Contents - Previous - Next

2. Dietary quality and linear growth

2.1. Existence of simultaneous multiple micronutrient deficiencies

Except in the case of iodine deficiency, where status is less dependent on the general adequacy of the food supply, there are several reasons to believe that multiple micronutrient deficiencies will exist in most other situations. First, if food is so scarce that energy intake is inadequate, then dietary quality is also likely to be poor. Specifically, a poor quality diet would contain few animal products, fruits and vegetables, and consist primarily of staples such as cereals, legumes or other plants (Allen et al., 1991). Such poor quality diets are associated with low intakes of several vitamins and minerals and poor mineral bioavailability. Poor dietary quality is a more common situation than actual food shortage in less wealthy countries, population groups and individuals. Also, morbidity is likely to cause depletion of several nutrients simultaneously, through anorexia or malabsorption.

2.2. Examples from the Nutrition CRSP

Data from the Nutrition Collaborative Research Support Program (CRSP) support the importance of dietary quality and the existence of multiple micronutrient deficiencies in developing countries. The Nutrition CRSP was a longitudinal study of the impacts of marginal malnutrition on the function of infants, preschool-children, schoolchildren and adults in Mexico, Kenya and Egypt 1. Multiple food intake measures on these individuals made it possible to explore relationships between the intake of specific foods and nutrients, and growth (among other functions). The project has been described in detail elsewhere (Allen et al., 1992a; Kirksey et al., 1992; Neumann, Bwibo & Sigman, 1992; Calloway, Murphy & Beaton, 1992).

1 The Nutrition CRSP was supported by US-AID Grants #DAN-1309-G-SS-1070-00 and DAN-1309-A-00-9090-00.

Similar to the situation in many developing countries, linear growth faltering occurred within a few months of birth and was substantially complete by about 18-22 months of age (Fig. 1). The growth pattern was similar in Mexico, Kenya and Egypt. The important question here is the extent to which this linear growth faltering is associated with dietary patterns or nutrient deficiencies. In Mexico and Kenya the staple foods are maize and legumes, while in Egypt wheat and rice provide most of the energy. Dairy products are eaten in very small quantities in all three locations, especially by Mexican and Kenyan schoolchildren. Other animal products, specifically meat and eggs, are eaten in low amounts especially in Kenya. However, among the CRSP preschool-children (and other individuals) within each country there was a nutritionally significant range in variation in dietary quality which permitted analysis of the relationship between dietary patterns and growth stunting.

Energy (food) shortage was not a problem in Egypt and Mexico based on evidence such as: energy intakes met recommended levels on average; energy intake of adult women increased substantially in early lactation, showing the availability of additional food; adult women were relatively fat and body mass index increased with each decade of life. Analysis of data from Mexico showed that household energy intake kept pace with household energy requirements, while the quality of household diets (consumption of animal products), and especially that of preschool-children's diets, did not (Allen et al., 1992; Backstrand et al., 1992). In Kenya, where periods of drought-induced food shortage occurred during the project, diets were low in both quantity and quality.

Beaton, Calloway & Murphy (1992) calculated the probability of protein and/or lysine deficiency in the three CRSP locations. In Egypt and Mexico protein and lysine intakes were adequate. There was a very small risk that protein and lysine intakes were inadequate for some Kenyan children, but had energy intake (food supply) been adequate, the prevalence of low intakes would have been negligible. Lysine intakes were about twice as high as the FAO/WHO recommendations, and tryptophan, threonine and sulfur amino acid intakes were even more adequate. It is unlikely, therefore, that protein or essential amino acid deficiencies were limiting the linear growth of these preschool-children.

Fig. 1. Height Z scores from birth to 145 months of age in the Nutrition CRSP.

In contrast to the adequacy of energy and protein intakes, none of the diets provided adequate amounts of vitamins or minerals. Calloway et al. (1992) calculated the probability of inadequate vitamin intakes in preschool-children, based on approximately two days of food intake per month for one year. Egyptian preschool-children frequently had inadequate intakes of vitamins A, E and riboflavin. Mexico appeared to be the worst situation in that there was a very high prevalence of inadequate intakes of vitamins A and E, riboflavin, B12, and ascorbic acid. In Kenya the main vitamin inadequacies were vitamins E, A and B12. In general, preschool-children in the CRSP countries who had low intakes of animal products had the lowest intakes of fat, and vitamins A, B2, B12, C and E (Backstrand et al., 1992).

Estimation of the adequacy of mineral intakes is complicated by their bioavailability, which is likely to be poorest when diets contain more foods that have a high content of phytate and fiber, and are low in enhancers of absorption such as animal protein and ascorbic acid. Murphy, Beaton & Calloway (1992) used algorithms that considered factors that could affect bioavailability of iron, zinc and copper in the preschool-children's diets, and concluded that between 31% and 65% would not meet their basal requirement for iron, and 12 to 43% would not absorb enough to prevent anemia. Virtually all Kenyan children and most Mexican children are expected to have inadequate intakes of zinc. Copper intake was not inadequate in any of the populations.

The reasonableness of these prevalence estimates is substantiated by biochemical data. Anemia and iron deficiency were highly prevalent in all three countries. At around 30 months of age, the percentages of children with low hemoglobin and low ferritin respectively were: Egypt, 38 and 73; Kenya, 40 and 74; and Mexico, 45 and 62 (Murphy, Beaton & Calloway, 1992).

In a subsequent study of 220 children, 18-30 months old, in the same region of Mexico, we found that 84% had anemia, 40% had deficient or low serum retinol, 59% had deficient ferritin, and 41% had low or deficient vitamin B12 (unpublished data). Although not measured, the presence of other micronutrient deficiencies is highly probable.

2.3. Associations between dietary quality and growth of preschool-children

The existence of simultaneous deficiencies made it difficult to detect associations between the intake of a single nutrient and growth. For example, in the Mexican preschool-children there was no significant relationship between size at 30 months (weight, length or weight-for-length) and each child's average intake of energy, protein, or other nutrients during the previous 12 months (Allen et al., 1992). Stronger, positive associations were found between the intake of specific foods and linear growth. The usual diet of taller children contained more animal products (especially milk and meat), and fewer maize tortillas than that of shorter children. These relationships persisted when socioeconomic status was controlled in analyses. Weight was less strongly related to the intake of specific foods.

These associations raised the question of why animal product intake was associated with the children's linear growth. Through a series of additional analyses that are described in detail elsewhere (Allen et al., 1992a), several important generalizations emerged from the Mexico data that are likely to apply to many populations with limited resources. One was the appreciation of the importance of considering dietary patterns rather than nutrient intakes. Those children who consumed proportionately more maize tortillas ate substantially less meat, dairy products, fruit and 'other plant' products (mostly refined cereals). Dairy product intake fell to almost zero when energy intake from tortillas exceeded 61%, a situation that occurred in about half of the preschool-children. Thus, it is difficult to ascribe growth stunting to lack of a single food group or excess of another, because a lower intake of, for example, dairy products is also associated with a lower intake of meat, fruits and vegetables and a higher intake of high phytate, high fiber tortillas. Rather, it is how the child ranks on this dietary pattern continuum that is important.

Children who ate more tortillas actually consumed more iron, thiamin, zinc, energy, calcium and niacin because of the high levels of these nutrients in maize. Because tortilla intake was negatively associated with growth, however, it seemed reasonable that the problem might be the poor bioavailability of minerals in this staple food. Indeed, higher intakes of fiber and phytate relative to iron and zinc did predict lower height at 30 months of age (Table 8).

In addition, those preschool-children who consumed the smallest amount of animal products (i.e. the lowest quartile of % energy from animal sources) had an inverse relationship between their weight and length gain during the 18-30 month age period (Allen et al., 1992b). In contrast, when more animal products were consumed, weight and length gain showed the expected positive relationship. Similarly, children in the lowest quartile of animal product intake had a negative, rather than the expected positive, relationship between their length at 30 months and the weight of their mother. These relationships indicate that poor dietary quality impairs the growth potential of young children. A similar conclusion was reached in the Egypt (Kirksey et al., in press) and Kenya (Neumann, Bwibo & Sigman, 1992) Nutrition CRSP projects (also see Neumann & Harrison in this issue).

Table 8. Correlations between nutrient bioavailability and children's size at 30 months 1





Heme iron

0.18 2



Available iron




Fiber: iron


-0.22 3



-0.29 3

-0.25 3


Phytate: zinc

-0.35 3



(Phytate X Ca): zinc

-0.30 3

-0.23 3


1 N = 83 children. Dietary data were average of 2 days per month for previous 12 months;
2 Spearman's correlation coefficients;
3 P < 0.05. From Allen et al. (1991).

3. Nutritional explanations of early linear growth faltering

The preceding analyses have shown that dietary quality is important for the linear growth of preschool-children between 18 and 30 months of age. However, in Fig. 1 it is apparent that the period of greatest growth faltering started very early, at about 3 months after birth, and was essentially complete well before the weaning period. This early growth faltering is common to most developing countries (Martorell & Klein, 1980; Waterlow, Ashworth & Griffiths, 1980). Between about 22 and 40 months of age, linear growth in the group as a whole is undergoing a slight 'catch-up' relative to rates in well-nourished reference children.

Fig. 2 illustrates the relative importance of nutritional factors and their relationship to linear growth failure during the first two years of life. The length Z scores are based on smoothed data from the Mexico Nutrition CRSP, but the values and pattern of change are similar to those in the Egypt and Kenya Nutrition CRSP, and in other populations (Martorell & Klein, 1980; Waterlow, Ashworth & Griffiths, 1980). The relative importance and timing of the nutritional factors is hypothetical, based on a review of the literature.

The main points illustrated in Fig. 2 are that: early growth faltering may be related to sub-optimal fetal endowment with nutrients during pregnancy so that stores are low at birth; this may be compounded by low levels of nutrients in breast milk of the same mothers the peak prevalence of diarrhea and morbidity occurs after growth faltering begins; and in the post-weaning period, when most children consume a diet similar to the rest of the household, there is a relatively faster rate of growth with some 'catch-up'. Interestingly, Martorell & Klein (1980) showed that during this period in the longitudinal INCAP study there was almost no beneficial effect of the nutrient supplements. Again, the major impact of malnutrition, and the greatest response to supplements, occurred prior to 18 months of age.

Fig. 2. Hypothetical model of nutritional influences on children's length, from pregnancy through 32 months of age

The role of perinatal and maternal nutrition in linear growth faltering therefore deserves more attention; there is almost no information on this topic in human populations. Maternal nutritional status is likely to affect fetal endowment of minerals and vitamins, and to predict the age at which stores of these nutrients become depleted postnatally. The size of the infant at birth is known to be a strong predictor of size at 6-8 months and resistance to growth faltering in the early postpartum period (Butte et al., 1992), though the reason for this is not understood. In addition, mothers who have sub-optimal nutritional status during pregnancy are likely to do so in lactation.

It is now generally accepted that maternal capacity to produce a normal volume of milk is relatively unaffected by moderate maternal malnutrition, so that inadequate milk production is unlikely to be a significant factor in early growth faltering. This was confirmed in a recent study by Butte et al. (1992) to determine whether growth faltering of Mexican-Indian infants was attributable to low intakes of breast milk. Compared to a well-nourished reference population in Houston, Texas, the Mexican infants were significantly shorter (even at 4 weeks of age) and their length velocity was significantly slower between 12 and 24 weeks. Supplemental feeding and intakes of other fluids were very low, negligible at 4 months, and unrelated to length or weight velocity. Compared to their Houston counterparts, the growth-faltering Mexican infants actually had higher or similar intakes of energy, higher intakes of protein and carbohydrate and lower intakes of fat from breast milk. The authors concluded that either a different growth-limiting nutrient was deficient in breast milk, or the infants had higher nutrient requirements. An alternative explanation might be poor endowment of micronutrient stores at birth.

There is little available information concerning the important question of whether low levels of specific nutrients in breast milk might contribute to infant growth faltering. Zinc (Karra et al., 1989) and iron (Lönnerdal, 1986) concentrations in breast milk seem to be relatively unaffected by maternal intake (including supplements) or status. In contrast, milk vitamin concentrations are relatively strongly associated with maternal intake and stores, especially in the case of water soluble vitamins (Kirksey, 1986; Allen, in press b). Vitamin A in the form of fortified MSG marketed through ordinary channels in Indonesia, increased the concentration of retinol in breast milk by about 16% but had no impact on infant growth (Muhilal et al., 1988).

4. The impact of diarrhea, infections and parasites on growth

Although there is clearly an association between infection, especially if accompanied by diarrhea, and growth faltering, the extent to which illness is generally responsible for endemic growth faltering is still debatable. Estimates show that perhaps one third of the total amount of linear growth failure can be ascribed to illness. For example, in Guatemala, children who were relatively free from diarrhea during the first seven years of their life were still about 13 cm shorter than well-nourished children in the United States - although 3.5 cm taller than children who had more diarrhea (Martorell et al., 1975a,b). Moreover, although diarrheal disease shows a pattern of increasing at the time when growth faltering is most severe (i.e. between 6 and 12 months) (Martorell et al., 1975a; Martorell, 1980); it is still highly prevalent between 18 and 24 months when growth faltering is less severe.

The possibility that subclinical infections or parasites are causing nutrient malabsorption cannot be discounted. For example, Lunn, Northrop-Clewes & Downes (1991) recently reported that impaired mucosal permeability explained 48% of linear growth faltering in Gambian infants. Asymptomatic infection with parasites such as Giardia, hookworm, Trichuris and Ascaris is common in infants and young children in developing countries, is associated with growth faltering, and may cause malabsorption of nutrients such as iron and vitamin A. Treatment for these parasites improves weight gain substantially but does not improve linear growth (Stephenson et al., 1993; Willett, Kilama & Kihamia, 1979).

To the extent that diarrhea, parasites and other illness are responsible for linear growth faltering, the problem must still be regarded as having a nutritional basis, mediated, for example through poor appetite (especially for non-breast milk foods), and malabsorption of nutrients including zinc and copper. It is still important to know which nutrients become growth-limiting in this situation in order to plan effective interventions. At present, insufficient attention has been paid to this issue.

Acknowledgements - The author gratefully acknowledges the assistance of Dr. J.R. Backstrand in the conceptualization and analysis of data from the Mexico CRSP, and his helpful suggestions for this manuscript.

Contents - Previous - Next