lence of iron deficiency anemia in infants and preschool-age children declined, on the basis of prevalence data collected under the CDC Pediatric Surveillance System, from 7 to 3 percent (Yip et al., 1987a,b). Preliminary data from NHANES III collected between 1988 and 1991 appear to confirm this trend (preliminary analysis of data on hemoglobin and serum ferritin concentration measures only), with a prevalence at or below 3 percent for both black and white children 1 to 5 years of age; the prevalence among Mexican American infants, but not young children, may be somewhat higher (A.C. Looker, National Center for Health Statistics, personal communication, June 1993).

For women of childbearing age, data do not show a similar drop in the prevalence of iron deficiency anemia. NHANES II and preliminary NHANES III data show that 4 to 10 percent of U.S. women of childbearing age have iron deficiency anemia on the basis of two or three abnormal values for the surveyed indicators of iron status (see above). The estimated prevalence is lower in non-Hispanic white women than in black women and women in one Hispanic sub-group. Women between 15 and 19 years of age have a prevalence of iron deficiency anemia similar to that of women between 20 and 44 years of age. For women between 20 and 44 years of age, a higher prevalence of iron deficiency anemia is associated with poverty, low educational attainment, and high parity (LSRO, 1984).

No national population survey data on iron deficiency anemia are available for pregnant women. However, data on low-income women are available from the CDC Pregnancy Nutrition Surveillance System, and the national WIC evaluation. The 1990 CDC survey reported prevalences of iron deficiency anemia of 10, 14, and 33 percent in the first, second, and third trimesters of pregnancy, respectively, for all pregnant women of all races (Kim et al., 1992). Black women exhibited a significantly higher prevalence of iron deficiency anemia than did women of other races. CDC data show that the prevalence of iron deficiency anemia in the low-income population has remained stable since 1979, a finding that the committee found particularly troubling.

Data from the 1985 National WIC Evaluation were consistent with most past studies—a significant negative relationship of initial hemoglobin with birth-weight (-23.6 g birth-weight per 1 g/dl increase in hemoglobin concentration [p<0.01] and a 0.96 percent increase in the rate of birth-weight under 2,500 g for each increase of 1 g/dl of initial hemoglobin [p<0.05]) (Rush et al., 1988). When these results were reexamined more closely for this report, there was no evidence of adverse relationship between a hemoglobin level under 10 g/dl or over 14 g/dl with adverse prenatal outcomes among whites. Among blacks, high hemoglobin (>14 g/dl) was associated with low-birth-weight in both first and second trimester participants, and low hemoglobin (<10 g/dl) was associated with low-birth-weight in first trimester participants, but numbers were small. For the much larger group of second trimester participants, there was no association of low hemoglobin with low-birth-weight (D. Rush, Tufts University, personal communication, June 1993).