake, not usual intake. Given that fruit and grain consumption varies from day to day, how high would a single-day intake of fruits and grains (or other food group) need to be to conclude, with some degree of certainty, that the child's usual intake makes him or her ineligible? To answer this question, it is necessary to know the day-to-day variance in the child's daily intake of fruits and grains. Using the 1994-1998 CSFII data for children ages 2 to 5 years, the panel estimated the day-to-day standard deviation of number of servings of fruits and grains to be 0.98 and 1.64 servings, respectively. Then the panel computed the mean intake based on one day of data that would result in rejection of the hypothesis that the child's usual intake does not meet the criterion. They did this under the assumptions that daily intake of fruits and grains for the child is normally distributed and that the child's day-to- day variance in intake is similar to the population estimate. For a confi- dence level of 97.5 percent, the calculation is the following: one day mean 2 1.96 x SD of daily intake + threshold,

OCR for page 83
ESTIMATING ELIGIBILITY BASED ON NUTRITION RISK CRITERIA 105 which in the case of fruits, results in one day mean2 1.96x0.98 +2. Thus, the one-day intake of fruits reported by the child would have to be at least 3.9 servings before the WIC staff member could confidently conclude that the child's usual fruit consumption meets the threshold of two servings per day. In the case of grains, the daily reported intake would need to be slightly higher than nine servings for the WIC staff to be confi- dent that the child is meeting the grain servings criterion. Clearly, a single 24-hour recall provides little information about the child's usual intake of the food. Therefore, a WIC field member would need to observe a very high intake on one day before she could be sure that, on the average, the child consumes enough of the food. Regardless of the instrument used by the WIC field staff, assessing dietary intakes for an individual is very challenging, even under the best of circumstances. With the inherent limitations of practical methods to assess dietary intake of individuals, it is arguably impossible for WIC field staff to distinguish the persons who do not meet the dietary risk criterion from those who do. COST-BENEFIT ANALYSIS OF ASSESSING THE DIETARY RISK OF WIC APPLICANTS FOR DETERMINING ELIGIBILITY Considering the limitations of methods to screen for dietary risk, the panel examined the costs and benefits of screening for dietary risk. It is possible that because of inaccuracies in the screening process for dietary risk, individuals who truly meet a dietary risk criterion for nutritional risk and who would benefit from the WIC program might be excluded from participating, while others who do not meet the criterion might be allowed . , in, to enroll. Two potential remedies could reduce the costs of these errors in dietary risk eligibility determination.4 One remedy would be to improve the accu- racy of the screening process. The other would be to presume that all cat- egorically and income-eligible individuals are at dietary risk an approach 4Benefits and costs here are defined broadly, including all the benefits of the program to society and all the costs to society associated with the program.

OCR for page 83
106 ESTIMATING ELIGIBILITYAND PAR TICIPATION FOR THE ~CPROGR~ that was recommended for women and children over age 2 years in the recent IOM report (Institute of Medicine, 20021. The principal way to improve the accuracy of the eligibility screen to assess the dietary component of nutritional risk would be to collect several additional days of information on an applicant's food intake using the best methods available. However, collecting this additional information would increase the burden on the applicant and increase the administrative costs of the program in the time and effort needed to collect and review the information. Increasing the burden on the WIC applicant might be a bar- rier to participation and thus result in an increased number of unserved people who are nutritionally at risk. Assuming that WIC benefits reduce nutritional risk in the eligible population, if fewer eligible individuals apply because of an additional burden, then fewer eligible people would receive the nutritional benefits of WIC and more people would be at nutritional risk. The panel finds the presumption of nutritional risk a more appealing approach to consider. This approach is consistent with the IOM recom- mendation to presume that all categorically and income-eligible women and children ages 2 to 5 years are at dietary risk (and thus at nutritional risk).5 If this remedy were applied, then it would no longer be necessary to account for nutritional risk in the estimates of the number of WIC-eligible individuals for budgetary purposes. Presuming that all are at nutritional risk could have at least one nega- tive and at least one positive effect. In particular, it could increase the pro- portion of participants who are not at nutritional risk and who thus would benefit less from the program.6 However, presuming that all are at nutri- tional risk in determining eligibility would eliminate the possibility of in- correctly denying eligibility to any applicants who are at risk and would benefit from the program. We illustrate these two possible effects of ignoring the nutritional risk screen with two examples one in which the nutritional risk screen is used 5The Institute of Medicine (2002) report emphasized that the assessment of nutritional risk remains valuable for tailoring the contents of the food package and the nutritional edu- cation and referral services that should be given to an individual. Moreover the assessment of anthropometric, biochemical, and medical/clinical risks is necessary for application of the priority system, should funding be insufficient to serve all who apply. 6Program data are unavailable to determine the percentage of applicants who are found ineligible based on lack of nutritional risk alone.

OCR for page 83
ESTIMATING ELIGIBILITY BASED ON NUTRITION RISK CRITERIA 107 and one in which the nutritional risk screen is ignored. For both examples, assume that 1,000 individuals are both categorically eligible and income eligible for the program, and that 95 percent of those categorically eligible and income eligible are truly at nutritional risk, as defined by meeting a dietary criterion for nutritional risk. Further assume that, because of limi- tations inherent in the screening procedure, the chance of excluding an individual who is truly at risk is 10 percent (sensitivity equals 90 percent), and the chance of incorrectly certifying an individual as nutritionally at risk is also 10 percent (specificity equals 90 percent). Considering the poor accuracy of the screening tests, it is highly unlikely that both the sensitivity and specificity would be this high. Thus, the calculations probably repre- CC 1 '' sent a nest case scenario. In the first example, when the nutritional risk screen is employed, 5 of the 50 truly ineligible persons would be screened as eligible and 95 of the 950 truly eligible persons would be screened as ineligible. A total of 860 individuals would be screened as fully eligible. These results are summa- rized in Table 7-5, part A. In the second example, when the nutritional screen is not employed to determine eligibility, all of the 1,000 individuals would be certified as fully eligible. Of these, 50 would not be truly eligible (part B). However, as can be seen by comparing part A with part B. the 95 at-risk individuals who would not have been certified on the basis of the inaccurate nutritional screen would now be eligible for benefits. Is it economically rational to presume that all are at nutritional risk and thus fully eligible? This depends on whether the net social benefits of providing WIC benefits to an additional 95 individuals who are at risk are greater than the net social costs of providing WIC benefits to the 45 indi- viduals who are not at risk and would not pass the nutritional risk screen. The panel formalized this cost-benefit calculation. Table 7-6 presents a set of the critical values that the net social benefits of the WIC program would have to be in order to warrant ignoring the costs associated with the presumption that all income-eligible individuals are at nutritional risk and thus fully eligible for WIC (see Appendix B for the formalization of this analysis). These different critical values are calculated assuming different true levels of the prevalence of nutritional risk in the income-eligible popu- lation, different levels of the relative predictive power of the screening pro- cedure (the ratio of the probability that someone who is truly not at risk is screened as at risk to the probability that someone who is truly at risk is screened as at risk), and different values of the net social benefits of WIC to

OCR for page 83
108 ESTIMATING ELIGIBILI~YANDPARTICIPATIONFOR THE WICPROGRAM TABLE 7-5 Effects of Using or Not Using a Screen for Nutritional Risk on the Number Found Eligible or Ineligible, by Their True Nutritional Risk Status A Numbers eligible and ineligible when the nutritional screen is used. Fully Eligible Based on Nutritional Risk Screen Truly at Nutritional Risk Yes No Total Yes 855 95 950 No 5 45 50 Total 860 140 1,000 B Numbers eligible and ineligible when the nutritional screen is not used. Fully Eligible (No Nutritional Risk Screen) Truly at Nutritional Risk Yes No Total Yes 950 0 950 No 50 0 50 Total 1,000 0 1,000 NOTE: Both panels assume that 95 percent of income-eligible populations are truly at nutritional risk and that the nutrition risk screen has a 90 percent sensitivity and 90 .^ . percent spec1~1c1ty. those fully eligible. Examining the table, if the true proportion of income- eligible persons at nutritional risk is 0.90 and if the probability of accu- rately screening someone who is truly not at risk equals the probability of inaccurately screening someone who is truly at risk, then the net benefits of WIC should be at least 1.56 to justify presumption of nutritional risk that is, for each dollar of program expenditures, program benefits must be equal to $1.56. As the screening procedure becomes more accurate (the relative probability of correctly identifying those not at risk increases- moving down columns), the net benefits of WIC must be larger to justify presumption of nutritional risk. As the true prevalence of nutritional risk in the population increases (moving from left to right across rows), the net benefits of WIC needed to justify presumption of nutritional risk decrease. Several studies have made estimates of the net benefits of the WIC

OCR for page 83
ESTIMATING ELIGIBILITY BASED ON NUTRITION RISK CRITERIA 109 TABLE 7-6 Critical Values, by Prevalence of Nutritional Risk and Hypothetical Values of the Accuracy of the Screening Procedure, of Net Social Benefits of WIC Needed to Ignore the Nutritional Screening Procedure Prevalence of Nutritional Risk in the Income-Eligible Population Accuracy of Screen 0.90 0.95 0.99 1.0 2.0 5.0 10.0 1.28 1.31 1.39 1.53 1.26 1.28 1.32 1.38 1.25 1.26 1.26 1.28 NOTE: Accuracy level of the screen for nutritional risk is measured as the probability the screen will accurately assess someone not truly at risk divided by the probability the screen will inaccurately assess someone who truly is at risk as not at risk. See Appendix B for details on how the net social benefit levels needed to ignore the screen are calculated. program. The most robust findings on the net benefits of the program in the WIC evaluation literature have examined the effect of WIC on preg- nancy outcomes. For example, a General Accounting Office (GAO) study (U.S. General Accounting Office, 1992) found that for every $1 spent on pregnant women, WIC saved $3.50 on medical and disability costs because there were fewer low-birthweight births. In a study that attempted to ac- count for selection bias in the GAO estimates, Devaney et al. (1992) found savings of $2.29 for every dollar of WIC expenditures. If the GAO esti- mates or the Devaney et al. estimates are correct, then it is clear that the net benefits of WIC for pregnant women are large enough to justify the pre- sumption of nutritional risk for eligibility purposes. Only if the true ben- efits of WIC are much lower than these estimates is it inadvisable to pre- sume all are at nutritional risk. For example, if the screening procedure can accurately identify those not at nutritional risk (predictive power ratio of at least 5), and if the true prevalence of nutritional risk is 90 percent, then a net benefit of 1.39 would not be great enough to justify ignoring the screen and presuming that all are at nutritional risk. Whether the presumption of nutritional risk should be made for cat- egorical groups other than pregnant women depends on four factors: as-

OCR for page 83
110 ESTIMATING ELIGIBILITY AND PAR TICIPATION FOR THE WIC PROGRAM gumptions about the net benefits of WIC participation to these groups, the percentage of income-eligible persons who are truly at nutritional risk, the accuracy of the screening method, and the assumptions about the excess burden of raising tax money to fund the program (see Appendix B). Our lower bound estimates of the prevalence of nutritional risk for women, infants, and children ages 2 to 5 years are well over 90 percent. Further- more, as the preceding section discussed, the dietary risk screen used to determine WIC eligibility has a high level of inaccuracy. Given these two factors even using lower bound estimates of the net benefits of WIC presuming that all are at nutrition risk, is justified. However, if new infor- mation about the prevalence of nutritional risk or of WIC's benefits be- comes available, or if a more accurate screen is found, this presumption should be reexamined. The calculations outlined here and in Appendix B give the framework for such an analysis. SUMMARY In this chapter, the panel critiqued current methods used to adjust the number of categorically and income-eligible persons to account for those who do not meet at least one criterion for nutritional risk. The chapter also presented lower bound estimates of the prevalence of nutritional risk and discussed the inherent limitations of accurate assessment of the dietary risk of an individual. Finally, the chapter examined cost-benefit ratios needed in order to presume that all income-eligible persons meet nutritional risk cri- teria and are therefore fully eligible for WIC. The cost-benefit analysis found that, based on estimates of the net benefits of WIC, ignoring the nutritional risk screen to determine eligibil- ity is justified. A nutritional risk screen would be justifiable, however, if a revised, highly accurate screen that correctly identifies individuals who are not at nutritional risk were available, and if the actual prevalence of nutri- tional risk was considerably lower than the current estimate. Lower bound estimates of dietary risk among income-eligible infants, women, and chil- dren ages 2 to 5 years all are at least 97 percent, and those children ages 1 to 2 are likely to be that high as well. CONCLUSION: Given very high estimates of the prevalence of nu- tritional risk among income-eligible populations, gross inaccuracies in screening procedures for dietary risk, and cost-benefit calculations of administering the screen, the panel concludes that a nutritional risk

OCR for page 83
ESTIMATING ELIGIBILITY BASED ON NUTRITION RISK CRITERIA 111 screen is not useful for determining eligibility. If USDA drops this aspect of eligibility determination, no adjustment for the prevalence of nutritional risk is needed to estimate eligibility. The IOM report recommends that all women and children ages 2 to 5 years who meet all other eligibility requirements should be presumed to meet the requirement of nutritional risk through the failure to meet dietary guidelines criterion (Institute of Medicine, 20021. The dietary guidelines used in the criterion do not apply to infants and children between the ages of 1 and 2, so the IOM recommendation does not specifically apply to children of these ages. However, if the recommendation is adopted, all in- fants will necessarily also be considered at nutritional risk because an infant whose mother was considered to be nutritionally at risk during pregnancy is also considered to meet nutritional risk requirements. Thus, an implica- tion of the IOM recommendation is that all infants will also be presumed to be at nutritional risk. If the IOM recommendation is not adopted by USDA, then the lower bound estimates ofthe prevalence of nutritional risk given earlier in this chapter should be used to estimate the number fully eligible for WIC. These lower bound estimates are: 0.97 for pregnant women, 1.00 for postpartum women, 0.97 for infants, and 0.98 for chil- dren ages 2 to 5. There are no data to make a lower bound estimate of the prevalence of nutritional risk among children ages 1 to 2. However, given that the diets of children at this age are probably not that different from the diets of children ages 2 to 5, and the many other criteria that could be used to confer nutritional risk of children at this age, the prevalence of nutri- tional risk among children ages 1 to 2 is also likely to be very high. If all income-eligible people are considered to be nutritionally at risk and no downward adjustment for nutritional risk is made to the estimates of those who are income eligible, the number of those estimated to be eligible for WIC will increase. Eligibility estimates for children will increase the most because the current adjustment factor for nutritional risk for chil- dren is 0.752 lower than that for any other group. In 1999, 6.4 million children were estimated to be income eligible for WIC and 4.8 million were estimated to be both income eligible and nutritionally at risk. USDA should periodically assess the findings leading to the conclu- sion that the nutritional risk screen is not useful to determine eligibility. Better data to measure the prevalence of nutritional risk may become avail- able, or if the program is highly successful at reducing nutritional problems or nutritional behaviors of the population otherwise improve, the preva-

OCR for page 83
112 ESTIMATING ELIGIBILITYANDPARTICIPATIONFOR THE WICPROGRAM fence of nutritional risk in the population may decrease. If it decreases significantly, and if the screen for nutritional risk becomes more accurate, screening would become more important in targeting WIC's benefits to intended groups. The eligibility estimates would then need to be adjusted accordingly (i.e., by the percentage of the income-eligible population at nutritional risk).