Anemia is estimated to affect 2.2 billion individuals worldwide, half of whom are estimated to have iron deficiency (WHO 1991). In most areas of the world, iron deficiency affects primarily infants and young children because of a higher iron requirement related to growth, and women of childbearing age as a result of menstrual loss and pregnancy. The effects of anemia include "retardation of physical and mental development, fatigue and low productivity at work, and ... impairment of reproductive functions" (WHO 1991). Factors that lead to iron deficiency anemia include inadequate bioavailable iron in the diet and poor absorption of iron due to the presence of inhibitors of iron absorption such as tannin in tea and phytates from plants (Scrimshaw 1991). In a 1990 nutrition survey among Palestinian children 6-35.9 mo of age living in refugee camps under the auspices of the United Nations Relief and Works Agency for Palestine Refugees in the Near East (UNRWA),⁴ the overall prevalence of anemia was 67% [95% confidence interval (CI): 66, 68]. Other surveys performed in the Middle East have estimated the prevalence of anemia in young children to range from 19 to 71% (Abdelnour 1991, Gofin et al. 1992, Hossain et al. 1995, Kocak et al. 1995, Lavon et al. 1985, Madanat et al. 1984). UNRWA provides health care for ~2.2 million refugees in the following five geographical Fields of Operation (hereafter referred to as "Fields"): Syria, Jordan, the West Bank, the Gaza Strip and Lebanon (UNRWA 1990). The purpose of this report was to determine factors associated with an increased probability of a child being anemic.

SUBJECTS AND METHODS

In each camp, a sample of 100-120 children was systematically selected from birth and immunization records from each of three age groups: 6-11.9, 12-23.9 and 24-35.9 mo. Selected children were requested to attend the clinic on a specific day. Data collected on children included an interview with the adult (usually the mother) on demographic, health and behavioral factors; weight measured with a beam balance infant scale (SECA, Columbia, MD) calibrated daily with children wearing only one layer of light clothing; recumbent length using a length board (Perspective Enterprise, Kalamazoo, MI); for children 24 mo of age, standing height measured using a tape fixed to a wall; and hemoglobin (Hb) determined with the use of a portable hemoglobin photometer (HemoCue AB, Helsingborg, Sweden) based on an azidimethemoglobin method with the accuracy verified daily by using a control sample. The HemoCue method has been shown to be comparable in both accuracy and precision with the standard cyanmethemoglobin method (Bridges et al. 1987, Johns and Lewis 1989).

Growth indices were calculated by using the international growth reference (WHO Working Group 1986) and the Epi Info software program (Sullivan et al. 1994) for height-for-age (HA), weight-for-age and weight-for-height Z-scores. For children < 24 mo of age, recumbent length was used in the anthropometric calculations; for children 2 y, standing heights were used. Low anthropometry was defined as a Z-score < 2 SD. Anemia was defined as a hemoglobin <110 g/L.

Members of a supervisory team visited camps each day during data collection to ensure that all equipment was calibrated correctly and to verify weight and height measurements on a sample of children. A clinic supervisor reviewed each questionnaire for accuracy. Data were entered into a computer each day, and forms with errors or unlikely values were returned to the clinic for corrections, in some cases resulting in children being brought back to the clinic and remeasured. Data were collected between May 6, 1990 and June 30, 1990.

Variables studied as potential factors associated with anemia were family size, tea consumption, consumption of cow's milk or powdered milk, breast-feeding, birth weight, mother's and father's education level, sex, history of a current or recent (within previous 2 wk) diarrheal, febrile or respiratory illness, birth weight, current weight and height. Potential risk factors for anemia were analyzed separately for each age group and Field. First, each of these factors as investigated using Epi Info (Dean et al. 1994). On the basis of preliminary analysis of questions relating to recent (within the previous 2 wk) or current illnesses, data were categorized into three mutually exclusive groups: children with both diarrhea and fever, children with diarrhea or fever (but not both) and children with neither diarrhea nor fever. Breast-feeding was defined in the following two ways: in the two younger age groups, breast-feeding status was defined as currently breast-feeding, breast-fed before but not currently, and never breast-fed. In the oldest age group, breast-feeding was categorized as previously breast-fed vs. never breast-fed.

For the frequency of variables and their confidence intervals, within each Field, each camp was treated as a stratum and each individual assigned a weight based on the number of individuals registered within the camp in their age group (i.e., 6-11.9 mo, 12-23.9 mo and 24-35.9 mo) divided by the number of individuals selected for the survey in the same age group. Data were analyzed using the Csample module of Epi Info Version 6.0 (Dean et al. 1994). On the basis of results of stratified analyses, important variables were analyzed with logistic regression using the Egret software program (Egret 1993). A backward elimination modeling method was used with all important two-way interactions initially in the model. First, interaction terms were removed one at a time based on their P-value; the least significant term was removed. Two-way interactions were kept in the model if they had a P-values 0.05. Once all nonsignificant interaction terms were removed, main effects not included in interaction terms were removed if they were nonsignificant (P-value 0.05) and not considered an important confounder.

RESULTS

Table 1. Number of children surveyed, percentage distribution and 95% confidence intervals of various factors by Field, children 6-35.9 mo of age, UNRWA Refugee Camps, 19901 [View Table]

The prevalence of anemia by the various potential risk factors is shown in Table 2. Children with the following characteristics tended to have a higher prevalence of anemia: those living in Syria, Lebanon and Gaza, younger age (6-23.9 mo), never breast-fed, male, having an illiterate mother, fever and/or diarrhea and low height-for-age.

Table 2. Prevalence of anemia with 95% confidence intervals by potential risk factors and field, children 6-35.9 mo of age, UNRWA Refugee Camps, 1990 [View Table]

The final logistic regression models by age are shown in Table 3. In the youngest age group, Field, breast-feeding history, stunting and illness history were associated with the prevalence of anemia, with the last two variables interacting. The interpretation of the interaction term is that among children of normal stature, having a febrile and/or diarrheal illness was not associated with anemia, but among stunted children, having one or both illnesses was significantly associated with anemia. For the 12- to 23.9-mo-old children, Field, breast-feeding status, sex and illness history were significantly associated with anemia. In this age group, never having been breast-fed, being male and having a fever and/or diarrhea were associated with a higher prevalence of anemia. Among the oldest children (25-35.9 mo), significant factors were Field, male sex, having an illiterate mother, stunting and having a febrile and/or diarrheal illness, with the last two interacting.

Table 3. Factors associated with the prevalence of anemia among Palestinian refugee children by age group, prevalence odds ratios (POR) and 95% confidence intervals (CI) from logistic regression models1 [View Table]

DISCUSSION

A number of studies have demonstrated that anemia represents a more severe form of iron deficiency, whereas milder forms have abnormal iron biochemistry values such as low serum ferritin and low transferrin saturation. It is estimated that using anemia as an indictor, the prevalence of iron deficiency would be twice the prevalence of anemia. In the Palestinian refugee camps, two thirds of the children had anemia, indicating that it is likely that most children, if not all, had some degree of iron deficiency.

Late infancy and early childhood are high risk periods for iron deficiency because of an increased iron requirement related to rapid growth and diets that are relatively low in iron content. Based on the discussions with UNWRA personnel and mothers of some study children by the investigators, several infant feeding practices among Palestinian refugees are likely to increase the risk of iron deficiency: These include breast-feeding after 6 months of age without iron supplementation, use of non-iron fortified cow's milk as weaning food, early and frequent consumption of tea, and relatively infrequent consumption of meat due to its high cost. The overall higher prevalence of anemia for children in Syria, Lebanon and Gaza is likely related to the poorer socioeconomic status compared with Jordan and the West Bank. Syria had the highest prevalence of stunting in children and the mothers are more likely to be illiterate.

Information from other studies suggests that exclusive breast-feeding is protective against iron deficiency for infants <6 mo of age, after which the iron concentration in breast milk may not meet the infant's iron requirement (Dallman et al. 1980). This is compounded by weaning foods that, in many settings, are usually low in bioavailable iron. As described in previous studies (Reeves et al. 1984, Yip et al. 1987), the presence of childhood illnesses is associated with a higher prevalence of anemia. In some children, the anemia was likely due to the combined effect of iron deficiency and recent acute illness. Although daily tea consumption was not found to be significant in this study, children consuming tea on a daily basis did tend to have a higher prevalence of anemia. The lack of significance of tea consumption may be due to the high frequency of other causes of anemia in this population. In addition, most (90%) of the children drank tea either daily or on an occasional basis. Comparing children who drank tea on a daily basis with those who did not drink tea, the prevalence odds ratio (POR) (and 95% CI), controlling for other factors in the models in Table 3, was 1.1 (0.9, 1.5) for children 6-11.9 mo, 1.2 (0.8, 1.7) for those 12-23.9 mo, and 1.7 (0.9, 3.5) for children 24-35.9 mo. This finding of a higher prevalence of anemia among tea drinkers is consistent with other reports on the effect of tea on iron status (Disler et al. 1975, Galan et al. 1985, Razagui et al. 1991, Rossander et al. 1979), but was not found to be significant in this population.

The results of this survey may not be representative of all Palestinian refugee children because of the nonrandom selection of camps in three Fields and the selection of camps near urban areas in the other two Fields. Weight-for-height information was collected but because <2% of the children were < 2 SD (Yip et al. 1990), there were too few "wasted" children in some subgroups. Birth weight information was collected, but in some camps the data were based on birth records, whereas in others they were based on maternal recall. Maternal recall tended to provide birth weight information rounded to the nearest 0.5 kg. Therefore, the birth weight information in this survey from some refugee camps was of questionable quality. Although male sex was associated with anemia in the older two age groups, the prevalence was only 2-3% higher than females. Factors associated with severe anemia (<90 g/L) were also investigated (Hassan 1992). The overall prevalence of severe anemia was 16% (95% CI: 14, 17); the data were too sparse in some subgroups to perform analyses similar to those described for anemia (< 110 g/L).

Recommendations to improve the iron status of children include the following: promotion of appropriate breast-feeding and weaning practices; iron fortification of commonly eaten foods; promotion of foods with high iron content, especially meat products, and foods with a high vitamin C content (which improves iron absorption); reduction of tea consumption among children; and consideration of oral iron supplementation for children 6-12 mo of age if dietary improvement cannot be instituted.

ACKNOWLEDGMENTS