Contents - Previous - Next

3.3. Outcomes of monofocal and multifocal interventions

The third source of information pointing to the inadequacies of a bivariate model is the difference between the results of studies on nutritional supplementation with and without a health and an education component. Nutritional supplementation of pregnant and lactating women, and of infants and preschool children, yielded mild effects on developmental outcomes. On the other hand, multifocal interventions which included nutrition supplementation, health care delivery, and educational stimulation resulted in impressive developmental benefits.

The Table (from JOOS and POLLITT, 1984) presents in summary form the main developmental test findings of four supplementation studies conducted in Bogota, Colombia; Guatemala; Harlem, New York; and Sui Lin, Taiwan8. The brief review of these four studies that follows, illustrates that food supplementation alone, among nutritionally at risk women and young children, had mild, albeit statistically significant developmental effects.

8A detailed report of the components of each one of these interventions and a comparative analysis of the findings are reported in JOOS and POLLITT (1984). See also references for each study for specific information regarding research design and findings.

The results (see Table) for the study conducted in Bogota, Colombia, are the mean scores on the general quotient (GQ) of the Griffiths test9 in the three groups that received only supplement and the untreated control group (WABER et al., 1981). The analyses up to 18 months showed independent effects of nutritional supplementation primarily in the subscales that reflect motor functioning (Locomotor and Performance) and the GQ. Comparison of the mean GQ and subscale scores of the supplemented and control groups shows that, while the supplemented groups have higher mean scores than the control group, these differences are not statistically significant for any subscales at any one age. Statistically significant differences were restricted to intergroup comparisons across ages. Also, there was no independent effect of nutrition on the Albert Einstein Scale of Sensory Motor Intelligence10. A subsequent analysis up to three years of age yielded essentially the same findings, except that the nutrition treatment had a significantly greater effect on scores of the females than of males.

9 For a review of the methods used for the behavioral assessment of the consequences of malnutrition see POLLITT (1984).

10 In contrast to most other psychometric techniques used in most other empirically-based studies of undernutrition and behavior, the Albert Einstein Scale of Sensory Motor Intelligence was constructed on the basis of Piaget's epigenetic theory of intelligence.

The second set of scores presented in the Table are the mean scores in the mean and motor subscales of the Composite Infant Scales used in a supplementation study conducted in Guatemala (KLEIN et al., 1976). At six months of age, the differences between low, medium and high supplementation groups were statistically significant only for the mental subscale. At 15 and 24 months the absolute differences were greater and significant for both subscales. Item analyses suggested that the impact of supplement ingestion was more associated with motor and manipulative items within each scale than with the more cognitive or linguistic items. As in the Bogota study, effects were more apparent in the motor subscales, though the absolute differences were more pronounced in the Guatemala data. No sex differences were found11.

11 Social behavior may represent an exception to the effects of the supplementation restricted to the preschool period. Among children B to 10 years old from three of the tour villages participating in this supplementation study, several aspects of social behavior (personal interactions with peers and adults, activity levels, response to the physical environment and affect) appeared to be related to nutrition in early childhood (BARRETT, YARROW and KLEIN, 1982). Controlling for socioeconomic status, high levels of supplementation between birth and two years of age were predictive of high levels of social involvement, happy and angry affect, and moderate levels of activity, while low supplementation was associated with passivity, dependency on adults, and anxious behavior.

A serious methodological problem in this Guatemalan study is the self-selection in the sample to determine frequency of participation in the supplementation program. Participation was likely to be influenced by the subjects' motivation to take advantage of the available supplement. This social behavioral variable could have determined, at least in part, the variations observed in social development (BARRETT, 1984).

Table. Mean scores on developmental scales


Ages in months

























































74 4**





New Yorkc














Mental (modified)





Motor (modified)





a S1 = Supplemented during pregnancy and infancy; S2 = Supplemented after 6-months old; S3 = Supplemented during pregnancy and until 6-months old; U = Unsupplemented controls.
b L = Low supplement intake (<5000 kcal/quarter for 14 quarters); M = Medium supplement intake (5000 to 10,000 kcal); H = High supplement intake (>10,000 kcal).
c S = "Supplement" group; C = "Complement"; U = Unsupplemented controls.
d S = High calorie and protein supplement; U = Placebo supplement.
e Average of male and female scores, Table 2, WABER et al., 1981.
*(p<.05) (F-test); ** (p<.01) (F-test); *** p = <.06 (2-tail t-test).

A third supplementation study was conducted in Harlem, New York (RUSH, STEIN and SUSSER, 1980). As indicated in the Table, there were no differences between the supplement, complement and control groups on either the mental or motor scales of the Bayley test at 12 months of age included in this study. The fourth and last study was carried out in the township of Sui Lin in Taiwan (JOOS et al., 1983). Treatment group differences in mental and motor scores (i.e., Bayley Scales) were tested with males and females combined and separated. The overall mental scores of the supplemented and the control group infants were not significantly different from each other. The motor scores of the supplement group (both sexes combined) were higher (p<.05) than those of the controls. However, analyses of the motor scores by sex did not show statistically significant differences between supplemented and unsupplemented children. Further, there were no significant correlations of volume of intake at any point in time (i.e., pregnancy and lactation) with any of the Bayley Scale scores.

This consistency of results in the four studies is striking considering that the selection of study sample, type of nutrition intervention, behavioral scales and ages at testing differed across the studies. Accordingly, the evidence from intervention studies suggests that the magnitude of developmental variability accounted for by nutritional variability alone is small, albeit statistically significant. These effects disappear in the school period and after terminating the supplementation program. The possibility of dormant effects that could be detected in later life still exists. Dormant effects were observed in early intervention studies among educationally disadvantaged children in the United States (Consortium for Longitudinal Studies, 1983).

In contrast to the mild effects of interventions restricted to nutrition supplementation, interventions that included health care and educational stimulation plus nutrition supplementation had major developmental effects (MCKAY et al., 1977). A study in Cali, Colombia, involving moderately to severely malnourished preschool children living in an impoverished urban community showed that the effects on cognitive developmental measures after four years of treatment were as large as .50 to .75 of a standard deviation. The size of these effects were particularly surprising because the children were enlisted when they were 42 months old. At this age they had already past the time of the so-called critical period of brain growth.

The difference in the magnitude of the effects between monofocal and multifocal interventions should not be surprising. The greater the improvements in the environmental conditions, the greater the developmental advantages for children. At issue here, however, is not the difference in itself but its magnitude, since it points to either additive or interactive effects. The former refers to the co-action of different inputs, whereas the latter reveals a synergistic effect, namely, that the health care and educational stimulation would potentiate the effects of the nutritional supplementation and vice versa. Unfortunately, the data available in the published reports are insufficient to allow for a comparative statistical analysis of the studies to determine the precise nature of the difference. In my view, the size of the effects in the Cali study is too large in comparison to that of the other studies to assume that they are simply additive.

An intervention study in Mexico (CHÁVEZ and MARTÍNEZ, 1982) is enlightening at this point because of the strong developmental effects obtained in the context of a study design that called for a nutritional supplement without additional inputs. The intervention began in pregnancy and continued throughout early childhood. The benefits were observed in cognitive measures in infancy and in the preschool period and later on in school attainment (CHÁVEZ and MARTÍNEZ, 1982a).

The investigators, as previously noted, intended to limit the experimental treatment to a nutritional supplement, but their reports strongly suggest that the experiment became a major social intervention. The conditions under which the supplement was delivered intensified the social interactions of the families and broadened their social contacts. In addition, the children that received the "treatment" acquired prestige or prominence within the community. The final outcome was that the intended intervention and the contextual response resulted in potent effects that, even in the absence of statistical comparisons between studies, were significantly greater than the effects observed in the other nutrition supplementation studies. It is thus reasonable to generalize that the differences already discussed between the benefits derived from the monofocal and the multifocal interventions are not determined only by the differences in the additive effect of the number of inputs. The interactions between nutrition and environmental factors are more likely to lie behind such differences, but these interactions have escaped the main-effect model.

In summary, the evidence presented on the three issues listed above indicates that the main-effect model is not conducive to an understanding of the developmental outcomes of children with a history of undernutrition. The direction in the trajectory of development of a child with a history of undernutrition depends not only on the nutritional history but on the environmental circumstances (e.g., attenuating circumstances, other risk factors) to which the organism is exposed, and the nature of its health history. Further, the notion of irreversible damage to brain function because of malnutrition during a so-called critical period is no longer tenable, as it is also no longer tenable in other areas of developmental psychobiology (GOTTLIEB, 1983). In fact, strong evidence points to the conclusion that unless "major and irreparable physiological insults" (HOROWITZ, 1987) have occurred, environmental conditions can modify the developmental effects of any biological or social risk factor to which the child is exposed in early life.

4. Suggestions for future research

For the purposes of this paper, a developmental risk factor was defined (see footnote 1) as a circumstance or event occurring in an early life period that increases the probability of diverting a child's growth and developmental trajectory from a course typically followed by children when physiological and emotional needs are met. Although the main-effect model has failed to give a conclusive answer to the issue of whether or not early undernutrition is a developmental risk factor, the evidence on the synergism between undernutrition and social-environmental factors strongly supports this contention. Through its interactions with illness and adverse family and socioeconomic conditions, undernutrition increases the probability of diverting the trajectory of mental development, a hazard not encountered by other children living under similar environmental and economic conditions who do not experience undernutrition. Moreover, and perhaps more importantly, undernutrition shapes behavioral adaptation into forms that are different from those of well-nourished children.

The evidence also points in the direction of research that integrates nutritional variability and context along a time dimension. Endemic infant and childhood malnutrition is a tragic manmade condition that provides a unique natural experiment to understand the biological basis of human behavior. Behaviorally-oriented research among chronically undernourished children has generally defined this condition as a biological insult, disregarding the behavioral implications of its long-term effects on the energy balance equation. A likely outcome of low energy expenditure, to maintain homeostasis, is reduction in activity, an important component of the behavioral repertoire of infants and young children (GOLDSMITH et al., 1987). As such, low energy expenditure is likely to shape interactions between the organism and the immediate environment (GRAVES, 1976, 1978; RICCUTI, 1981; BEATON, 1983; VALENZUELA and ARAYA, 1987), it would be of particular developmental interest to determine the ways in which the undernourished organism establishes transactions12 with caretakers, that is, the ways in which the energy-deficient child influences behaviors of "others" so that the energy demands are in keeping with its energy balance.

12 For present purposes I use the term transaction following RUTTER's definition (1983): "Transactional effects differ from all the interaction effects considered thus far in that it is not a question of one variable potentiating, reducing, or altering the effect of some other variable or outcome, but rather of one variable changing the other" (p. 305).

On the other hand, if research interest stems from consideration of applied issues (public health) then the greatest payoff lies in questions aimed at identifying and operationalizing those environmental factors that, in conjunction with undernutrition, increase children's developmental risk. Evidence from research on other biological risk factors suggests that cumulative risk through the additive and interactive effects of social and biological risk factors in time might be the best predictor of final outcome. Accordingly, there could be significant public health payoff in the identification of family or community "risk" variables13.

13 These could be conceptualized as time-independent or time-dependent. The former refers to structural variables, such as parental illiteracy, which are not likely to change throughout the child's infancy and early childhood. Time-dependent variables vary as the child grows older. These variations could occur in a systematic way. For example, rainy seasons with higher demands on maternal work could be considered a time-dependent variable that varies systematically within a 12-month period and represents a growth risk to infants and young children (ADAIR and POLLITT, 1985). Increases in family size do not follow a specific schedule; they generally result in caretaking practices within the household which could have adverse consequences for young children.

A recognition that it is necessary to change from bivariate to multivariate models is not a guarantee that the main effect model and bivariate equations will not be the prescriptions used in research on undernutrition and behavioral development. It does, however, provide a conceptual umbrella to understand better the developmental meaning of the data that are generated from research with different conceptual and methodological approaches.

Contents - Previous - Next