3

Approach to the Task

For this interim report, the committee was tasked with collecting the information and data needed to support recommendations for potential modifications to the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) food packages. These recommendations will appear in the final, phase II report. In this section, the committee’s approach to information collection and data analysis is reviewed. The approach included

• Convening public workshops;
• Conducting literature searches;
• Analyzing food and nutrient intakes and diet quality of WIC and WIC-eligible (low-income and for women, also pregnant, breastfeeding, or postpartum) populations;
• Developing an approach to WIC food package food, nutrient, and cost profiles;
• Conducting a food expenditure analysis;
• Developing approaches to sensitivity and regulatory impact analyses to be conducted during phase II;
• Visiting WIC sites and shopping for WIC foods; and
• Reviewing public comments.

WORKSHOPS

For phase I of this review, two public workshops were held. The first, held on October 15, 2014, specifically supported the information-gathering

process for the first report of this three-report series, Review of WIC Food Packages: An Evaluation of White Potatoes in the Cash Value Voucher: Letter Report (IOM, 2015). The agenda for this workshop is available in Appendix H. The second workshop, “Methods and Approaches to the Assessment of WIC Food Packages,” was held in Washington, DC, on March 12, 2015, and included a public comment session on March 13, 2015. The agenda for this workshop is available in Appendix H. Presentations from both events are available on the Institute of Medicine (IOM) Web page for this study.¹ A public comment session was also held in Irvine, California, on June 26, 2015. Two additional workshops will be held in phase II to focus specifically on topics that relate to the development of the final report and its recommendations.

LITERATURE AND REPORT REVIEW

Comprehensive Literature Reviews

The committee was tasked with conducting a comprehensive literature review² to gather evidence to support its final recommendations. The first step was development of a draft of key research questions based on the statement of task (see Chapter 1, Box 1-1), the literature review questions developed for the letter report (IOM, 2015), and other topics outlined by the U.S. Department of Agriculture’s Food and Nutrition Service (USDA-FNS) for committee consideration. In collaboration with IOM staff and committee consultants, committee members refined the key questions, as well as the literature search strategy, study eligibility criteria, and the synthesis of search results, using an iterative process.

The key questions were organized by topic area:

1. Nutritional status of WIC populations;
2. Health status of WIC populations;
3. Breastfeeding promotion;
4. The role of WIC food packages in preventing food insecurity;
5. Racial or ethnic differences in infant/child feeding practices and personal food intake patterns;
6. Market availability of current WIC foods;
7. Administrative feasibility and efficiency for vendors; and

__________________

¹ Study details can be accessed at the following Web page: http://iom.nationalacademies.org/Activities/Nutrition/ReviewWICFoodPackages.aspx.

² Time and resources were inadequate to carry out a full systematic review. Specifically, the last two steps of a systematic review process were not completed: (1) risk of bias evaluation and (2) evidence synthesis (which includes evaluation of the strength of the evidence).

8. Barriers and incentives for WIC participants, potential participants, and their families.

Literature Search Strategy

Electronic literature searches of studies indexed in MEDLINE, PubMed, Agricola, CINAHL (Cumulative Index to Nursing and Allied Health Literature), ERIC (Educational Resources Information Center), PsychINFO, and Scopus (including Embase) were conducted. First, a broad search was conducted to identify all studies including WIC programs or WIC populations without restrictions to any outcome or study design. Searches were conducted using the National Library of Medicine’s Medical Subject Headings (MeSH) keyword nomenclature. All relevant studies with human subjects that were published in the English language from 2005 onward were identified. Duplicate citations across databases were removed before screening. Separate search strategies were developed to identify studies conducted among low-income populations living in the United States. The MEDLINE database was searched using a combination of search terms relating to Medicaid, poverty, and low income, plus search terms relating to firstly, culture or race/ethnicity and diet or feeding behavior or, secondly, food access or accessibility, food environment, food costs, store, and vendor. Furthermore, another Medline search strategy was developed for identifying interventional breastfeeding studies conducted among low-income populations living in the United States using the combinations of the low-income search with additional MeSH terms for culture and continental population groups and a broad search for breastfeeding, infant nutrition, and human milk. The full search strategy is described in Appendix I, Table I-2. The search was repeated before report completion to identify newly published papers.

Study Selection

Abstrackr software (abstrackr.cebm.brown.edu), EndNote, and Microsoft Excel were used to manage the search outputs, screening, and data abstraction. After a training session to ensure understanding of the inclusion and exclusion criteria, title/abstract screening was conducted in duplicate using a screening form that listed the inclusion and exclusion criteria and allowed selection of reasons for exclusion. A third reviewer reconciled the discrepant title/abstract selections. Full-text articles of all accepted title/abstracts were then retrieved and screened by one reviewer based on the study eligibility criteria. Second-level screening of full text articles was conducted by two reviewers and differences reconciled by a third reviewer. The literature search and study selection flow and study eligibility criteria

for each key question are both described in Appendix I (Figure I-1 and Table I-1, respectively).

Challenges with Evaluating WIC-Specific Data

Since its creation, it has been difficult to evaluate the effect of WIC participation on any outcome with a study design that is suitable for causal inference. Only limited experimental options are available (e.g., random assignment of a WIC service area to delayed start of a new benefit) because random assignment of individuals to receive or not receive WIC benefits is not considered ethical. In the 1980s, Rush and his colleagues used studies of several different designs (e.g., historical, longitudinal cohort, and cross-sectional), each with different weaknesses, to provide a comprehensive assessment of the WIC program (Rush et al., 1988a,b,c,d). Such a large and comprehensive study has not been repeated. As a result, nearly all studies reviewed for this report compare WIC participants to a group of nonparticipants or use a pre-post design (relative to a change in the food package). These study designs are not sufficient for causal inference. Kreider et al. (2016) used nonparametric partial identification methods to jointly account for selection and measurement problems and evaluate the causal impacts of WIC on food insecurity in children, using the National Health and Nutrition Examination Survey (NHANES) data. Their methods offer an alternative approach and bound the average treatment effects of WIC on observed outcomes.

A challenge to analyzing WIC-specific data is a phenomenon known as selection bias, which occurs when individuals who choose to participate in a program are different from eligible individuals who choose not to participate. These differences can be either observable or unobservable. With many social assistance programs, participants are likely to be negatively selected, that is, less well off, for example with less education or less wage income (compared to nonparticipants). This leads to results that make it appear that the program is not as effective as it really is. Conversely, participants may be positively selected for unobserved or unobservable characteristics, such as motivation or the eagerness to keep their children healthy (Besharov and Germanis, 2001). This leads to results that are biased upward that make it appear that a program, such as WIC, has more positive effects than it really does. For WIC specifically, positively biased effects could also result from longer-lasting pregnancies, with longer pregnancies increasing the chances that WIC-eligible women will enter the program,

and also giving them a longer time period over which to benefit from the program.³

Using 1992–1999 data from the Center for Disease Control and Prevention’s (CDC’s) Pregnancy Risk Assessment Monitoring System, Bitler and Currie (2005) conducted a survey of mothers at 6 months postpartum and found that WIC participating women were negatively selected for several observable characteristics compared to WIC-eligible, nonparticipating women whose birth was paid for by Medicaid. Specifically, they found that WIC participants were less educated, less likely to be married, more likely to be of minority race, more likely to be teen mothers, less likely to report the father’s information on the birth certificate, more likely to be obese, more likely to use public assistance and less likely to have wage income in the past year, and more likely to have had a previous low birth weight or premature infant if not a first-time mother. More recently, in a study of birth records from New York City, Currie and Rajani (2015) examined women who were pregnant more than once but who chose to participate in WIC only for one birth. They found that WIC pregnancies were more likely when women were younger, unemployed, unmarried, or had experienced a bad previous birth outcome. When there is negative selection on observable factors, as shown in these two studies, it seems likely that there is also negative selection on at least some unobservable factors (e.g., the woman’s propensity to have negative birth outcomes outside of any conditions that can be measured by the researcher) as well. There is little reason to expect that there is solely an upward bias in the reported program effects because of the likely cumulative effect of negative selection on these factors (Altoni et al., 2005, 2008).

Evaluation of WIC participant outcomes before and after the 2009 adoption of the new food package is complicated by the fact that adoption of the new package took place at the tail end of a recession and at a time when families were facing the worst labor market since the deep recession of the early 1980s. The American Recovery and Reinvestment Act of 2009 provided the funds necessary to increase the maximum benefit level of Supplemental Nutrition Assistance Program (SNAP) of about 15 percent (EOPUS, 2014). Inasmuch as the SNAP recipients are automatically eligible for WIC, many WIC participants also receive SNAP benefits. Among those

__________________

³ One important possible source of bias that is prominent in the recent WIC literature is gestational age bias. For example, suppose two women are similar on every dimension but for idiosyncratic reasons, one gives birth at 7.5 months and the other at 9 months. The woman with the premature birth would have enrolled in WIC at 8 months had her pregnancy lasted to 8 months, and the second woman does enroll at 8 months. A comparison of prenatal WIC use and gestation would lead to the mistaken conclusion that WIC participation caused longer gestation.

who were receiving both benefits, food expenditures and consumption may have changed because SNAP increased the maximum benefit level.

Identification of Relevant Reports

In addition to the literature search described above, relevant IOM reports and government reports related to the task, also published since 2005, were identified and evaluated. The USDA Economic Research Service (ERS), FNS, and Agricultural Research Service (ARS) websites were searched for reports relevant to WIC and other topics identified as relevant by the key questions.

Additional Literature Searches

Additional literature searches were conducted to address specific chapter topics, for example, to identify information to support a review of relevant nutrition-related health risks in Chapter 6, to understand food allergies, and other food intolerances, and to understand the health effects of fruit juice or high-fat dairy in Chapter 9, as examples.

Special Task: Approach to Identifying Literature on Functional Ingredients

The committee was asked to consider the current science on functional ingredients added to foods for adults, children, and infants, particularly infant formula (see Chapter 9 for a review of infant formula developments since the 2006 review of food packages). This information will be used in phase II to consider how USDA-FNS might approach the inclusion of foods containing these ingredients in the WIC food packages. A unique search was conducted to address this task. The functional ingredients investigated were those currently added to infant formula, because this is the item in the WIC food packages of primary interest to USDA-FNS with respect to these ingredients. The literature search used common names for ingredients, along with expanded variations. Health effects of these ingredients relevant to the WIC population (women, infants, and children) were considered.

From an initial broad literature search, the committee narrowed the evidence base to three sources of information on health effects: (1) statements from authoritative bodies on nutrition and health (e.g., American Academy of Pediatrics [AAP], Academy of Nutrition and Dietetics [AND], American Heart Association [AHA], Agency for Healthcare Research and Quality [AHRQ]); (2) U.S. Food and Drug Administration [FDA] qualified health claims; and (3) Cochrane Reviews. Search results were retained only if they related to dietary and/or supplemental sources of a functional

ingredient. Evidence related to enteral or parenteral administration was excluded, as were outcomes not anticipated to affect a large portion of the WIC population (e.g., gout) as well as outcomes not anticipated to be affected by the short-term, supplemental nature of the WIC food packages (e.g., cardiovascular disease).

NUTRIENT AND FOOD INTAKE: EVALUATING ADEQUACY

The committee was tasked with estimating nutrient intake and intake adequacy in the WIC population based on recommended Dietary Reference Intakes (DRIs) and comparing food intakes to those recommended in the 2015 Dietary Guidelines for Americans (DGA), bearing in mind that the purpose of WIC is to provide supplemental food to correct for nutritional intake inadequacies. This section describes the methods used to assess the prevalence of inadequate and excess nutrient intake in the WIC subpopulations and, for this phase I report, compare food intakes to the recommended food patterns in the Scientific Report of the 2015 Dietary Guidelines Advisory Committee (2015 DGAC report) (for the phase II report, they will be compared to the 2015 DGA).

Dietary Reference Intakes for Micronutrients

The different types of DRI standards for nutrients are described in Box 3-1. For the past two decades, IOM committees have been developing and releasing nutrient intake recommendations to update the DRIs (see Appendix J, Tables J-1a through 1c (IOM, 1997, 1998, 2000a, 2001, 2002/2005, 2005, 2011a). The most recently updated DRIs were for calcium and vitamin D (IOM, 2011a). Wherever possible, the IOM DRI reports present a review of the available science base for quantitative recommendations and the amount of each nutrient needed to meet the nutritional requirements to maintain health in apparently healthy individuals, grouped by age and sex, in the United States and Canada. For this report, the Estimated Average Requirement (EAR) and Tolerable Upper Intake Level (UL) were applied to assess the nutrient intakes of the various WIC population subgroups; the Adequate Intake (AI) value was applied in cases where an EAR has not yet been determined. The EAR is appropriate for population or group-level evaluations of nutrient adequacy. Mean intakes at or above the AI imply a low prevalence of inadequacy in the group (IOM, 2000b).

Dietary Reference Intakes for Macronutrients and Energy

DRIs for Macronutrients

Macronutrients include carbohydrate, protein, and fat. These nutrients have associated DRIs known as the Acceptable Macronutrient Distribution Ranges (AMDRs) (for children and adults only), and may also have an EAR or AI value. For protein, an EAR has been established for individuals 6 months of age and older (see Appendix J, Table J-1c), but only an AI for infants younger than 6 months. Protein intakes are assessed using these values. For carbohydrate and total fat, intakes of women and children are compared to the AMDR, but intakes of infants are compared to the AI.

Although the IOM (2002/2005) report recommended limiting the amounts of saturated fat and cholesterol for all individuals more than 2 years of age, analyses of these macronutrients in this report are based on updated recommendations in the 2015 DGAC report (USDA/HHS, 2015). The latter report indicates limits for saturated fat, and does not specify a limit for cholesterol intake. Cholesterol intake was therefore not evaluated in this report.

Estimated Energy Expenditure

Comparing food group intakes to those recommended in the 2015 DGAC report required calculating Estimated Energy Requirements (EERs) for the various WIC subgroups. A 2002 IOM committee developed equations to derive EERs that balance total energy expenditure at a level of physical activity consistent with health and support growth rates in children that are compatible with a healthy body size and composition (IOM, 2002/2005). In children, the EER was calculated based on an individual’s age, body weight, height, and activity level. For adults, the EER was calculated based on age, gender, body weight, height, and physical activity level. The EER calculations applied in this report assumed a low-active physical activity level (PAL) for women and children 2 to 5 years of age. The EER for pregnant and breastfeeding women also includes energy needs associated with the deposition of tissue or the secretion of milk. This committee used these equations. For pregnant women, the second trimester of pregnancy was assumed to cover all stages of pregnancy because a woman’s specific stage of pregnancy at the time her intake was assessed is not recorded in NHANES. For breastfeeding women, the EER assumed the first 6 months postpartum. Recent research suggested that the IOM (2002/2005) formula may overestimate energy needs for children (Butte et al., 2014), although this finding is yet to be validated broadly. Interpretations of data in this report were considered in light of these recent findings.

Recommended Limits for Other Dietary Components

The 2015 DGAC report recommended limiting intake of added sugars to not more than 10 percent of total energy intake. In July 2015, the FDA issued a proposed rule for the inclusion of percentage of calories from added sugars on the Nutrition Facts label (FDA, 2015), indicating that regulatory action is underway to support limits on added sugars intake. For sodium, the 2015 DGAC panel set an upper limit of 2,300 mg per day (in agreement with the established IOM UL) for adults, and a goal of less than the established DRI (UL) for other age groups (USDA/HHS, 2015). For children age 1 to 3 years, this is 1,500 mg per day and for children 4 to 8 years, this is 1,900 mg per day (IOM, 2005).

Using the DRIs to Assess Nutrient Adequacy

The committee used the DRIs to assess nutrient adequacy, which involved examining both inadequate and excessive intakes of nutrients. The methods applied in this report are the same as those used in IOM (2006) and originally designed by Nusser et al. (1996) and Carriquiry (1999) (see Appendix C of IOM [2006]). Brief descriptions of the approaches are provided here, with modifications noted as appropriate. Nutrients analyzed for this report are listed in Appendix J, Table J-2.

Estimating Usual Intake Distributions

Assessing nutrient adequacy involves, first, estimating usual distributions of intake. The Iowa State University (ISU) method proposed by Nusser et al. (1996) and applied in the IOM (2006) report for determining usual intake distributions is generally accepted in the nutrition community, and several software packages are now available to generate the mean and variance of usual intake as well as percentiles of intake of the user’s choosing. For this report, PC Software for Intake Distribution (PC-SIDE) was used to implement the ISU method (nutrients). To estimate the distribution of dietary components consumed episodically (food groups and subgroups), the Statistical Program for Age-adjusted Dietary Assessment (SPADE), a method similar to the National Cancer Institute method was implemented (Dekkers et al., 2014). These software packages are specifically designed for estimating the usual intake distributions of populations, and are not appropriate for application to individuals (IOM, 2000b).

Assessing the Prevalence of Inadequate Nutrient Intake with EARs

In all of the statistical analyses, intake data were weighted to population values by using survey weights associated with survey participants. Fractional jackknife replicate weights (Fuller, 2009) were used to estimate standard errors of estimated percentiles. Usual nutrient intake distributions were estimated using methods that account for the statistical properties of the data (intra-individual variation and reported data that are not normally distributed [Nusser et al., 1996; IOM, 2000b]). Beaton (1994) and Carriquiry (1999) suggested that the prevalence of inadequate intakes in the group can be estimated by the proportion of persons in the group whose usual intakes do not reach the EAR for the nutrient. This approach is known as the EAR cut-point method.

A difficulty arises when one wishes to estimate prevalence of inadequacy in a group that includes persons from groups that have different EARs. If the sample size is too small to carry out separate analyses for each group, it is possible to proceed as proposed by IOM (2000b). This approach for estimating prevalence of inadequacy when combining population subgroups with different EARs consists of rescaling daily intakes for one of the population subgroups so they can be compared to the EAR of the other group (a similar re-scaling was used in IOM, 2006). This approach was applied to two of the population subgroups of interest in this work: children aged 2 to less than 5 years, and women aged 19 to 50 years of age. Neither of these two groups aligns with the DRI gender and age groups; this is particularly true for women. As a result of low sample sizes, pregnant, breastfeeding, and postpartum (not breastfeeding) women were grouped into single analytic samples by WIC participation and income status. The resulting prevalences of inadequacy must be interpreted carefully when the EARs for the groups that are being combined are very different. For example, the EAR for iron for pregnant women is approximately three-fold that for lactating (breastfeeding) women 19 to 30 years of age.⁴ Thus, the overall prevalence of iron inadequacy for the combined group may conceal a relatively high prevalence among pregnant women and a much lower prevalence among lactating women. For iron specifically, another caveat is that requirements are not normally distributed for women, mostly because of menstrual losses of iron. As a result, the EAR cut-point method cannot be used to estimate the prevalence of inadequacy of iron. Inasmuch as most of the women in the analytical sample were either pregnant or breastfeeding and the sample size was small, the EAR cut-point method was nonetheless implemented. These limitations were considered when interpreting the data.

In addition to analyzing nutrients in reference to EARs, means and usual

__________________

⁴ The EARs for iron during pregnancy and lactation are 22 and 6.5 mg per day, respectively.

intake distributions were also determined for nutrients with AIs (IOM, 2006). Interpretation of intake differs for nutrients with AIs in that only limited inferences can be made about the prevalence of nutrient inadequacy. If a mean intake level is equal to or exceeds the AI, it is likely that the prevalence of inadequacy is low, but no conclusion can be drawn about the prevalence of inadequacy for a mean intake level that falls below the AI (IOM, 2000b). For this reason, in this report the prevalence of inadequacy was not evaluated for nutrients with AIs.

Note that only AIs are available for infants 0 to less than 6 months of age, therefore the prevalence of inadequacy of any nutrient could not be calculated for this age group.

Assessing the Prevalence of Excessive Intakes

Excessive intakes of micronutrients were assessed by comparing observed nutrient intake to the UL for that nutrient, as described in IOM (2006). Not all nutrients have ULs and, for some nutrients, the UL is based on intake of supplements that were not evaluated for this report. In this report, the probability of exceeding the UL was determined only for retinol, vitamins C and B6, calcium, iron, phosphorous, zinc, copper, and selenium. Inasmuch as there is no evidence of adverse effects from the consumption of folate, vitamin E, niacin, and magnesium naturally occurring in food, the ULs for these four nutrients are set in reference to intake from supplements, fortificants, or pharmacological agents only (IOM, 1997, 1998, 2000a). Therefore, intake relative to the UL was not evaluated for folate, vitamin E, niacin, and magnesium. Excess zinc intake was not considered of concern for formula-fed infants or children 1 to less than 2 years because the method used to set the UL resulted in a narrow margin between the Recommended Daily Allowance (RDA) and the UL (IOM, 2001). For other age groups, there exists no evidence for adverse effects from zinc naturally occurring in food (IOM, 2001), and the committee considers infant formula (and zinc provided therein) to be tightly regulated for safety by the FDA. Excess retinol intake was not considered of concern because of a similarly narrow margin between the UL and the RDA (IOM, 2001). Toxicity from excess consumption of retinol rarely occurs without supplemental intake (IOM, 2001).

Special Case: Vitamin D

Both dietary intake and sun exposure contribute to an individual’s vitamin D status. It is generally agreed that dietary intake of vitamin D is of limited value in the evaluation of vitamin D adequacy because the relationship between the two is nonlinear (IOM, 2011a). Further, the current USDA Food and Nutrient Composition Database does not separate vita-

min D from 25-hydroxyvitamin D (25(OH)D) in foods. This results in an underestimate of the bioequivalent vitamin D in foods because 25(OH)D is four to five times more bioequivalent than is the parent form of vitamin D (Cashman, 2012; Cashman et al., 2012).

In contrast, serum 25(OH)D captures both total dietary intake of parent vitamin D and 25(OH)D and sun exposure and has been validated as a biomarker for assessing vitamin D adequacy (IOM, 2011a; Taylor et al., 2013). Data on adults aged 19–70 years from NHANES 2005–2006 indicate that approximately 71 percent of the U.S. population consumes less than the EAR for dietary vitamin D, but the prevalence of inadequacy assessed by 25(OH)D is only about 19 percent (Taylor et al., 2013). Food package content of vitamin D will be determined in phase II, primarily to serve as a reference point for food package changes (i.e., if, during phase II, the committee determines that foods containing vitamin D should be added to the WIC packages, the potential difference from baseline dietary intake can be estimated). Only vitamin D intake data are presented only for infants 0 to less than 12 months of age in this report because serum 25(OH)D data are not available for this group. Data on serum 25(OH)D were available for individuals ages 1 year and older for NHANES survey years 2005–2006 (see the next section in this chapter for a description of the NHANES survey).

Assessing the Prevalence of Inadequate and Excessive Consumption of Macronutrients

As noted above, for macronutrients, protein intakes were compared to recommended intakes in g/kg/d but, for carbohydrates and fats in most age subgroups, the proportions above and below the AMDR were estimated. AMDRs are expressed in terms of percentage of total calories contributed by the macronutrients. Carbohydrate intakes below the AMDR are not considered of concern given lack of evidence for harm. Because the 2015 DGAC report emphasized saturated and not total fat (USDA/HHS, 2015), intakes of total fat exceeding the AMDR were likewise not considered to be of concern.

Comparing Food Intakes to Dietary Guidelines

The DRIs serve as the basis for nutrient targets in the DGAs. Recommended food patterns developed as part of the DGA consider nutrient requirements (as specified by the DRIs) as the foundation, in combination with usual dietary intake patterns of Americans (see Appendix E-3.1 of USDA/HHS, 2015). The committee was tasked with evaluating nutrient and food intake of the WIC-eligible population in comparison to both

the DRIs and the DGA. The DGA cover only individuals ages 2 years and older, therefore, a review of authoritative guidance other than the DGA was conducted for individuals less than 2 years of age.

Dietary Guidance for Individuals Ages 2 Years and Older

The food patterns indicative of a healthy diet are developed by the USDA every 5 years and released as new DGA. For this report, the committee applied the recommendations and food patterns outlined in the 2015 DGAC report (USDA/HHS, 2015), which provides the scientific underpinnings for development of the 2015 DGA (anticipated for release in early 2016). For the phase II report, the 2015 DGA will serve as the basis for recommendations, superseding use of the 2015 DGAC report.

Table 1-5 in Chapter 1 illustrates the food patterns recommended in the 2015 DGAC report for various energy intake levels. To evaluate the diets of all children 1 to less than 5 years of age, the committee applied a weighted food pattern (a 1,000 kcal pattern weighted 1:3 with the average of 1,200- and 1,400-kcal patterns [IOM, 2011b], referenced herein as the “1,000–1,300 kcal weighted diet”). This approach generated a single food pattern that could be applied across all children, simplifying the analysis.⁵ For all WIC women, a 2,200-kcal pattern was applied, which was the mean calculated EER among WIC women in the NHANES analyses conducted for this report.

Also as described in Chapter 1 (see Table 1-6), the 2015 DGAC report identified the following shortfall nutrients: vitamins A, D, E, and C; folate; calcium; magnesium; fiber; potassium; and iron for adolescent and premenopausal women. The 2015 DGAC report further identified a subset of these (vitamin D, calcium, potassium, and fiber, as well as iron for adolescent and premenopausal women) as nutrients of public health concern because they are linked to specific adverse health outcomes (USDA/HHS, 2015). The committee paid particular attention to the adequacy of intake of these nutrients.

Dietary Guidance for Infants and Children, 0 to 24 Months of Age

The DGA do not provide dietary guidance for individuals from birth to 24 months of age, although the possibility of expanding the DGA to include these individuals is currently being explored (Raiten et al., 2014). In this report, the adequacy of food intakes of infants and children 1 to less than 2 years of age could not be evaluated using a dietary pattern due to small

__________________

⁵ Ultimately, the sample sizes were too low for children 1 to less than 2 years of age to generate adequately precise data for “% below recommendations” and only means are presented in Chapter 5.

sample sizes, but rather, mean intakes were compared across subgroups and to other nationally representative data. The committee searched and compiled dietary guidance information for these age groups from AAP, AND, the World Health Organization (WHO), and other sources. This guidance is presented in detail in Table 3-1.

TABLE 3-1 Dietary Guidance for Infants and Children Less Than 2 Years of Age

^a There is some controversy regarding whether exclusive breastfeeding meets energy requirements of infants at 6 months of age in developed countries (Fewtrell et al., 2007). Fewtrell et al. (2007) states, “A reasonable interpretation of the available scientific data is that there are currently insufficient grounds to confidently recommend an optimal duration of exclusive breastfeeding of 6 as opposed to 4–6 months for infants in developed countries.”

^b Concern regarding the possible risk from intake of excessive mercury or other contaminants is offset by the neurobehavioral benefits of an adequate DHA intake and can be minimized by avoiding the intake of predatory fish (e.g., pike, marlin, mackerel, tilefish, swordfish) (AAP, 2014).

SOURCES: As indicated in the Reference column.

Inadequacy or Excess: The Basis for Concern

The committee was tasked with developing nutrient intake adequacy estimates referenced to the DRIs. On a population level, inadequate or excessive intake of any nutrient is usually considered to be of concern when present in 2.5 percent or more of the population of interest (IOM, 2003). This percentage should translate to an equivalent prevalence of impaired function or adverse effect. For example, a 5 percent prevalence of dietary iron inadequacy should translate to a 5 percent prevalence of low iron stores. For this report, a 5 percent threshold was applied (as in IOM, 2011b). This is a slightly relaxed standard, which accounts for some of the uncertainty in setting the EARs, as well as some of the generally accepted errors associated with dietary assessment. The same threshold was applied to proportions of the population with intakes falling above or below the AMDR, or above the UL. For nutrients with an AI, an assessment of adequacy cannot be made. Rather, it can only be stated that the mean usual intakes above the AI implies a low prevalence of inadequacy (IOM, 2000b).

Food group intakes can be compared to recommended food patterns for a specific energy level, as described previously. Because the food patterns are designed to ensure nutrient intakes that meet almost all of the RDAs, it would be ideal if almost everyone in a population reported usual diets that conformed to the food patterns. However, this goal is almost never achieved, so the committee chose a less restrictive approach in selecting foods group intakes that should be improved: if 50 percent or more of the population falls below the recommended level, then improving intake should be a priority. This approach improves on past assessments that prioritized food groups with mean or median intakes below the recommendation, but that did not quantify the percentage of the population with low intakes.

NUTRIENT AND FOOD INTAKE IN THE WIC POPULATION

Nutrient and food intakes in the WIC-eligible population were estimated using NHANES 2005–2008 and 2011–2012. The intent of these analyses was to identify priority nutrient and food group needs that could be addressed by making additional changes to the food packages. The methods of these analyses are described here. The results are discussed in Chapter 4 (nutrient intake) and Chapter 5 (food intake).

Dataset

The primary source of data on food and nutrient intake of the U.S. population is the What We Eat in America (WWEIA) component of NHANES

(USDA/ARS, 2005–2008, 2011–2012). The survey data used for this report were dietary intake data (foods and nutrients from food sources only, not dietary supplements⁶) collected using the Automated Multiple-Pass Method,⁷ and demographic information, including age, gender, and physiological status (e.g., pregnant, breastfeeding, or postpartum women [0–1 year after delivery]⁸). The only filter applied to create the analytic datasets was the indicator DR1DRSTZ (or DR2DRSTZ for day 2), which identified complete and reliable records. No outliers were removed. By and large, the published NHANES databases have few missing values, in particular for nutrient intake. The population survey weights were applied to all analyses, generating estimated intake values representative of the U.S. population, including by income categories. However, participation in programs such as WIC is not considered in the survey design (Johnson et al., 2014). In addition, pregnant, breastfeeding, or postpartum women are not oversampled (Johnson et al., 2014), which results in small sample sizes for these physiological states, apart from narrowing to low income.

Food intake data for each survey respondent were translated to USDA equivalent values using the Food Patterns Equivalent Database (FPED), a file that identifies the food group and subgroup intakes associated with the DGA recommendations (USDA/ARS, 2013). A reasonability check was conducted to compare the output for this report to the nationally representative WWEIA data. The food groups selected for analyses are presented in Appendix J, Table J-3.

Utility of NHANES Datasets for Addressing the Task

The committee was tasked with assessing the nutrient and food group intakes of the WIC-eligible population. USDA-FNS also requested an evaluation of intakes before and after 2009 food package changes, and a comparison of WIC participants to eligible non-WIC participants. USDA-FNS required full implementation of the 2007 (interim rule) food package changes by October 2009, and most states implemented the changes at

__________________

⁶ At the request of the study sponsor, USDA-FNS, dietary supplement intake was excluded from the analysis. The purpose of the WIC food packages is to improve nutrient intakes from foods alone. It would not be appropriate to assume that all WIC participants are taking specific supplements or to design the food packages based on such an assumption. Thus, although the committee recognizes that dietary supplements can provide additional nutrients, it was important to examine intakes from foods alone.

⁷ The Automated Multiple-Pass Method is a computerized method for collecting interviewer-administered 24-hour dietary recalls. In NHANES it is applied in person for the first day, and by telephone for the second day of data collection.

⁸ Women were selected from NHANES if coded as breastfeeding, or if not breastfeeding, but coded as 0 to 5.9 months postpartum. Some women reporting WIC participation did not report being pregnant, breastfeeding, or postpartum.

some point between issuance of the 2007 interim rule and the October deadline (USDA/FNS, 2012). Given the number of complications with dividing the NHANES 2009–2010 data,⁹ the committee estimated prepackage change intakes using NHANES 2005–2008.

The WIC identifier for the NHANES 2011–2012 dataset was not available at the time of this analysis. Therefore, a comparison of nutrient or food intakes among WIC participants before the 2009 food package changes to those after the changes could not be conducted. Moreover, the comparison of WIC participant intakes to WIC-eligible nonparticipants could be conducted only with the NHANES 2005–2008 release.¹⁰ The pre/post comparison will be available in the phase II report, in which NHANES 2011–2012 will be analyzed using WIC participant and WIC-eligible nonparticipant subgroups as the sample sizes allow.

For each WIC subgroup comparison, the committee evaluated the population subgroup sizes to determine which combinations of individuals relevant to the task would allow adequately robust sample sizes. Oversampling of some NHANES population subsets has been discontinued (CDC, 2014), which was a concern for several of the WIC subgroups of interest because small subgroup sizes may result in statistically unreliable population-level estimates.¹¹ The committee’s initial goal was to analyze WIC participants¹² and WIC-eligible nonparticipants in subgroups of infants (formula-fed or breastfed), children (1 to less than 2 and 2 to less than 5 years of age), and women (19 to 50 years of age, eligible being pregnant, breastfeeding, or postpartum). These subgroups allow for comparison of nutrient and food intake of all individuals who participate in WIC compared to individu-

__________________

⁹ NHANES respondents are assigned weights specific to the 2-year datasets. Separation of a 2-year dataset requires re-computation of population weights, which was beyond the scope of this study. It also required knowledge of the location of the participant and the dates of the interviews. Both of these variables are unpublished to preserve privacy of participants.

¹⁰ In addition to the difficulties with separation of the NHANES 2009–2010 dataset noted in footnote 7, this period spanned the change in food packages. It was therefore not considered appropriate for either the pre- or post-food package change assessments.

¹¹ The committee determined that a mean usual intake can be calculated within 3 percent of the true value (95 percent confidence interval) with a minimum of 17 individuals, for most nutrients. This minimum is not adequate for accurate calculation of population-level intake adequacy.

¹² Capturing WIC participation is dependent on accurate reporting in NHANES. The committee’s comparison of the weighted total number of recipients reporting WIC as well as extensive experience with reporting of programs like WIC suggest that WIC use is underreported. There is also a challenge in identifying the low-income group as eligible: The concept of income reported in NHANES does not correspond to state-level income requirements for eligibility. Some individuals may be income ineligible but may still legitimately participate in the program if adjunctively or automatically eligible due to participation in Medicaid, Temporary Assistance for Needy Families (TANF), or the Supplemental Nutrition Assistance Program (SNAP).

als who qualify but do not participate in the program. Inspection of the data in the survey years of interest (2005 through 2012) indicated that modification of these initially outlined population subgroups was required. Table 3-2 details the limitations of NHANES for developing these initially designed population subsets and the modifications made to accommodate the limitations.

Following careful consideration of these limitations, the committee designed the final population subgroups that would be analyzed for this report (see Table 3-3). Subgroups identified as low income include all individuals with income ≤ 185 percent of the poverty-to-income ratio (PIR) (based on PIR guidelines in HHS, 2015, and USDA/FNS, 2015). The WIC subgroups include only individuals reported as being on WIC in the NHANES survey (these individuals may or may not have a PIR of ≤ 185 percent). There are two reasons for inclusion of any income level in the WIC group: (1) income could change within the certification period, but the individual remains in the program at the new income level, and (2) the objective is primarily to evaluate the effect of the food package, not the effect of income. WIC-eligible non-participating individuals were identified in the survey by not reporting being on WIC, but with a PIR of ≤ 185 percent and for women, having a qualifying physiological state (e.g., pregnant, breastfeeding, or postpartum).

TABLE 3-2 Limitations of the NHANES Datasets Relevant to the Task and Resulting Subgroup Modification

NOTES: non-WIC = WIC-eligible nonparticipants; WIC = individuals participating in WIC.

^a NHANES (National Health and Nutrition Examination Survey) includes a Food Security Survey Module that contains an identifier for individuals currently receiving WIC benefits and those who received WIC benefits in the past 12 months. This identifier can be used to identify subgroups of individuals receiving WIC with WIC-eligible women not receiving WIC benefits and also with low-income women who are not currently pregnant, lactating, or postpartum (i.e., eligible for WIC).

^b The typical age distribution for WIC participation is 18–34 years (USDA/FNS, 2013a).

^c This information has been updated since the initial release of this report.

TABLE 3-3 NHANES Sample Sizes of Population Subgroups Selected for Nutrient and Food Intake Analyses: Phase I

NOTES: BF = breastfeeding, FF = formula fed; P = pregnant; PP = postpartum up to 1 year. Numbers may differ between the nutrient and food intake analyses because 2 days of food intake data are required to estimate usual intakes for food. At the time of analysis, the WIC indicator was not available for NHANES 2011–2012. Population subgroups for phase II may vary from what is presented here, depending on the “WIC” and “non-WIC” sample sizes in NHANES 2011–2012.

Adjustment for Small Sample Sizes

As indicated in Table 3-2, some of the sample sizes were small. The committee determined that means for subgroups other than women were adequately precise, despite sample sizes as small as 19. For example, to estimate mean usual intake of calcium for infants ages 0 to less than 6 months, a minimum sample size of 17 infants is required to obtain an estimate that is no more than 20 mg below or above the true mean with 95 percent certainty. For zinc, a minimum of seven infants is required to estimate the mean usual intake within 0.2 mg of the true value. This is because the estimated variance of usual intake tends to be small, in particular for infants. For quantities (i.e., “% Inadequacy”) other than means, the required sample sizes are significantly larger.

For women, some samples remained small and the variance large despite combining all pregnant, breastfeeding, and postpartum individuals into one group. To generate more robust nutrient intake estimates of the ratio of the within- to the between-person variance in intake, the method of Jahns et al. (2005) was applied. In this method, the variance ratio estimated from the subgroup intake data is combined with a ratio estimate obtained from the group of all women. To do this, an estimate of within-person variance (external variance) is generated using PC-SIDE to assess intake information of all low-income, pregnant, lactating, or postpartum women in all survey years. An internal ratio estimate is obtained separately for each subgroup. A new within- to between-person variance ratio, is then computed as a weighted average of the external and internal variance ratio estimates. On average, the external variance was weighted by 100, and the

A = Individuals identified as participating in WIC at the time of the survey, NHANES 2005–2008.

B = WIC-eligible nonparticipants (≤ 185% of the poverty income ratio; for women also P, BF, or PP), NHANES 2005–2008.

C = All individuals ≤ 185% of the poverty income ratio, NHANES 2011–2012.

SOURCES: NHANES 2005–2008 and 2011–2012 (USDA/ARS, 2005–2008, 2011–2012).

internal variance was weighted by the number of women in the subgroup who provided 2 days of information. When this number is small (as in the case of pregnant or lactating women in 2011–2012), the external variance plays a larger role in the combined estimate. The resulting estimates are less subject to the large degree of variability in the within-person variance estimate that can be introduced by a small sample size. Both means and the “% Inadequacy” have improved reliability.

For the analysis of episodically consumed foods, small samples add enormous challenges. Neither the National Cancer Institute (NCI) method nor SPADE (used here) results in reliable estimates of distributions of usual food intake when the sample size is small and the proportion of zero consumption is large. In many cases, the programs fail to converge, and no estimation beyond the usual intake mean is possible. Further, neither of the two approaches (NCI or SPADE) permit combining an external and an internal within-person variance estimate when estimating the intake distribution, so the approach followed for nutrients (described above) cannot be implemented for foods. Consequently, with the small sample sizes that were available for women, and the large proportion of zero intakes observed for many of the food subgroups, estimates of the proportion of usual intakes of foods below recommendations are less reliable. Estimates of mean food intake are, however, adequately precise and only these are presented for women (Dekkers et al., 2014).

TABLE 3-4 Tasks Related to Infant Formula Requirements in the Food Packages and the Approach

NOTE: DRI = Dietary Reference Intake; EER = Estimated Energy Requirement; NHANES = National Health and Nutrition Examination Survey.

Tasks Specific to Infant Formulas

In addition to the science supporting functional ingredients in infant formulas, the IOM committee was asked to evaluate three additional aspects of infant formula requirements in the food packages: energy concentration, iron concentration, and volume provided. The three tasks and the evaluation approach are outlined in Table 3-4.

Assessing Diet Quality

The committee was tasked with evaluating the diet quality of WIC-eligible subpopulations using the Healthy Eating Index-2010 (HEI-2010) (Guenther et al., 2013; see Box 3-2) and one additional index of the committee’s choosing. A second index was developed, as detailed in the Letter Report (IOM, 2015):

Briefly, the committee developed a Nutrient-Based Diet Quality (NBDQ) index based on the mean probability of adequacy for the 9 shortfall nutrients, calculated for each individual (see Box 3-2).¹³ The possible scores range from 0 to 100. This approach is very similar to that recently published by Verger et al. (2012), except that the NBDQ includes only shortfall nutrients as defined by the 2015 DGAC report. When tracked with energy intake, the association between the NBDQ index and energy intake was not strong, which suggests that the index is a summary measure that predicts dietary quality beyond simply being a measure of overall energy intakes (see Appendix K, Figures K-1 through K-3). Further details of the committee’s development of NBDQ are described in Appendix K. The NBDQ was applied to all subpopulations excluding infants.

Because it is based on the DGA food patterns, which apply only to individuals ages 2 and older, the HEI was likewise applied only to individuals ages 2 years and older (see Appendix K, Table K-1). The NBDQ was applied to individuals aged 1 year and older because nutrient adequacy can be defined for these individuals based on the EARs or AIs.

Statistical Comparisons in NHANES Analyses

For this report, the only statistical testing of hypotheses conducted by the committee were for a difference between means of WIC participants and eligible non-WIC participant subgroups. Participants in the 2011–2012 NHANES were not included in statistical comparisons because individual samples in these years represented a different time period and the available data combined both WIC participants and eligible nonparticipants. As a result, data from 2011–2012 did not provide an appropriate comparison

__________________

¹³ There are ample precedents for the use of a composite nutrient adequacy index. Mean adequacy ratios have been used for many years and have more recently been updated to reflect the DRIs. The NBDQ is essentially the same as the indexes used in several published studies (Foote et al., 2004; Murphy et al., 2006).

group. In all cases, pairwise t-tests were applied with estimated standard errors that account for the complex design of the NHANES surveys. Tests were implemented for differences in means of the usual intake distributions of nutrients and foods, for the prevalence of inadequate intakes, and for overall mean HEI scores. The NBDQ index, constructed as a combination of estimated percentage of adequacy of nutrients with and without an EAR

was not included in the statistical comparisons because an estimate of the standard error of the mean index requires approximations that are justified only in large samples. Because of the lack of reliability of reported energy intake values (Subar et al., 2015), statistical comparisons were likewise not applied to this measure. A p-value of less than or equal to 0.05 was considered statistically significant.

APPROACH TO DEVELOPMENT OF THE BASELINE FOOD PACKAGE: NUTRIENTS AND COST

Several of the committee’s tasks related to dietary intake estimation and food package costs required an evaluation of baseline packages representative of the foods acquired through the WIC food packages. Accurately representing baseline package composition is fundamental to subsequent (phase II) assessment of changes in nutrient intake, food intake, and cost. The methods used to construct baseline food packages and evaluate their costs are summarized here. The approach used here parallels that applied in the 2006 WIC report (IOM, 2006), but it will use updated food options and selection (redemption) data.

Baseline Food Package Composition and Nutrient Profiles

Each of the food packages prescribed by WIC (see Appendix D, Tables D-1 and D-2) includes specific food categories (e.g., milk or breakfast cereals) with specifications for foods allowed under each category (i.e., skim or 1 percent milk, breakfast cereals with ≤ 6 g sugar per serving). The set of prescribed food categories constitutes the “package” under the revised 2009 food packages. For some food packages, only one choice of food is offered (i.e., whole milk as the “Milk” for children 1 to 2 years of age). However, for other food packages or ages, multiple choices are available within one food category (e.g.,, either skim milk or 1% milk could be chosen within the category of “Milk” for women). To create a baseline “Milk” category from which to evaluate dietary intake and cost changes, the committee will develop a composite of the available options. For example, the committee considered milk choices based on the regulations defining allowed substitutions and rates of substitutions, USDA-FNS studies of state allowed substitutions (USDA/FNS, 2011) and state data on redemptions (which were available from some of the states that are using electronic benefit transfer [EBT] for redemption of WIC benefits). State data on redemption of issued WIC foods is useful for this purpose because it provides information about the proportions (weights) of “Milk” category redemptions that are skim, 1%, yogurt, soy beverage, tofu, or cheese among the available substitutions. These data will be used to develop reasonable selections (allo-

cations) for specific foods. Information from redemption shares as well as allowable substitutions and state options will be used to determine the proportions of each type of food in a food category (e.g., the “Milk” category is 50 percent skim milk, 40 percent 1% milk, 10 percent low-fat yogurt). Nutrient and costs for each food “category” will then be determined from the proportion-weighted component of foods.

The baseline composite food categories containing foods purchased with a cash value voucher (CVV) were computed differently than for other WIC food categories. Because the CVV can be used to purchase many different fruits and vegetables, the composition of baseline representative CVVs for the different categories were computed as weighted averages of several specific items based on their rates of purchase. The contribution (weight) of vegetables (e.g., broccoli) to each vegetable category (e.g., dark greens) will be determined by USDA’s Center for Nutrition Policy and Promotion (CNPP) for use by the 2015 DGAC report (Personal communication, P. Britten, USDA/CNPP, September 24, 2014).

For each composite food category, the relative proportions of different options will be used to construct nutrient profiles. The protocol for estimating these nutrient profiles will be similar to that used in the previous evaluation of WIC food packages (IOM, 2006). Food composition data will be obtained from the USDA National Nutrient Database for Standard Reference, Release 27 (USDA/ARS, 2014). For some foods, nutrient data from USDA will be used without modification (e.g., whole milk). For most foods, however, weighted composite nutrient data will be created, for example, the nutrient profile for “milk” will be composed of nutrients contained in all the various types of milk and milk substitutes included in the baseline composite milk food category, weighted accordingly.

Nutrient profiles for the composite fruit and vegetable food groups and subgroups will be created based on weighted contributions of only those individual fruits and vegetables contributing 5 percent or more to each group or subgroup. Although CVVs can be used to purchase fruits and vegetables in canned, frozen, fresh, or dehydrated forms, depending on state regulation, for baseline compositions used in the phase II report, allocation for most fruits and vegetables was assumed to be in fresh form, because all states are required to allow purchase of this form. This, as well as the relative proportions of foods in the WIC food categories (i.e., types of milk), maybe revised in phase II pending the availability of additional redemption data.

Baseline Prices

To evaluate the costs of the baseline food packages, the committee will need to determine a baseline time period to use for the evaluations. Although July 2015 would be appropriate as this date occurred after imple-

mentation of allowing the purchase of white potatoes with the CVV, as well as the substitution of whole grain pasta (allowed effective May 2014), price and other product and program data for 2015 are limited at the current time (e.g., the yogurt substitution deadline is ongoing) and some data are not available at the time of this report. Therefore, 2014 price and other program data will be used for the initial phase II analysis, with an update to 2015 price data later in phase II.

The average price of each food category in the WIC food package will be determined by assessing prices for qualifying foods (USDA/FNS, 2013b). The same approach will be used for infant formulas. Baseline price data for all food products except fruits and vegetables are available from retail scanner data (from the Information Resources Incorporated, Chicago, Illinois, through a third-party agreement with the ERS). These data will be supplemented, when needed, by other sources such as the Bureau of Labor Statistics national average price data, Internet sources, or local store price data. For fruits and vegetables, ERS price data will be used. Recently, the ERS updated its computation of prices for fruits and vegetables using market purchase data from retail sales data for 2013 (USDA/ERS, 2015a). These 2013 prices will be updated to the most current (2014 or 2015) prices using the relevant Consumer Price Index (CPI) for fresh or processed fruit and vegetables (BLS, 2015).

Determining the Cost and Redemption Rates for the Baseline Package

The cost of the baseline packages will be determined by multiplying the amounts of foods (which vary by package size) by their prices. Available redemption data will be evaluated, with adjustments applied to account for differences among the specific packages. Because redemption data do not account for different redemption rates between women and children for some products (e.g., ready-to-eat cereals), the effects of this variation will be further investigated in the phase II sensitivity analysis. Calculation of program costs for each baseline package will be based on cost, redemption rates, number of participants and, for infant formula, the rate of state contract rebate. All of this information will be presented in phase II.

Limitations of Redemption Data

There are several limitations to the application of redemption data for development of baseline food package nutrient profiles and costs. First, redemption data are not differentiated by package (e.g., food redeemed from a children’s package, or from a woman’s package). Second, it is not possible to extract preferred rates of substitutions (e.g., the substitution of cheese for a portion of milk). Some substitutions may affect cost or nutri-

tional composition. For example, the price and nutritional composition of milk per ounce differs from the price and composition of cheese per ounce. Finally, available state redemption data are limited in applicability on a national level, although the data might provide insights into preferences or product availability. The committee will weigh merits and limitations of the available data in determining the relative product shares for foods in the representative WIC packages.

SENSITIVITY ANALYSIS APPROACH

The committee is tasked with conducting a sensitivity analysis in phase II to assess the effect of potential food package changes on nutrient composition of each package relative to the DRIs, food groups and subgroups relative to the 2015 DGA recommendations, and cost. Changes in nutrients, food groups, and costs will be determined for each proposed change in the food package relative to the baseline composite food packages described above. The planned approach for this analysis is outlined here.

Developing a List of Potential Package Changes

To evaluate the effect of changes to the food packages, the committee first plans to develop a list of potential changes. This could include, for example, changes in food categories (e.g., specific foods added, increased or decreased quantities, changes in the value of the CVV) and changes in combinations of the package components (i.e., allowable substitutions and alternates, with respective changes in substitution or redemption assumptions). Combinations will be tested and compared to the “baseline food package” to ensure that any changes being considered are, overall (for the WIC program), cost neutral or not more than 10 percent above or below the current level of funding.

Testing Changes to the Food Packages

The committee plans to consider food package changes based on consideration of the totality of evidence. The sensitivity analysis will determine the effect of any change on nutrient intake, food intake, and cost. For all WIC food categories within the baseline food packages, the committee plans to evaluate options to add/eliminate/increase/decrease/alter the baseline composition. The effects of each food change will be assessed at the food package level (i.e., how each food package recipient would be affected) for changes in nutrient intake, food group (i.e., dairy) and food subgroup (i.e., milk) intake, and cost. For each option explored, an assumption will be assigned regarding any change in the “weight” of the foods within the

composite packages. For example, if a new food were added, would it be expected to change redemptions of the foods in that composite package?

As with the nutrient profiles for the baseline composite food packages, nutrient data for each food change will come from the USDA Standard Reference Database, Release 27 (USDA/ARS, 2014). Should major changes to the food packages be considered, the amount of change in nutrient intake will be evaluated in terms of its effect on the risks of nutrient inadequacy by adjusting the intake distribution by the amount of the nutrient change. For minor changes, the amount of change in nutrient intake will be assessed without looking at distribution shifts. Changes in food group and subgroup intake will be evaluated with respect to changes in the degree to which 2015 DGA food group recommendations are met. Finally, cost changes will be evaluated for all food and combination changes.

Qualitative Assessment of Food Package Changes

The committee plans to consider additional dimensions that could be affected by changes to the food packages. These include the effects of changes on participation (uptake) for the package and/or effects on the redemption rates of each package. The likely effects will be based on available data on current redemption rates and literature reviewed. These changes will be important to consider when conducting the Regulatory Impact Analysis (RIA) (see below), and major changes may be included as an option in the RIA.

Variations from Cost Neutral

While the committee was tasked with ensuring overall cost neutrality for recommended changes to the WIC food packages, they were also asked to offer prioritized recommendations in the event that USDA-FNS’s WIC funding is either above or below the cost-neutral level. These priorities will appear in the phase II report.

FOOD EXPENDITURE ANALYSIS

The committee was tasked with the planning and implementation of a food expenditure analysis for the WIC population using nationally representative purchasing and price data. A summary of the data sources is described here, details of the analysis are presented in Appendix L, and the results discussed in Chapter 10. A portion of this task included determining expenditures on food groups. This task will be completed in phase II. The Information Resources Incorporated (IRI) household panel scanner and the National Household Food Acquisition and Purchasing Survey

(FoodAPS) data were acquired in phase I, however the process was lengthy and did now allow adequate time to conduct analysis of food group data for the expenditure analysis. In addition, the work required to match foods acquired (FoodAPS) to the USDA food groups is extensive and was not feasible in the time allotted to produce the phase I report.

Data Sources

Sources of Purchasing Data

Nationally representative data on food expenditures by WIC households are limited. However, data collected as part of the USDA’s FoodAPS have recently been released (USDA/ERS, 2015b). Using these data, the committee compared shopping patterns of WIC participants, based on categorical eligibility and self-report, to low-income and higher-income nonparticipants. FoodAPS is a nationally representative survey of 4,826 American households, covering 14,317 individuals, that provides detailed information about foods purchased or otherwise acquired for consumption at home and away from home between April 2012 and January 2013. The survey includes identifiers for households reporting participation in WIC and reports whether a WIC voucher was used in a food acquisition transaction.

Another source of data available for analysis of food product purchase is in the 2011 and later IRI household panel scanner data on household purchases from retail stores. The data cover the 48 continental states. Participating households use a scanner at home to record retail food purchases after shopping and the resulting information includes items purchased, quantities bought, amount of money paid, and date of purchase. Household scanner data panelists are instructed to scan all purchases from all outlets, including supermarkets, supercenters, club stores, convenience stores, drugstores, farmers’ markets, and other types of retail facilities. The household panel scanner data provide information on the purchases of a large number of households and can be used to assess expenditures and quantities of detailed products that may be evaluated in determining likely costs of baseline and alternative package foods. Sample weights will be applied to derive nationally representative estimates of retail food purchases and unit values (prices) for all households across the contiguous United States. The primary subpopulation of interest in the IRI household panel scanner dataset is low-income households. In addition, households with young children present can be identified.

Sources of Price Data

For the analysis conducted in this report, two sources of price data were available: IRI retail scanner data and USDA ERS data on fruit and vegetable prices (USDA/ERS, 2015a). As described previously, these are the same data sources used to determine prices for the baseline composite food packages. The IRI scanner data allow estimation of quantity-weighted prices for aggregated food groups representative of WIC package foods. Price data developed for the Thrifty Food Plan with food group quantities updated to reflect the 2010 DGA are not available. As with price data used for determining prices of the baseline composite food packages, all prices will be updated to the 2014 base year using the Bureau of Labor Statistics (BLS)-CPI for food at home.

Information on household food expenditures comes from sources listed in Table 3-5. The sources not available in time for delivery of this report will continue to be pursued for phase II, and the committee is open to the identification of additional resources. Analysis of food expenditures conducted during phase I focused on the reported expenditures (transactions to purchase and acquire food) in the FoodAPS.

APPROACH TO THE REGULATORY IMPACT ANALYSIS

The committee developed an approach for a RIA to be conducted during phase II and based on the approach detailed in the Office of Information and Regulatory Analysis document, “Regulatory Impact Analysis: A Primer” (OIRA, 2011). The objective of the RIA will be to evaluate the effect of the committee’s recommended changes in WIC food packages on program participation, value of selected food packages, and program cost and administration. Details of the proposed RIA approach are presented in Appendix M.

NATIONWIDE DISTRIBUTION AND COSTS OF FOOD

Also during phase II, the committee will gather information on the nationwide costs and distribution of foods (including low-income neighborhoods). Part of the purpose of this is to ensure that the new food packages are efficient for nationwide distribution. Particularly, all of the specific changes recommended for the WIC food packages should be based on consideration of whether it is feasible to make the recommended foods available, from both the perspective of federal/state administration in allowing local agencies to make substitutions (i.e., select combinations from among the WIC-approved foods) and the perspective of vendors that directly provide the foods included in the packages. Variability in seasonal availability,

TABLE 3-5 Availability of Nationally Representative Price and Expenditure Datasets as of November 2015

NOTE: FoodAPS = National Household Food Acquisition and Purchasing Survey; IRI = Information Resources Incorporated; TFP = Thrifty Food Plan; USDA-CNPP = U.S. Department of Agriculture-Center for Nutrition Policy and Promotion; USDA-ERS = U.S. Department of Agriculture-Economic Research Service.

seasonal pricing, and types of vendors available in different locales (e.g., supermarket versus trading post) will be factored into the recommendations. Issues of local distribution (e.g., availability of neighborhood grocery outlets) will be considered. All output will be provided in the final report.

COMMITTEE WIC SITE VISITS AND SHOPPING EXPERIENCE

USDA-FNS asked that the majority of committee members visit a state WIC clinic and experience shopping as a WIC participant prior to development of the phase II report. Between March and June 2015, committee members visited a total of 14 WIC sites and vendors either in their home state, another state, or both. The visits were organized to ensure geographic and cultural diversity, a balance of sites issuing paper vouchers versus using EBT, committee member availability, site staff availability, and activity at the site (e.g., days of greater participant flow and provision of group education). A list of sites visited by city and state is presented in Table 3-6. The committee members adhered to the following agenda during site visits:

• Become familiar with the flow of clinic operations and intake.
• If possible, observe a WIC enrollment from start to finish. Alternatively, observe a WIC certification appointment from start to finish.

TABLE 3-6 WIC Sites Visited by the Committee to Review WIC Food Packages

• If occurring at the time of the visit, observe a group education class.
• If occurring at the time of the visit, observe a prenatal and/or breastfeeding class.
• Observe the orientation to WIC foods and use the voucher/EBT card.
• If a breastfeeding peer counselor is available, learn about delivery of such services at that site.
• Obtain an EBT card or voucher to complete the shopping experience.
• Visit a local WIC authorized vendor to locate and purchase WIC foods.

Committee members prepared written reports and shared their experiences during a closed meeting. A summary of the committee’s key observations is presented in Appendix N.

REFERENCES

AAP (American Academy of Pediatrics). 2001. Provisional section on breastfeeding: WIC program. Pediatrics 108(5):1216-1217.

AAP. 2012. Breastfeeding and the use of human milk. Pediatrics 129(3):e827-e841.

AAP. 2014. Pediatric Nutrition, 7th Edition. Edited by R. E. Kleinman and F. R. Greer. Elk Grove Village, IL: American Academy of Pediatrics.

AAP, Committee on Nutrition. 2008. Lipid screening and cardiovascular health in childhood. Pediatrics 122(1):198-208.

AAP, Committee on Nutrition. 2010. Diagnosis and prevention of iron deficiency and iron-deficiency anemia in infants and young children (0–3 years of age). Pediatrics 126(5):1040-1050.

Altoni, J. G., T. E. Elder, C. R. Taber. 2005. Selection on observed and unobserved variables: Assessing the effectiveness of Catholic schools. Journal of Political Economy 113(1):151-184.

Altoni, J. G., T. E. Elder, C. R. Taber. 2008. Using selection on observed variables to assess bias from unobservables when evaluating Swan-Ganz catheterization. American Economic Association Papers and Proceedings 98(2):345-350.

AND (Academy of Nutrition and Dietetics). 2015. Position of the American Dietetic Association: Promoting and supporting breastfeeding. Journal of the Academy of Nutrition and Dietetics 115:444-449.

Beaton G. H. 1994. Approaches to analysis of dietary data: Relationship between planned analyses and choice of methodology. American Journal of Clinical Nutrition 59(Suppl 2):53S-61S.

Besharov, D. J., and P. Germanis. 2001. Rethinking WIC: An evaluation of the women, infants, and children program: Washington, DC: AEI Press.

Bitler, M. P., and J. Currie. 2005. Does WIC work? The effects of WIC on pregnancy and birth outcomes. Journal of Policy Analysis and Management 24(1):73-91.

BLS (Bureau of Labor Statistics). 2015. CPI detailed report: Data for January 2015. Washington, DC: U.S. Department of Labor. http://www.bls.gov/cpi/cpid1501.pdf (accessed June 10, 2015).

Butte, N. F., W. W. Wong, T. A. Wilson, A. L. Adolph, M. R. Puyau, and I. F. Zakeri. 2014. Revision of Dietary Reference Intakes for energy in preschool-age children. American Journal of Clinical Nutrition 100(1):161-167.

Carriquiry, A. L. 1999. Assessing the prevalence of nutrient inadequacy. Public Health Nutrition 2(1):23-33.

Cashman, K. D. 2012. The role of vitamers and dietary-based metabolites of vitamin D in prevention of vitamin D deficiency. Food and Nutrition Research 56.

Cashman, K. D., K. M. Seamans, A. J. Lucey, E. Stöcklin, P. Weber, M. Kiely, and T. R. Hill. 2012. Relative effectiveness of oral 25-hydroxyvitamin D3 and vitamin D3 in raising wintertime serum 25-hydroxyvitamin D in older adults. American Journal of Clinical Nutrition 95(6):1350-1356.

CDC (Centers for Disease Control and Prevention). 2014. National Health and Nutrition Examination Survey: Sample design, 2011–2014. Hyattsville, MD. http://www.cdc.gov/nchs/data/series/sr_02/sr02_162.pdf (accessed September 30, 2015).

Currie, J., and I. Rajani. 2015. Within-mother estimates of the effects of WIC on birth outcomes in New York City. Economic Inquiry 53(4):1691-1701.

Dekkers, A. L., J. Verkaik-Kloosterman, C. T. van Rossum, and M. C. Ocke. 2014. SPADE, a new statistical program to estimate habitual dietary intake from multiple food sources and dietary supplements. Journal of Nutrition 144(12):2083-2091.

EOPUS (Executive Office of the President of the United States). 2014. The economic impact of the American Recovery and Reinvestment Act five years later: Final report to Congress. Washington, DC: Executive Office of the President of the United States. https://www.whitehouse.gov/sites/default/files/docs/cea_arra_report.pdf (accessed August 24, 2015).

FDA (U.S. Food and Drug Administration). 2015. Factsheet on the new proposed nutrition facts label. http://www.fda.gov/Food/GuidanceRegulation/GuidanceDocumentsRegulatoryInformation/LabelingNutrition/ucm387533.htm (accessed August 26, 2015).

Fewtrell, M. S., J. B. Morgan, C. Duggan, G. Gunnlaugsson, P. L. Hibberd, A. Lucas, and R. E. Kleinman. 2007. Optimal duration of exclusive breastfeeding: What is the evidence to support current recommendations? The American Journal of Clinical Nutrition 85(2):635S-638S.

Foote, J. A., S. P. Murphy, L. R. Wilkens, P. P. Basiotis, and A. Carlson. 2004. Dietary variety increases the probability of nutrient adequacy among adults. Journal of Nutrition 134(7):1779-1785.

Fuller, W. 2009. Sampling statistics. Hoboken, NJ: John Wiley & Sons.

Guenther, P. M., K. O. Casavale, J. Reedy, S. I. Kirkpatrick, H. A. Hiza, K. J. Kuczynski, L. L. Kahle, and S. M. Krebs-Smith. 2013. Update of the healthy eating index: HEI-2010. Journal of the Academy of Nutrition and Dietetics 113(4):569-580.

HHS (U.S. Department of Health and Human Services). 2015. 2015 Federal poverty guidelines. http://aspe.hhs.gov/2015-poverty-guidelines (accessed September 8, 2015).

IOM (Institute of Medicine). 1997. Dietary reference intakes for calcium, phosphorus, magnesium, vitamin D, and fluoride. Washington, DC: National Academy Press.

IOM. 1998. Dietary reference intakes for thiamin, riboflavin, niacin, vitamin B6, folate, vitamin B12, pantothenic acid, biotin, and choline. Washington, DC: National Academy Press.

IOM. 2000a. Dietary reference intakes for vitamin C, vitamin E, selenium, and carotenoids. Washington, DC: National Academy Press.

IOM. 2000b. Dietary reference intakes: Applications in dietary assessment. Washington, DC: National Academy Press.

IOM. 2001. Dietary reference intakes for vitamin A, vitamin K, arsenic, boron, chromium, copper, iodine, iron, manganese, molybdenum, nickel, silicon, vanadium, and zinc. Washington, DC: National Academy Press.

IOM. 2002/2005. Dietary reference intakes for energy, carbohydrate, fiber, fat, fatty acids, cholesterol, protein and amino acids. Washington, DC: The National Academies Press.

IOM. 2003. Dietary reference intakes: Applications in dietary planning. Washington, DC: The National Academies Press.

IOM. 2005. Dietary reference intakes for water, potassium, sodium, chloride, and sulfate. Washington, DC: The National Academies Press.

IOM. 2006. WIC food packages: Time for a change. Washington, DC: The National Academies Press.

IOM. 2011a. Dietary reference intakes for calcium and vitamin D. Washington, DC: The National Academies Press.

IOM. 2011b. Child and adult care food program. Washington, DC: The National Academies Press.

IOM. 2011c. Early childhood obesity prevention policies. Washington, DC: The National Academies Press.

IOM. 2015. Review of WIC Food Packages: An evaluation of white potatoes in the cash value voucher: Letter report. Washington, DC: The National Academies Press.

Jahns, L., L. Arab, A. Carriquiry, and B. M. Popkin. 2005. The use of external within-person variance estimates to adjust nutrient intake distributions over time and across populations. Public Health Nutrition 8(1):69-76.

Johnson, C. L., S. M. Dohrmann, V. L. Burt, and L. K. Mohadjer. 2014. National Health and Nutrition Examination Survey: Sample design, 2011–2014. National Center for Health Statistics. Vital and Health Statistics 2(162):1-33.

Kreider, B., J. V. Pepper, and M. Roy. 2015. Identifying the effects of WIC on food insecurity among infants and children. Southern Economic Journal.

Murphy, S. P., J. A. Foote, L. R. Wilkens, P. P. Basiotis, A. Carlson, K. K. White, and K. M. Yonemori. 2006. Simple measures of dietary variety are associated with improved dietary quality. Journal of the American Dietetic Association 106(3):425-429.

NHLBI (National Heart, Lung, and Blood Institute). 2011. Expert panel on integrated guidelines for cardiovascular health and risk reduction in children and adolescents: Summary report. Bethesda, MD: NIH. http://www.nhlbi.nih.gov/health-pro/guidelines/current/cardiovascular-health-pediatric-guidelines/summary (accessed May 24, 2015).

Nusser, S. M., A. L. Carriquiry, K. W. Dodd, and W. A. Fuller. 1996. A semiparametric transformation approach to estimating usual daily intake distributions. Journal of the American Statistical Association 91(436):1440-1449.

OIRA (Office of Information and Regulatory Affairs). 2011. Regulatory impact analysis: A primer. Washington, DC: Office of Management and Budget, The White House. https://www.whitehouse.gov/sites/default/files/omb/inforeg/regpol/circular-a-4_regulatoryimpact-analysis-a-primer.pdf (accessed August 26, 2015).

Raiten, D. J., R. Raghavan, A. Porter, J. E. Obbagy, and J. M. Spahn. 2014. Executive summary: Evaluating the evidence base to support the inclusion of infants and children from birth to 24 mo of age in the Dietary Guidelines for Americans—the B-24 project. American Journal of Clinical Nutrition 99(3):663S-691S.

Rush, D., M. R. Kurzon, W.B. Seaver, and D.S. Shanklin. 1988a. The National WIC Evaluation: evaluation of the Special Supplemental Food Program for Women, Infants, and Children. VII. Study of food expenditures. American Journal of Clinical Nutrition 48(2 Suppl):512-519.

Rush, D., J. Leighton, N. L. Sloan, J. M. Alvir, D. G. Horvitz, W. B. Seaver, G. C. Garbowski, S. S. Johnson, R. A. Kulka, J. W. Devore et al. 1988b. The National WIC Evaluation: evaluation of the Special Supplemental Food Program for Women, Infants, and Children. V. Longitudinal study of pregnant women. American Journal of Clinical Nutrition 48(2 Suppl):512-519.

Rush, D., J. Leighton, N. L. Sloan, J. M. Alvir, G. C. Garbowski. 1988c. The National WIC Evaluation: evaluation of the Special Supplemental Food Program for Women, Infants, and Children. II. Review of past studies of WIC. American Journal of Clinical Nutrition 48(2 Suppl):394-411.

Rush, D., J. M. Alvir, D. A. Kenny, S. S. Johnson, and D. G. Horvitz. 1988d. The National WIC Evaluation: evaluation of the Special Supplemental Food Program for Women, Infants, and Children. III. Historical study of pregnancy outcomes. American Journal of Clinical Nutrition 48(2 Suppl):412-428.

Subar, A. F., L. S. Freedman, J. A. Tooze, S. I. Kirkpatrick, C. Boushey, M. L. Neuhouser, F. E. Thompson, N. Potischman, P. M. Guenther, V. Tarasuk, J. Reedy, and S. M. Krebs-Smith. 2015. Addressing current criticism regarding the value of self-report dietary data. Journal of Nutrition 145(12):2639-2645.

Taylor, C. L., A. L. Carriquiry, R. L. Bailey, C. T. Sempos, and E. A. Yetley. 2013. Appropriateness of the probability approach with a nutrient status biomarker to assess population inadequacy: A study using vitamin D. American Journal of Clinical Nutrition 97(1):72-78.

USDA/ARS (U.S. Department of Agriculture/Agricultural Research Service). 2005–2008. What we eat in America, NHANES 2005–2008. Beltsville, MD: USDA/ARS (accessed December 15, 2014).

USDA/ARS. 2011–2012. What we eat in America, NHANES 2011–2012. Beltsville, MD: USDA/ARS (accessed December 15, 2014).

USDA/ARS. 2013. Food patterns equivalents database 2007–08: Methodology and user guide. Beltsville, MD: USDA/ARS. http://www.ars.usda.gov/SP2UserFiles/Place/80400530/pdf/fped/FPED_0708.pdf (accessed December 15, 2014).

USDA/ARS. 2014. USDA national nutrient database for standard reference, release 27. Beltsville, MD: USDA/ARS. http://www.ars.usda.gov/ba/bhnrc/ndl (accessed December 15, 2014).

USDA/ERS (U.S. Department of Agriculture/Economic Research Service). 2015a. Fruit and vegetable prices. Washington, DC: USDA/ERS. http://ers.usda.gov/data-products/fruitand-vegetable-prices.aspx (accessed June 9, 2015).

USDA/ERS. 2015b. FoodAPS national household food acquisition and purchase survey. Washington, DC: USDA/ERS. http://www.ers.usda.gov/data-products/foodaps-nationalhousehold-food-acquisition-and-purchase-survey.aspx (accessed August 26, 2015).

USDA/FNS (U.S. Department of Agriculture/Food and Nutrition Service). 2011. WIC food packages policy options study, final report. Alexandria, VA: USDA/FNS. http://www.fns.usda.gov/sites/default/files/WICFoodPackageOptions.pdf (accessed December 9, 2014).

USDA/FNS. 2012. Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) erroneous payments to vendors: Annual estimates for FY 2011. Alexandria, VA: USDA/FNS. http://www.fns.usda.gov/sites/default/files/WICErroneousPayment_2011.pdf (accessed August 26, 2015).

USDA/FNS. 2013a. WIC participant and program characteristics 2012 final report. Alexandria, VA: USDA/FNS. http://www.fns.usda.gov/sites/default/files/WICPC2012.pdf (accessed December 20, 2014).

USDA/FNS. 2013b. Fiscal year 2010: WIC food cost report. Alexandria, VA: USDA/FNS. http://www.fns.usda.gov/sites/default/files/WICFoodCost2010_0.pdf (accessed June 8, 2015).

USDA/FNS. 2015. WIC income eligibility guidelines. http://www.fns.usda.gov/wic/wic-income-eligibility-guidelines (accessed August 26, 2015).

USDA/HHS (U.S. Department of Agriculture/U.S. Department of Health and Human Services). 2015. The report of the Dietary Guidelines Advisory Committee on the Dietary Guidelines for Americans, 2015, to the Secretary of Agriculture and the Secretary of Health and Human Services. Washington, DC: USDA/HHS. http://www.health.gov/dietaryguidelines/2015-scientific-report/PDFs/Scientific-Report-of-the-2015-DietaryGuidelines-Advisory-Committee.pdf (accessed May 24, 2015).

Verger, E. O., F. Mariotti, B. A. Holmes, D. Paineau, and J. F. Huneau. 2012. Evaluation of a diet quality index based on the probability of adequate nutrient intake (PANDIET) using national French and U.S. dietary surveys. PLoS ONE 7(8):e42155.

WHO (World Health Organization). 2009. Infant and young child feeding. Geneva, Switzerland: World Health Organization.