PEDIATRICS Vol. 107 No. 4 April 2001, pp. 677-682

Continuation of the Decline in Prevalence of Anemia in Low-Income Infants and Children in Five States

Bettylou Sherry, PhD, RD^*, Zuguo Mei, MD, MPH^*, and Ray Yip, MD, MPH

From the ^* Maternal and Child Nutrition Branch, Division of Nutrition and Physical Activity, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, Georgia, and  Centers for Disease Control and Prevention, on assignment to UNICEF Beijing, Beijing, People's Republic of China.

	ABSTRACT

Top Abstract Methods Results Discussion References

Objective.  To examine whether there is a continuation of the decline in prevalence of anemia among low-income infants and children 6.0 to 59.9 months old from the early 1980s to the mid-1990s.

Study Design.  Cross-sectional trend analysis of data from the Centers for Disease Control and Prevention's Pediatric Nutrition Surveillance System from the 5 states (Colorado, New Mexico, Oklahoma, Utah, and Vermont) that have been using the same laboratory method for anemia screening since 1984 or earlier.

Results.  The overall prevalence of anemia decreased substantially in each state from the early 1980s to the mid-1990s as follows: Colorado by 52%; New Mexico by 75%; Oklahoma by 67%; Utah by 57%; and Vermont by 48%. In each state, the prevalence of anemia declined for children of different age groups, birth weights, genders, type of pediatric care visit (screening or follow-up), and most race/ethnic groups.

Conclusions.  The decline in the prevalence of anemia initially observed in the 1980s continued well into the 1990s. This decline is likely attributable to better iron nutrition related to greater usage of iron-fortified products and possibly better iron bioavailability in some of the food products.  Key words:  anemia, low-income children, breast feeding, iron-fortified formula, iron-fortified cereal.

From the 1960s through the 1970s, the prevalence of anemia was notably high, ranging from 15% to 30% among US infants and young children.^1-3 For the most part, Vazques-Seoane et al² and Miller et al³ attributed the anemia to iron deficiency and to the low iron content in the commonly consumed infant dietformula and cereal. On recognition of dietary factors such as the widespread use of whole cow's milk in infants' diets as contributors to iron-deficiency anemia, the American Academy of Pediatrics recommended that infants who were not breastfed be given iron-fortified formula and that low-iron formula or whole cow's milk be avoided.⁴ The Special Supplemental Nutrition Program for Women, Infants, and Children (WIC), which began in 1972 and provides supplemental foods to mothers, infants, and children of low-income families, adheres to the American Academy of Pediatrics recommendations by providing iron-fortified formula to infants. The WIC Program enrollment grew dramatically during the 1970s and 1980s. A steadily declining trend of anemia throughout the 1980s among both low-income and middle-class children^5-7 seemed to reflect the increased use of iron-fortified formula and cereal.⁸ By the early 1990s, the prevalence of iron deficiency anemia had reached a relatively low level, as has been shown by the third National Health and Nutrition Examination Survey (NHANES III).⁹ The NHANES III (conducted in 1988-1994) documented the prevalence of iron deficiency anemia as 3% for 1- to 2-year-olds and <1% for 3- to 5-year-olds.⁹

It is reasonable to expect among infants and children participating in the WIC program that the prevalence of anemia would stabilize after the initial years of rapid decline. As the WIC program increased its capacity, over time state programs have reached a steady state of participation or in the last few years, some programs have decreasing enrollment patterns likely attributable to welfare reform issues. However, a recent review of pediatric anemia among WIC participants in Vermont found that the prevalence continued to decline through 1994.¹⁰ The purpose of this study was to assess whether this observation was also true for WIC participants in other states.

	METHODS

Top Abstract Methods Results Discussion References

This trend analyses of cross-sectional data examines the prevalence of anemia from the early 1980s through the mid-1990s in low-income infants and children from 6 to 59.9 months old who participated in WIC and other publicly funded health and nutrition programs. The data come from the Pediatric Nutrition Surveillance System (PedNSS), which is managed by the Centers for Disease Control and Prevention (CDC). States routinely submit their clinic-based data on the anemia, anthropometric, and breastfeeding status of these infants and children to the PedNSS at CDC, which aggregates and summarizes the data to provide national, state, and local surveillance information. In 1995, 45 states, Indian reservations, and US territories contributed data to the PedNSS. Our study has been approved by the CDC Institutional Review Board for Human Subjects.

Most states used both hemoglobin and hematocrit testing to screen for anemia, and the proportion of those using the hemoglobin test has increased throughout the 1990s. Mixing results of the 2 screening tests and the proportion of the results derived from each test makes it difficult to measure the real trend of the prevalence of anemia. For this reason, we selected the states that have been using only the hematocrit screening method over the years to ensure comparability in the trend of prevalence from a single anemia screening method. This is particularly important because the common cutoff used for defining anemia based on hemoglobin testing yields a different estimate from the one based on hematocrit.¹¹

Colorado, New Mexico, Oklahoma, Utah, and Vermont met the following a priori criteria for inclusion: 90% of hematology data for anemia screening was based on the hematocrit assay, >5000 hematocrit records per year in the years of study (states submitting complete data for a year normally have >5000 records), consistency in the number of records per year (a difference of no more than ± 3 times the mean number of records per year in sequential years), and data submitted to the PedNSS each year from at least since 1984 through 1993 to 1995. Colorado had data for 1979 through 1995, New Mexico for 1983 through 1995, Oklahoma for 1983 through 1993, Utah for 1984 through 1995, and Vermont for 1981 through 1995. These states provided us with a range in poverty status. Data on the proportion of children <18 years old living in poverty in 1993 shows the following percentages: Utah 14%; Colorado and Vermont 17%, Oklahoma 25%, and New Mexico 31%, with the US national 23%.¹² Thus, we included 2 states with a high proportion of children in poverty, 2 medium-poverty states, and 1 low-poverty state. Data were not available for states with higher incidence of immigrants¹³ among whom the prevalence of anemia might be higher.

Although the equipment brand varied by state and over time, all of the states in our study used microhematocrit assays throughout the study period. All of the states followed training protocols for new employees and routinely scheduled quality assurance reviews by both state and clinic staff to check for the use of proper procedures.

The definition of a positive screening hematocrit test is based on the criteria for anemia recommended by the CDC: 32.9% for children 12.0 to 23.9 months old and  33% for children 24.0 to 59.9 months old.¹⁴ We used the same criteria for infants 6 to 12 months old as recommended for children 12.0 to 23.9 months old. The data have been adjusted for altitude as recommended by the CDC.¹⁴

For these analyses, we deleted records with missing hematocrit data. The mean (and range) of missing data for the study period was 23% (7%-35%) for Colorado, 5% (5%-11%) for New Mexico, 3% (1%-12%) for Oklahoma, 3% (<1%-13%) for Utah, and 6% (3%-10%) for Vermont. Colorado had a relatively high proportion of missing data of submitted data from routine clinic visits in addition to WIC certification/recertification examinations; these other routine clinic visits did not include anemia screening. The other 4 states submitted only WIC participant certification/recertification data. Another reason for the missing data in Colorado, New Mexico, and Vermont is they use skip patterns for monitoring hematocrit; one example would be if at the previous visit a child was not anemic, they would not redo for 1 year. Thus, little bias resulting from missing data existed in our analyses.

After we deleted the records with missing hematocrit data, we randomly selected 1 measured hematocrit value per child per year for analyses to prevent bias from an oversampling of children who were evaluated more than once a year. The annual prevalence of anemia was calculated by state and was stratified by age, gender, race or ethnicity, and birth weight status to identify potential confounding factors. Trends in prevalence by screening and follow-up visits are also presented. Because the data include all children enrolled in WIC, there is no sampling error, and significance tests are inappropriate. The 5 states have different populations, protocols, priorities served, and risk factors for enrollment criteria for WIC; thus, our results are presented by state rather than as an aggregated summary.

Two measures were used to examine the quality of the hematocrit data: the biologically plausible range of data, and the standard deviation of the hematocrit values. We set the biologically plausible range for hematocrit values at 24% to 51%, as this was high enough to exclude potential miscodes of hemoglobin data yet included values expected in normal healthy individuals. All of the states had <1% of their data outside of the biologically plausible range. The standard deviations of the hematocrit values were examined after we deleted the biologically implausible values. We used 2.3% to 3.3% as the acceptable range within which we would expect to find the values.¹⁵ This range was based on the hematocrit standard deviation of the second NHANES, which is 2.3%,¹⁶ and the hemoglobin median standard deviation found in our PedNSS data for 1993, which is 1.1 g/dL (×3 for the hematocrit equivalent = 3.3%).¹⁵ In 4 states, the range of hematocrit standard deviations by state each year was within our acceptable range; in Vermont, the range was lower than expected (2.15%-2.45%). This low range may reflect the racial and ethnic homogeneity (ie, predominately white) of the state's population.¹⁷

	RESULTS

Top Abstract Methods Results Discussion References

In all 5 states, the number of records submitted to PedNSS increased over time (Table 1). New Mexico had a large change in age composition of program participants over time: initially 95% of their participating children were 6.0 to 23.9 months old, and by 1995 37% were in this age category. In 1989, New Mexico increased service to include all older children that had medical, anthropometry, and hematology risk factors (oral communication, J. Peacock, MS, RD, WIC Director, New Mexico Department of Health, Santa Fe, New Mexico, July 1996). All states consistently had an approximately equal gender distribution of program participants. Colorado and New Mexico had a much higher proportion of Hispanics than the other states did, and Vermont had the highest percentage of white participants (~98%).

Mean hematocrit values increased in all age categories in all states (except children 36.0-47.9 months old in New Mexico) (Table 2). These increases in hematocrit values reflect decreases in the prevalence of anemia: in each state, we found that prevalence of anemia decreased not only for each age group and with increasing age, but also for each gender, for each race or ethnicity (with 1 exception), and for each birth weight class (Table 3). The prevalence of anemia for both screening visits and follow-up visits declined over the years of study, and a higher prevalence of anemia was found during the screening visits. Overall, the relative decline in prevalence of anemia in each state ranged from 42% (Vermont) to 81% (New Mexico).

Through the mid-1990s, New Mexico, Oklahoma, and Utah had large and relatively consistent decreases in the prevalence of anemia (Fig 1). Colorado and Vermont, which had lower initial levels of initial prevalence, exhibited more modest declines in prevalence.

View larger version (15K): [in this window] [in a new window]   Fig. 1.   Prevalence of anemia among low-income infants and children in 5 states participating in the CDC PedNSS.

We examined racial and ethnic differences in the prevalence of anemia in 4 states (Vermont was excluded, because of its racial homogeneity). We noticed that in Colorado, New Mexico, and Oklahoma, blacks had the lowest or second-lowest prevalence of anemia during the mid-1980s, but had the highest prevalence relative to other groups by the mid-1990s (Fig 2) and in Colorado and Oklahoma had an increase in prevalence in the early 1990s, despite decreases in all other groups. We also noted that in Utah, Native Americans had the highest prevalence of anemia in all study years except 1993. 

View larger version (19K): [in this window] [in a new window]   Fig. 2.   Prevalence of anemia by race or ethnicity among low-income infants and children in 4 states participating in the CDC PedNSS.

	DISCUSSION

Top Abstract Methods Results Discussion References

Data from low-income infants and children participating in WIC and other publicly funded health and nutrition programs in 5 states document a large and relatively consistent decline in the prevalence of childhood anemia from the early 1980s through the mid-1990s. Our findings are consistent with previous research on both low-income and middle-class children^5,6,18 during the 1980s and the single analysis that examined this trend in prevalence through 1994 in Vermont.¹⁰ Although the literature cited used varied cutoffs for defining anemia, the decreasing trend in prevalence and the relatively low prevalence of anemia documented by NHANES III⁹ still are relevant to our findings. The large size of the data set used for our analysis and the use of a single screening tool lend credence to our finding that a decline in prevalence in childhood anemia has continued through the mid-1990s and that the phenomenon may be widespread.

Although our study that is based on PedNSS data does not include testing specific for iron deficiency (eg, serum ferritin concentration), improvement of iron status can, to a large extent, be attributed to increased use of iron-fortified formula in the United States and the decrease in use of cow's milk.^8,18 In the Ross Laboratories Mother's Survey, a large national mail survey of mothers who have recently given birth, the data showed that among the respondents who gave their infants any kind of formula at birth, the proportion of infants who were fed iron-fortified formula increased from nearly 70% in 1988% to 91% in 1998 (oral communication, A.S. Ryan, PhD, Ross Laboratories, Market Research Information, Cleveland, Ohio, October 1999). It is also possible that a slight increase in the incidence of breastfeeding¹⁹ may have contributed to the decline in the prevalence of anemia because of the relatively high absorption of iron in breast milk.^20,21

Although the iron content of the formula has remained the same (12 mg/L) over the years, there are evidence that the bioavailability or absorption of added iron has improved. This is suggested in a recent study by Walter et al²² that found that noniron-fortified formula (2 g/L) is protective against iron deficiency anemia. In addition, Walter et al²³ demonstrated that, at 8 months of age, the prevalence of iron deficiency anemia was significantly lower in infants fed iron-fortified cereal than in infants fed noniron-fortified cereal. This study was the first to show that iron-fortified cereal was effective in preventing iron deficiency anemia. The iron used for infant cereal has become better absorbed as the particle size of the elemental iron used has become smaller and has been shown to result in better absorption.^24,25 The increased use of iron-fortified cereals would impact anemia in infants as well as in the older age groups studied.

Additional evidence to support the increased use of iron-fortified products as a contributing factor to the decrease in the prevalence of anemia is the increased participation in the WIC program during the 1970s and 1980s. States now serve more priority levels, which means that children with risk factors other than anemia, inadequate growth, or health factors are included, which in turn effectively reduces the proportion of children who are anemic when they enter the WIC program. The high iron content of the WIC foods can also prevent anemia among those who might be at risk. Data from the NHANES II and III suggest that this may be the case. Comparisons between NHANES II and III show basically no change in the prevalence of iron deficiency based on the MCV model in 1- to 2-year-olds and 3- to 4-year-olds.²⁶ However, although the low-income children in both surveys have a higher prevalence of anemia than the overall prevalence for their age groups, the difference is much less in NHANES III than it was in NHANES II.²⁶ WIC, because it is such a major program and it provides iron-rich foods, is one logical potential contributor to lowering the prevalence of iron deficiency in low-income children. Vermont has always had the lowest prevalence of anemia and they have been able to serve all priorities with the exception of a short period in part of 1991 and 1992. Other states varied in what priorities they served; all increased with the exception of Oklahoma, which still serves only priorities 1 to 3.

One limitation of our study is that we were only able to include 5 states because they were the only ones that had continuously used the hematocrit assay. It is possible that our findings may not be applicable to states with very different demographics, such as those with a high proportion of immigrants. Another limitation is that data were not available to demonstrate that the anemia was caused by iron deficiency, yet the changes in prevalence of anemia over time are likely to reflect improvement in iron nutritional status. However, we cannot rule out the impact of other causes of anemia or other changes.

It seems that the increased use of iron-fortified formula and cereal have contributed to better iron nutrition status. These have added more iron to the infant's and older child's diet. This encouraging trend has brought the prevalence of anemia to a relatively low levelnear or below 5% in all 5 states that we studied. At this level, the majority of anemia cases are not related to iron deficiency or other pathologic conditions. If we use the 5th percentile of a healthy sample to define the cutoff for anemia, we would expect, just by definition, 5% of the population to be classified as anemic. Under such conditions of low prevalence, the indication of routine screening for anemia to detect those children with higher risk for iron deficiency is no longer useful.¹⁴ For this reason, routine screening for anemia by WIC and other publicly funded health and nutrition program deserves reassessment. If, indeed, a given population is found to be free from iron deficiency anemia, the current standard practice of routine screening for iron deficiency using the hematocrit or hemoglobin test may no longer be necessary.

The US success story of the WIC program in preventing iron deficiency among low-income infants and children serves as a good case example of primary prevention through pediatric nutrition. As evidence shows that the prevalence of childhood anemia is continuing to decrease, if a population is found to be free of iron deficiency anemia, current screening practices may indeed, warrant reassessment. The PedNSS data at the state level could be used to identify states, counties, or clinics with a low prevalence of anemia and modification of screening practices could be considered to target those at higher risk for anemia.

	ACKNOWLEDGMENTS

We thank the staff of the Department of Health for the 5 states included in these analyses, who helped make our documentation of their state activities more complete. We also thank Laurence Grummer-Strawn, PhD, for his thoughtful review of this manuscript, and Ellen Borland for her diligent preparation of the data sets.

	FOOTNOTES

Received for publication Jan 11, 2000; accepted Aug 28, 2000.

Reprint requests to (B.S.) Maternal and Child Nutrition Branch, Division of Nutrition and Physical Activity, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Mail Stop K-25, 4770 Buford Hwy, NE, Atlanta, GA 30341-3724. E-mail: bls6{at}cdc.gov

	ABBREVIATIONS

WIC, Special Supplemental Nutrition Program for Women, Infants, and Children; NHANES, National Health and Nutrition Examination Survey; PedNSS, Pediatric Nutrition Surveillance System; CDC, Centers for Disease Control and Prevention.

	REFERENCES

Top Abstract Methods Results Discussion References

1. US Department of Health, Education, and Welfare. Ten State Nutrition Survey, 1968-1970, Volumes I-IV. Washington, DC: US Government Printing Office; 1972. DHEW Publ. Nos. (HSM) 72-8130 to 72-8132
2. Vazquez-Seoane P, Windom R, Pearson HA Disappearance of iron deficiency anemia in a high risk infant population given supplemental iron. N Engl J Med 1985; 313:1239-1240 [Medline]
3. Miller V, Swaney S, Deinard AS Impact of the WIC program on the iron status of infants. Pediatrics 1985; 75:100-105 [Abstract/Free Full Text]
4. Committee on Nutrition, American Academy of Pediatrics Iron fortified formula. Pediatrics 1971; 47:786 [Abstract/Free Full Text]
5. Yip R, Binkin NJ, Fleshood L, Trowbridge FL Declining prevalence of anemia among low-income children in the United States. JAMA 1987; 258:1619-1623 [Abstract/Free Full Text]
6. Yip R, Walsh KM, Goldfarb MG, Binkin NJ Declining prevalence of anemia in childhood in a middle-class setting: a pediatric success story? Pediatrics 1987; 80:330-334 [Abstract/Free Full Text]
7. Centers for Disease Control and Prevention. Pediatric Nutrition Surveillance System 1997 Full Report: Atlanta, GA: US Department of Health and Human Services, Centers for Disease Control and Prevention; 1998
8. Ryan AS, Martinez GA, Yip R. Changing patterns of infants feeding in the United States: evidence to support improved iron nutritional status in childhood. In: Hercberg S, Galon P, Dupin H, eds. Recent Knowledge on Iron and Folate Deficiencies in the World: International Meeting, Paris, July 11-13, 1989, Vol. 197. Paris, France: Colloque INSERM; 1990:631-640
9. Looker AC, Dallman PR, Carroll MD, Gunter EW, Johnson CL Prevalence of iron deficiency anemia in the United States. JAMA 1997; 277:973-976 [Abstract/Free Full Text]
10. Sherry B, Bister D, Yip R Continuation of decline in prevalence of anemia in low-income children: the Vermont experience. Arch Pediatr Adolesc Med 1997; 151:928-930 [Abstract/Free Full Text]
11. Dallman PR, Simes MA, Stekel A Iron deficiency in infancy and childhood. Am J Clin Nutr 1980; 33:86-118 [Free Full Text]
12. Children's Defense Fund. Children in the states: 1998 data. Available at: http://www.childrensdefense.org/states.html. Accessed July 13, 2000
13. Bureau of Census. State population estimates and demographic components of population change: April 1, 1990 to July 1, 1990. Available at http://www.census.gov/population/estimates/stat/st-99-2.txt. Accessed July 13, 2000
14. Centers for Disease Control and Prevention. Recommendations to Prevent and Control Iron Deficiency in the United States. MMWR Morb Mortal Wkly Rep. 1998;47(No. RR-3)
15. Centers for Disease Control and Prevention. Enhanced Pediatric Nutrition Surveillance System: User's Manual. 1996 Revision. Atlanta, GA: Centers for Disease Control and Prevention; 1996
16. Fulwood R, Johnson CL, Bryner JD, et al. Hematological and Nutritional Biochemistry Reference Data For Persons 6 Months-75 Years of Age: United States, 1976-80;Vital and Health Statistics. Series 11. No 232. Hyattsville, MD: National Center for Health Statistics; 1982. DHHS Publication No. (PHS) 83-1682
17. Life Sciences Research Office. Assessment of the Iron Nutritional Status of the US Population Based on Data Collected in the Second National Health and Nutrition Examination Survey, 1976-1980. Bethesda, MD: Federation of American Societies for Experimental Biology;1984
18. Zeigler EE, Fomon SJ Strategies for the prevention of iron deficiency: iron in infant formulas and baby foods. Nutr Rev 1996; 54:348-354 [Medline]
19. Ryan AS. The resurgence of breast-feeding in the United States. Pediatrics. 1997;99(4). URL: http://www.pediatrics.org/cgi/content/full/99/4/e12
20. McMillan JA, Oski FA, Lourie G, Tomarelli RM, Landaw SA Iron absorption from human milk, simulated human milk, and proprietary formulas. Pediatrics 1977; 60:896-900 [Abstract/Free Full Text]
21. McMillan JA, Landaw SA, Oski FA Iron sufficiency in breast-fed infants and the availability of iron from human milk. Pediatrics 1976; 58:686-691 [Abstract/Free Full Text]
22. Walter T, Pino P, Pizarro F, Lozoff B Prevention of iron-deficiency anemia: comparison of high- and low-iron formulas in term healthy infants after six months of life. J Pediatr 1998; 132:635-640 [CrossRef][Medline]
23. Walter T, Dallman PR, Pizarro F, Effectiveness of iron-fortified cereal in prevention of iron deficiency anemia. Pediatrics 1993; 91:976-982 [Abstract/Free Full Text]
24. Rios E, Hunter RE, Cook JD, Smith NJ, Finch CA The absorption of iron as supplements in infant cereal and infant formulas. Pediatrics 1975; 55:686-693 [Abstract/Free Full Text]
25. Fomon SJ Bioavailability of supplemental iron in commercially prepared dry infant cereals. J Pediatrics 1987; 110:660-661 [CrossRef][Medline]
26. National Center for Health Statistics. Healthy People 2000 Review, 1997. Hyattsville, MD: Public Health Service; 1997