Contents - Previous - Next

Functional significance of the covariance between protein energy malnutrition and iron deficiency anemia¹

Epidemiological characteristics
Iron and development
Weaknesses in design
Literature cited

ERNESTO POLLITT²

¹Prepared for the International Dietary Energy Consultative Group (IDECG) Task Force workshop on malnutrition and behavior at the University of California, Davis, CA, December 6-10, 1993. This workshop was supported by IDECG, the Nestle Foundation, Kraft Foods and the International Union for Nutritional Science. Guest editor for this supplement was Ernesto Pollitt, Department f Pediatrics, University of California, Davis, CA, 95616

²To whom correspondence should be addressed: Department of Pediatrics, School of Medicine, Program in International Nutrition, University of California, Davis, CA 95616.

Department or Pediatrics, School of Medicine, Program in International Nutrition,
University of California, Davis, CA 95616

In 1968 Derrick B. Jelliffe published a book on Child Nutrition in Developing Countries containing the following description of protein-calorie malnutrition:

The term PCM is, in fact, a collective term and refers to a variety of different clinical forms of malnutrition. These can be best visualized in the form of a triangle, which is intended to show that there is a gradation between the normal, healthy, well-fed child at the apex of the triangle and severe PCM at its base. In between, mild and moderate degrees occur. Two types of severe PCM - nutritional marasmus and kwashiorkor - can be easily recognized. The term PCM is used for this group of conditions because all Of them are due to a diet low in protein but with different levels of intake of carbohydrate Calorie. (p. 75; italics in original.)

This statement on the etiology of protein-energy malnutrition³ (PEM) guided research on the effects of PEM on cognitive development when the observational and experimental studies that: contributed most to the data currently available were launched. Severe PEM was characterized by a well-defined set of clinical signs and symptoms for marasmus, kwashiorkor or a combination of both, and mild-to-moderate malnutrition was primarily defined by physical growth retardation (Alleyne et al. 1977). The PEM was recognized not as a distinct clinical entity, but as a syndrome that generally included micronutrient deficits. Little attention was given to the role that these deficits had in the growth and developmental delays observed among malnourished children (see, for example, Scrimshaw and Gordon 1968).

³Over the years the term protein-energy malnutrition replaced the term protein-calorie malnutrition, which was originally coined by Derrick B. Jelliffe.

New information from field studies in the developing world shows, however, that at least some populations previously considered at risk of mild-to-moderate PEM, particularly because of the high prevalence of linear growth retardation, have adequate intakes of energy and protein. An illustration of a discrepancy between perception and reality is found in the CRSP study of economically impoverished communities in three countries (Egypt, Kenya, Mexico) that were originally selected as targets for a study on the functional effects of limited energy intake (Calloway et al. 1992). The energy intake of the children and adults in the target communities of Egypt and Mexico exceeded requirements (Allen 1993, Beaton et al. 1992). In Kenya (but not in the other two countries), energy intake covaried positively with social and economic status, which suggested that intake was not constrained by resources in the other two countries (Allen 1993). Further, the protein intake of toddlers in

Egypt was adequate. Only in Kenya was there any evidence of protein deficiency; in those cases where energy intake was sufficient, protein intake was also sufficient to meet requirements. These data failed to support any hypothesis that the growth and development of the children, at least in the communities in Mexico and Egypt, are influenced by inadequate protein and energy intake (Beaton et al. 1992). Therefore, the data could not answer the question that originated the study, namely, the functional effects of deficits in energy intake.

On the other hand, the intakes of some micronutrients were below requirements in many children of the communities in the three countries. For instance, 35% of toddlers in Egypt and 43% in Mexico had iron intakes that were insufficient to prevent anemia (Murphy et al. 1992). In Kenya 13% of the toddlers had low iron intakes. Further, 57% of the children in Kenya and 25% in Mexico had inadequate zinc intake, whereas in Egypt this was true for only 10% of the children.

No claim is made here that the intakes of protein and calories are above international recommendations in all societies where the prevalence of PEM has been estimated to be high or where, on average, children are much smaller than international reference standards. Ample documentation exists, for example, on the limited intake of weanling children in different parts of the developing world (see below). At issue here is that a few years ago it seemed reasonable to select the communities chosen for the CRSP study and similar communities to study the relationships between PEM and development.

In a related area of research, new studies have been published on the developmental effects of iron deficiency anemia (IDA). As discussed below, IDA increases the probability of deviations in the trajectory of motor and mental development of children from a normal developmental path. Data on toddlers, preschoolers and school-age children collected in double blind, randomized trials show that iron deficiency is associated with developmental delays and that iron supplementation improves performance of subjects with this nutrient deficit (Pollitt and Metallinos-Katsaras 1990). Also, as reported by Golub et al. (1995) in this supplement, there is suggestive evidence that zinc deficiency could also alter cognitive development. Accordingly, it is justified to raise the question of whether micronutrients, in general, and iron and zinc, in particular, acted as confounders in the studies on the functional consequences of PEM.

In this paper the focus is on iron and the functional significance of the association between PEM and iron deficiency anemia. In particular, the argument is presented that a disregard for such an association is no longer warranted in attempting to look at the effects of any one of these two conditions on human development. Three sources of data are reviewed: similarities in epidemiological characteristics, effects of iron on development and weaknesses in the designs of studies on PEM.

Epidemiological characteristics

This section briefly reviews similarities in causation and in developmental periods of increased vulnerability. In each of these subsections evidence is presented that speaks to the covariance between PEM and iron deficiency anemia.

Causation. Energy and protein intakes that are below requirements, as well as infections, are commonly the most important proximal determinants of PEM. In severely impoverished communities in developing countries with high prevalence of growth retardation, the dietary intakes of young children are often characterized by an extremely low consumption of meat, poultry and fish, particularly among weanling infants and toddlers. In Bangladesh breast milk was the main source of nutrients up to 30 mo of age, with cereals as a second source (Brown et al. 1982). Few children consumed meat and eggs, even among those older than 24 mot The total intakes of energy and protein were below international recommendations. Ingestion of energy was considered particularly limited and judged to be barely adequate for resting metabolism in some children.

Primary causes of IDA in infants and toddlers are a limited intake of heme iron and a high intake of dietary factors that inhibit nonheme iron absorption (for example, phytates, tea and coffee.) The intake of products (for example, meat, fish and poultry) that represent the main sources of heme iron in human consumption is often negligible, and the main source of iron among infants and toddlers is breast milk (see, for example, Brown et al. 1982). However, as noted below, although iron from breast milk is highly absorbable, it is insufficient to prevent anemia after the first 6 mo of life (Cohen et al. 1993).

Developmental risk. One developmental period of high risk for an inadequate intake of macro- and micronutrients is at ~ 4-6 mo of age because some nutritional properties of human milk are no longer sufficient to meet physiological needs, and weaning foods are introduced that are vectors for infection. After the first 4 mo of life the protein intake of breast-fed infants generally falls below requirements (Fomon 1991, Heinig et al. 1993, Waterlow and Thompson 1979); however, the uptake of breast milk protein may be used more efficiently (Butte et al. 1992). Further, because of poor sanitation and hygienic practices, the introduction of other foods increases the probabilities of infection at weaning. For example, during the first 6 mo of life in Peru, the incidence rate of diarrhea was higher among infants who were no longer breast feeding than for those who were exclusively or partially breast fed (Brown et al. 1989).

Iron stores are saturated during the first 3 mo of life, but they are often depleted by 4-6 mo as demands for iron to prevent anemia increase (Duncan et al. 1985). The Recommended Guidelines for the Prevention of Iron Deficiency Anemia of the US Food and Nutrition Board (Earle and Woteki 1993) specify that breast-fed infants should receive a source of iron (iron-fortified infant cereal, meat or supplemental iron at 1 mg ~ kg^-1. d^-1) beginning at 4 mo of age. Iron requirements for 4- to 12-moold infants (120 Dg kg^-1 • d^-1) are almost double those of toddlers aged 13-24 mo (56 Dg-kg^-1-d^-l). Total body iron does not change appreciably in an average infant during the first 4 mo of life. However, it changes from ~ 250 g in a 6-kg baby to ~ 420 g in a 10-kg baby (Dallman 1993). Although, as noted, the iron in breast milk is highly bioavailable and meets physiological needs during the first 4-6 mo of life, it may not suffice after that.

In Honduras, the number of anemic 6-mo-old infants who were exclusively breast-fed was higher than those who had received complementary foods beginning at 4 mo of age (Cohen et al. 1993).

Iron and development

This section focuses on justifying the claim that iron could, in fact, alter cognition in ways similar to the alterations that have been attributed to PEM. This justification is found in studies on the role of iron in the brain and in studies on the developmental delays of iron-deficient children. Iron is found throughout the different regions of the brain, and its highest concentrations are found in the globus pallidus, red nucleus, caudate nucleus, putamen and substantia nigra. Minor concentrations exist in the cortex and cerebellum (Hill 1989, Smith 1990). Of importance is that the uptake of iron into the brain is at its peak during periods of fast neuronal growth (Jacobson 1963) and that iron deficiency anemia in rodents alters the process of myelination (Morris et al. 1992). Moreover, intracellular iron is involved in the synthesis, uptake and degradation of neurotransmitters known to affect information processing (for review see Beard et al. l 993). It has been proposed that the alterations in dopamine and y-amino-butyric acid (GABA) receptors that follow iron depletion mediate the neurodevelopmental changes that have been observed (Youdim et al. 1989).

Correlational data show a striking consistency among studies in the association between IDA and comparatively poor performance in mental and motor development scales among infants and toddlers. Intervention studies are classified into two groups. One includes trials on the effects of oral or intramuscular iron 7-10 d after the intervention, testing the hypothesis that the depletion of cellular iron in the brain alters mental and motor test performance. The other includes studies on the effects of the supplement up to 4 mo after the initiation of the treatment. Because the hemoglobin of infants originally diagnosed as IDA generally falls within the normal limits after 8 wk of iron intervention, this second group of studies is not able to address whether the effects of iron are mediated by mechanisms related to changes in hematology or cerebral iron.

Studies focusing on effects of iron supplementation over a 2- to 4-mo period are categorized into those that followed a strict experimental design and those that used quasi-experimental designs (which did not include anemic subjects exposed to a placebo condition). Two double blind, randomized clinical trials that tested the long-term developmental effects of iron therapy on IDA infants found evidence consistent with the hypothesis that iron deficiency anemia causes delays in mental and motor development (Aukett et al. 1986, Idjradinata and Pollitt 1993). For example, in one study, 12- to 18-mo-old IDA infants exposed to a 4-mo iron oral intervention had, on average, a 20-point incremental change in their Bayley-MDI from a pre- to a post-treatment evaluation. Conversely, the MDI increment in the IDA infants who received placebo was negligible (Idjradinata and Pollitt 1993). The changes in motor development were even greater than those in mental development among anemies treated with iron.

In the studies lacking a control anemic group, iron repletion therapy for 12 wk did not improve the mental development score of the anemic subjects (Pollitt 1993). This failure to reverse the delay was particularly clear among cases of moderate anemia (Hb < 100 g/L) that did not show an improvement of their anemia. On the other hand, those subjects whose iron stores were repleted had full reversal of the delay in motor development. Because of design limitations, these studies failed to provide a basis for any conclusive inferences.

Randomized trials with preschool and school children yielded findings similar to the two studies cited above that were based on experimental designs (Pollitt and Metallinos-Katsaras 1990). In short, there is strong evidence leading to the conclusion that iron deficiency has an adverse effect on performance on tests that tap a wide range of behaviors and cognitive processes.

Weaknesses in design

Although research on the effects of mild-to-moderate PEM in human populations has continued unabated during the last 15 y, a substantial portion of the work, particularly the quasi- and experimental studies, was done in the late 1960s and early 1970s (Brozek and Schurch 1984, Pollitt 1988). Correlational studies relied on anthropometry criteria to select subjects and classify them as well nourished or at risk of PEM. Quasi- and experimental studies were principally concerned with the effects of protein and energy supplementation on growth and development. Neither approach, correlational or experimental, dealt satisfactorily with the role of micronutrients in general or of iron in particular as a potential confounder. This problem is illustrated in the following two influential studies in Guatemala.

Cravioto et al. (1966) conducted a study in the village of Magdalena in the Department of Sacatepoquez that yielded what are probably the most influential correlational data in this area of research during the last 40 y. They assessed the relationship between anthropometric measurements and performance in tests on intersensory integration among school-age children because of the concern that protein-calorie malnutrition affected not only stature but also the capacity to learn (p, 319). At issue here are not the well-established weaknesses of a correlational design but, rather, the operational definition of the independent variables and their determination of which culprits were responsible for the observed deficits in neurointegration.

On the basis of a dietary survey conducted in 1950 and three more in 1963, the authors concluded that the "present diet continues to be protein-poor and is not significantly more adequate than that consumed 13 yrs ago" (p. 3361. Tall and short children from 6 to 11 y of age living in Magadalena were compared on a test of neurointegrative development (requiring matching of forms on the basis of information from different sensory modalities). The tall children performed significantly better than the short group, suggesting the possibility that nutritional deficiencies in the short children accounted for their poorer performance. Aware of the fact that the stature differences might be associated with factors other than nutrition, these investigators attempted to evaluate the possible role of parental stature and of differences in rate of maturation. In addition, they were concerned with the possible influence of various socioeconomic, educational and other environmental factors on intersensory test performance. In this connection, they also tested a control group of tall and short, upper-middle-class urban children and obtained their parental heights, as well as information on the environmental background of the rural children. Their results indicated that parental height was not significantly related to children's height in the rural group, although there was a slight trend in this direction for father's height. In the urban upper-middle-class group, however, father's height was significantly related to children's height. This comparison was taken to suggest that, for the rural children, variations in stature are determined more by nutritional variation than by genetic endowment, whereas the opposite is true for the urban upper-middle-class children. Because there were no significant psychological test score differences between the tall and short urban children, the case of nutritional influences on the contrasting test performance of the tall and short rural groups was seen as strengthened.

The Comment section at the end of the brief monograph by Cravioto et al. (1966) did not identify deficits in the intake of protein and energy as being solely responsible for the poor neurointegrative test performance of the short rural children. Infection and a "failure to have received appropriate amounts and kinds of food (primary malnutrition)..." were considered as explanatory factors. Nonetheless, in the section that discusses possible mechanisms behind the observed effects, the authors refer exclusively to data that suggest that protein depletion impairs the structure and growth of the brain (p.. 358).

Two arguments could be used in favor of the assumption that iron deficiency played the role of a confounder in this study. First, as with stunting, which has its origins in early life (Martorell et al. 1990), cognitive deficits observed in middle childhood were traced back to iron deficiency in early life (Lozoff et al. 1991, Palti et al. 1983). The other is that a majority of studies concerned with iron deficiency and physical growth reported significant covariance between these two conditions, and some studies reported a salutary effect of iron supplementation on physical growth among anemic children. For example, in Guatemala, Peragallo-Guarda (1984) found an inverse association between degree of anemia and stature; in Central Java (Chwang et al. 1988) there was evidence of increased growth velocity among iron-deficient anemic children who received iron supplementation; and recently in Kenya, Lawless and collaborators (1994) found that iron supplements over a 14-wk period resulted in improved weight.

The second example is the quasi-experimental trial on the effects of early supplementary feeding on growth and development conducted by staff from the Institute of Nutrition of Central America and Panama in Guatemala (Read and Habicht 1992, Habicht and Martorell 1992). The trial intended to test the effect of improving protein status among children in four rural villages in the Department of El Progreso because of the concern at the time with protein as the major limiting factor in the diets of children in the Third World. The study design involved two villages that received either a high protein/high calorie beverage (Atole) and two other villages that received a placebo (that is, a low calorie drink called Fresco). The two groups (Atole/Fresco) received the same amount of micronutrients per unit of volume of the respective beverages.

In a motor development scale administered at 24 mo and on a general cognitive test battery at 48 and 60 mo of age, and with controls for social and economic status in place, the children who received Atole obtained better scores than the children who received Fresco (Engle et al. 1992). In adolescence, statistically significant differences were also observed between groups in tests of complex mental abilities (for example, reading comprehension, vocabulary and arithmetic) and information processing (Pollitt et al. 1993).

Of importance here are the data on the actual amounts of Atole and Fresco consumed and the relationships between these amounts and the performance of the subjects in the tests administered. For this purpose analyses were conducted of the amount of supplement consumed by the pregnant women and their offspring during the first 2 y of life and of the relationship between levels of consumption and developmental test performance (Oh S.-Y., Pollitt, E. & Martorell, R., unpublished results). Effects of pre- and postnatal intervention were analyzed separately, and comparisons were made for three ages (6, 15 and 24 mo) and for each product and level (high/low) of supplementation. The analyses used energy intake as the primary indicator of supplementation because energy/protein and energy/micronutrient ratios were constant in each supplement. Prenatal supplementation was defined as mean daily supplementary energy intake of mothers of the children during pregnancy On the basis of the median value for maternal supplementary energy intake within both Atole and Fresco villages, children were divided into high- and low-energy supplement. Postnatal supplementation was defined as mean daily child supplementary energy intake from 3 mo of age to time of testing.

Recall that the calorie content of the Atole and Fresco differed, but micronutrients were directly proportional to the volume of both drinks. For the purposes of the analyses the nutrient contents of the supplements consumed by the four subgroups were calculated as percentages of the US recommended dietary allowance per day. There is no reason to assume that physiological importance of percentage units are equivalent across nutrients. The intent in the approach adopted was to use the RDAs for descriptive purposes.

As observed in Table 1, the highest intakes of energy, protein and iron were consistently observed in the High/Atole group, whereas the opposite was true for the Low/Fresco group, both for the pre- and postnatal supplementation. Because of such multiple differences, it is impossible to discriminate between effects of particular nutritional factors. The High/Atole and Low/Fresco groups are, in turn, the subgroups showing the largest differences in mental and motor scores, in relationship to both periods of supplementation. Of 12 comparisons (tests (mental/motor), age (8, 15 and 24 mo) and period (pre/postnatal)], 8 were statistically significant (p < 0.05). None of the other subgroup comparisons showed such a pattern. Moreover, the comparisons of the High/Atole and the High/Fresco do not yield an answer to the particular question of the effects of protein because these two groups also differed, both pre- and postnatally, in iron intake. Developmental test performance differed in the expected direction, in relation to prenatal supplementation, in the mental evaluation at 15 mo and in the motor evaluation at 15 and 24 mot In relationship to the postnatal assessment, the children differed in their motor development at 15 and 24 mot

TABLE 1
Protein, energy and iron supplement (percent of RDAs) of high and low energy groups in the Atole and Fresco groups

None of the group differences of nutrient intakes from supplementary feeding showed an acceptable degree of concordance with group differences of developmental outcomes. In fact, the degree of discordance was such that it precluded any conclusive inferences regarding the differential effects of energy, protein and-iron. Neither can the role of other micronutrients be discounted.

Briefly, the two studies reviewed provided no conclusive evidence that the observed potential effects of malnutrition were distinctly related to a deficit in energy and protein intake. Theoretically, the differences observed as a function of body size and supplement intake could be explained by differences in iron status. A recent study in Mexico (Vega-Franco et al. 1991) addressed this particular issue of iron as a confounder in an analysis of the relationship between PEM and intelligence. The authors compared the intelligence quotient of 129 6- to 11-y-old children who fell above and below 90% of weight/height (reference standard not specified,, controlling for iron deficiency anemia. Once the effect of anemia was partialled out, the differences in IQ between the anthropometric groups were no longer observed.

Although no definitive conclusions are warranted, the data presented suggest the need for caution in the interpretation of findings from published studies on the effects of mild-to-moderate PEM on development. Because of methodological weaknesses in studies on PEM and cognition and of the established effects of iron deficiency on mental and motor development, it is plausible that studies attributing the effects of PEM were mistaken. Such effects could have been produced by iron deficiency anemia. The published data, however, do not allow for a retrospective assessment of possible additive or interactive effects from both conditions.

Contents - Previous - Next