Persistence and Emergence of Anemia in Children During Participation in the Special Supplemental Nutrition Program for Women, Infants, and Children FREE

IRON-DEFICIENCY ANEMIA occurs most commonly in children aged 9 to 18 months and can have lasting effects on cognitive development.^1- 4 The prevalence of iron-deficiency anemia has decreased dramatically during the past few decades, likely because of a national increase in breastfeeding, standardization of the composition of iron-fortified formulas, and delay in the introduction of whole milk until age 1 year. By providing participants with iron-fortified formula and cereal and vouchers for iron-containing foods the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) has also played a substantial role in this decline.⁵

Infants and children 5 years and younger are eligible for enrollment in WIC if family income is below 185% of the US Poverty Income Guidelines and if they are at nutritional risk. Nutritional risk may be medically based, such as being anemic or underweight, or diet based, such as an inadequate diet. To prove nutritional risk, participants must provide a 24-hour diet history, recent weight and height, and a recent hemoglobin or hematocrit level. There is national variability in the frequency of blood tests required for WIC participation. In 1998, the US Department of Agriculture, which oversees the WIC program, recommended that all state WIC agencies follow the Centers for Disease Control and Prevention (CDC) recommendation to screen for anemia once between ages 9 and 12 months, once 6 months later, and then annually from ages 2 to 5 years.⁶ Illinois requires anemia screening every 6 months for WIC recertification.

Using a computer database of WIC participants from 3 Chicago sites, we examined the prevalence and persistence of anemia. We sought to identify factors, such as sex, age, race or ethnicity, birth weight, and weight for height, that could be used to focus anemia screening in this population. In addition, we hypothesized that anemic children identified by the WIC program would develop normal hemoglobin concentrations over time.

In 1995, Illinois implemented Project Cornerstone, a computer database used to maintain WIC participant visit information, including demographics, anthropometric data, and laboratory test results. We analyzed Project Cornerstone data for all infants and children aged 6 months to 5 years who visited any of 3 Chicago WIC sites between March 3, 1997, to March 2, 1999. The Illinois Department of Public Health and the administrators of the WIC sites from which the data were obtained made this anonymous data set available to us.

The analytic data set included the following variables: date of birth, date of visit, race, ethnicity, hemoglobin level, hematocrit level, date of hemoglobin and hematocrit measurement, weight, height, and birth weight. For each child, a race or ethnicity variable was determined: Hispanic, non-Hispanic black, non-Hispanic white, or other. Only visits with a hemoglobin measurement when the child was aged 6 to 59 months were considered in analyses. Only hemoglobin values were used (rather than hemoglobin and hematocrit levels) because most visits did not have a recorded hematocrit value. Furthermore, visit data were excluded if the date of the hemoglobin test reported for that visit was after the visit date or if the hemoglobin value was outside an extreme range (<5.0 and >20.0 g/dL). Data from a subsequent visit that was within 3 months of a previous visit were excluded. Extreme birth weight values (<500 or >6000 g) were considered suspect and were excluded from analysis.

Height and weight measurements were taken at health care provider visits and were recorded on forms presented at the time of the WIC visits. Measurement percentiles and z scores were determined using Epi Info software.⁷ Computed weight-for-height z scores for each child were compared across visits. Anthropometric data on children with a z score change greater than 1 between consecutive visits (suggesting a spurious measurement) were excluded. Anthropometric data on children who had only 1 visit were not considered in the analyses.

Data were grouped by age at first measurement in the study using 2 methods. The first grouping method, 6 to 23 and 24 to 59 months, follows the age separation used by WIC and the CDC to define their cutoff values for anemia. The second grouping method, 6 to 8, 9 to 23, 24 to 35, and 36 to 59 months, was used to examine age trends that might affect the development of anemia. In analyses involving race or ethnicity, only 3 race or ethnicity groups were considered: Hispanic, non-Hispanic black, and non-Hispanic white. Initial analyses examined data cross-sectionally and included only a first visit for each child in the data period.

The data were analyzed using 2 definitions of anemia. First, the national WIC hemoglobin cutoff values of less than 11.0 g/dL for children aged 6 to 23 months and less than 11.1 g/dL for children aged 24 to 59 months were used for all children. In addition, non-Hispanic black children were evaluated using the CDC recommended cutoff value for non-Hispanic black children (<10.6 g/dL for children aged 6-23 months and <10.7 g/dL for those aged 24-59 months). These data sets were designated Anemia1. Second, a more stringent hemoglobin cutoff value to define anemia was used for each group of children (Hispanic and non-Hispanic white children: <10.5 g/dL at 6-23 months and <10.6 g/dL at 24-59 months; and non-Hispanic black children: <10.1 g/dL at 6-23 months and <10.2 g/dL at 24-59 months) to avoid misclassification of patients.⁵ These data sets were designated Anemia2. The binomial proportion test was used to evaluate the change in percentage of anemia between the 2 definitions. χ² Tests were used to investigate associations between categorical variables. The Mantel-Haenszel test was used to examine an association between anemia and the 4 age groups because age group was an ordinal variable in the analyses. For all analyses, statistical significance was defined as P<.05.

Logistic regression models were used to determine the relationship between anemia and the following independent variables: sex, race or ethnicity (Hispanic, non-Hispanic black, and non-Hispanic white), age group (6-8, 9-23, 24-35, and 36-59 months), birth weight (<2500 vs ≥2500 g), and weight-for-height z score. Data were also evaluated longitudinally to determine the incidence of anemia over multiple visits and the utility of anemia at the first visit to predict anemia at later visits. Approval for this study was obtained from the institutional review board of Children's Memorial Hospital.

The initial data set included 7065 infants and children aged 6 to 59 months who had 17 283 visits for which a hemoglobin measure (valid or not) was recorded. After review of the data, 243 visits were excluded for the following reasons: extreme hemoglobin values (n = 68), an inappropriately short interval (<3 months) between visits (n = 107), and the date of the laboratory test was reported to be after the WIC visit (n = 68). Thus, 7053 children with 17 040 visits were included in the analysis. The demographics of the children and the characteristics of the visits are given in Table 1.

Figure 1 demonstrates the distribution of the hemoglobin values of the non-Hispanic black children compared with those of all the Hispanic and non-Hispanic white children. When the CDC cutoff value for anemia for black children was used, the percentage of non-Hispanic black children defined as anemic decreased statistically significantly to 8.0%, compared with 15.8% using the WIC cutoff value (binomial proportion test, P<.001). In the univariate analyses, only age was statistically significantly associated with Anemia1 (χ², P = .01); sex, race or ethnicity, birth weight, and weight-for-height z score were not associated with anemia. Anthropometric data for 2.2% of the children were excluded from analysis owing to changes in the weight-for-height z scores in excess of 1 SD between 2 visits.

Using the less restrictive definition, anemia was more common in the 6- to 23-month-old children than those aged 24 to 59 months (17.9% vs 11.7%; χ², P = .001). The prevalence of anemia at the first visit in the 4 previously defined age groups is shown in Figure 2; there was an inverse relationship between anemia and age (Mantel-Haenszel, P<.001). Children aged 6 to 8 months were 3.3 times more likely than children aged 36 to 59 months to be anemic (95% confidence interval [CI], 2.4-4.5). Children aged 9 to 23 months and 24 to 35 months were each at 2.0 times greater risk than 36- to 59-month-old children (9-23 months: 95% CI, 1.4-2.8; 24-35 months: 95% CI, 1.3-2.8).

Using the more restrictive definition, anemia at the first visit was more common in 6- to 23-month-old children than in those aged 24 to 59 months (9.0% vs 5.2%; χ², P = .001). The prevalence of anemia in the 4 previously defined age groups is shown in Figure 2; an inverse relationship between anemia and age existed in this subgroup of patients as well (Mantel-Haenszel, P<.001). Children aged 6 to 8 months were 3.3 times more likely than those aged 36 to 59 months to be anemic (95% CI, 2.4-4.7). Children aged 9 to 23 months and 24 to 35 months were each at 2.0 times greater risk than 35- to 59-month-old children (9-23 months: 95% CI, 1.4-2.9; and 24-35 months: 95% CI, 1.3-2.9).

We next examined the longitudinal changes in the incidence and persistence of anemia by age and race using only data from children who had at least 3 visits during the study (n = 2926). Figure 3 displays these longitudinal changes using both the less and more stringent definitions of anemia for children aged 6 to 23 months; Figure 4 displays similar data for children aged 24 to 59 months. The 6- to 23-month-old children were more likely than those aged 24 to 59 months to remain anemic or to develop anemia at a second visit after having had an initial normal hemoglobin concentration using either definition (P = .001). However, there was a great deal of crossover between the anemic and nonanemic states in both age groups regardless of the definition of anemia used. Race was not a significant factor in predicting which children would either remain anemic or develop anemia during the study.

In the 6- to 23-month age group, anemia developed in 11.5% of the children at the second visit and in an additional 7.1% (113 of 1585 children) at the third visit. Approximately one third of the children who were anemic at either the first or second visit remained anemic at the subsequent visit. Moreover, 6.6% of the children (113 of 1402 in the 6- to 23-month-old group and 35 of 807 in the 24- to 59-month-old group) who had 2 previous normal hemoglobin measurements developed anemia at a third visit, and 1.8% (40 of 1927 in the 6- to 23-month-old group and 6 of 999 in the 24- to 59-month-old group) remained anemic for 3 consecutive visits.

Using the more stringent definition, anemia developed in 6.4% of the nonanemic 6- to 23-month-old children at the second visit and in an additional 3.9% (68 of 1757 children) at the third visit. Nearly 19% of initially anemic children in both age groups remained anemic at the second visit, and nearly one third of these children were still anemic at the third visit.

The mean (SD) time between the first and second visits was 6.9 (2.5) months; the mean (SD) time between the second and third visits was 6.8 (1.7) months. Using Anemia1 definitions, we looked at the changes in hemoglobin concentration for children who had at least 2 visits (n = 6103). Children who had a normal hemoglobin measurement at a first visit and who were anemic at a second visit had a mean (SD) decrease in hemoglobin concentration of 1.6 (1.2) g/dL; they were below the WIC cutoff value by a mean (SD) of 0.7 (0.7) g/dL. Of 505 children who had a normal hemoglobin level at the first visit and who became anemic at any subsequent visit, 77.4% were less than 1.0 g/dL below the WIC cutoff value. Of children who were first anemic at the third or fourth visit after having had a normal hemoglobin level at the previous visit, 78.9% and 86.7%, respectively, were still within 1.0 g/dL of the cutoff value.

The prevalence of anemia in this population was as high as 17.9% in selected subgroups. By a second visit, 10.7% of children had developed anemia, and 30.5% of initially anemic children remained anemic while participating in WIC. Most anemic children, however, were below the cutoff value by less than 1.0 g/dL of hemoglobin. Although their anemia may represent iron deficiency, this can only be an assumption without further assessment of iron status.

Factors that have been thought to contribute to WIC's success in lowering anemia in the past include (1) improved iron intake owing to the receipt of vouchers for iron-containing foods, (2) extensive nutrition counseling, and (3) frequent screening for anemia.⁵ Contrary to our hypothesis, many children in the longitudinal analyses either developed anemia or remained anemic over time, indicating that participation in WIC may be less effective in preventing and treating anemia than previously reported.

Several potential reasons for these findings exist. First, although WIC participants are encouraged to attend 2 nutrition education sessions in each 6-month period to be recertified, the efficacy of this education has not been well established. Second, children in WIC who are anemic are encouraged, but not required, to seek medical attention for appropriate treatment. Although health care providers perform hemoglobin testing, there is no system in place that ensures that families return to a physician for needed therapy. Finally, the choice of an appropriate definition of anemia may partly explain its high prevalence in our population. We chose a conservative approach and used the WIC cutoff value for anemia for Hispanic and non-Hispanic white children and the CDC's lower definition of anemia for non-Hispanic black children. Because most of the children who became anemic in this study had levels less than 1.0 g/dL below the WIC cutoff value, many may have been only transiently anemic and misclassified as such. In addition, the relationship between anemia and iron deficiency in this cohort is unknown. Intercurrent illnesses may have contributed to the anemia in some children.⁸

Determining an effective screening test for iron deficiency in children has proved to be difficult. Screening for iron-deficiency anemia in high-risk children using dietary and health history, as recommended by the Institute of Medicine and the CDC, has been shown to have only modest sensitivity and poor specificity.⁹ As anemia is a late manifestation of iron deficiency, children may be iron deficient and have normal hemoglobin or hematocrit values.¹⁰ Although hemoglobin and hematocrit measurements are well standardized and easily performed, they are not ideal screening tests for iron deficiency, especially if used alone without the aid of other red blood cell indices such as the mean cell volume or the red blood cell distribution width.

New strategies need to be implemented to ensure follow-up and treatment of anemia. One such strategy could be to require proof of physician awareness and treatment of anemia for WIC recertification. Alternatively, formation of a multistep screening process might be developed in which recertification of an anemic child would require measurement of either serum ferritin or complete red blood cell indices. A recent position paper by the Provisional Section on Breastfeeding of the American Academy of Pediatrics¹¹ highlighted the need to establish uniform procedures within WIC that would ensure the medical follow-up and treatment of anemic children. Perhaps even more controversial would be relocation of the WIC program out of the Department of Agriculture and into the Department of Health and Human Services, where mandatory linkages between WIC sites and health clinics might ensure closer follow-up and treatment.

In conclusion, despite frequent screening, food supplementation, and nutritional counseling, the prevalence of anemia among infant and child WIC participants was high. Very young children were at highest risk. A significant percentage of WIC recipients either developed anemia or remained anemic while enrolled in the program. These data support the use of a frequent screening strategy, such as that recommended by the CDC, in this high-risk population. Consideration should be given to implementation of mandatory follow-up by WIC or health care providers of all anemic infants to ensure accuracy of the diagnosis and, for most, repletion of their iron stores and resolution of their anemia.

Accepted for publication May 29, 2002.

This study was presented in part at the Midwest Society for Pediatric Research Annual Meeting, Chicago, Ill, September 17, 1999, and at the Pediatric Academic Societies Annual Meeting, Boston, Mass, May 15, 2000.

We thank Erlinda Binghay, RD, MPH, of the Near North Health Corp in Chicago for assistance in obtaining the data set.

Previous studies have shown that the prevalence of iron-deficiency anemia has decreased substantially because of the provision of iron-containing infant formula and cereal to children enrolled in WIC. The utility of repeated measurements of hemoglobin concentrations while receiving WIC supplementation has not been addressed. This study shows that many children either develop anemia or remain anemic while receiving WIC benefits. This highlights the need for mandatory follow-up by WIC.

Corresponding author: Robert Listernick, MD, Division of General Academic Pediatrics, Children's Memorial Hospital, Box 16, 2300 Children's Plaza, Chicago, IL 60614 (e-mail: boblist@northwestern.edu).