|Early Supplementary Feeding and Cognition (Society for Research in Child Development, 1993, 123 pages)|
This Monograph reports results from a study of the effects of supplementary feeding of newborn to 7-year-old children on their subsequent performance as adolescents and young adults (11-24 years) on a battery of psychoeducational and information-processing tests. The study, which began in 1969, was conducted in a nutritionally at-risk population in the Department of El Progreso, Eastern Guatemala. The intent was to test the hypothesis that protein deficiency in infants and children delays mental development.
We report the effects of the nutritional intervention on cognition during infancy, the preschool period, and, more extensively, adolescence and early adulthood. Issues of differential effects from pre- versus postnatal supplement are excluded; similarly, the current analyses do not attend to the possible differential effect of specific nutrients.
This section describes the key theoretical and research issues in international nutrition and in developmental biology and psychology that were dominant at the time the study was launched (1969). The discussion aims to justify the study design that was chosen and the research questions of concern here.
International Nutrition and the Protein Gap
Cicely Williams's (1933) description of kwashiorkor and her suggestion that its etiology could be found in a deficiency of "amino acids or protein" became a landmark; thereafter, protein deficiency became the object of intensive study and international concern. 1 For at least the following four decades, protein was considered to be the major limiting factor in the diets of most undernourished children in the less developed countries, particularly in Africa. Illustrative is the 1953 publication of an Expert Committee on Nutrition from the UN Food and Agricultural Organization and the World Health Organization (cited in McLaren, 1974). Without the necessary epidemiological data, it nevertheless stated that kwashiorkor was "the most serious and widespread nutritional disorder known to medical and nutritional sciences." Thus, it is not surprising that the United Nations instituted a Protein Advisory Group in 1955 and that groups of experts in international nutrition provided strong recommendations to develop local protein-rich foods. Moreover, evaluations of lysine fortification programs were also launched in target populations considered protein deficient (McLaren, 1974).
1 Kwashiorkor is the result of a
protein-deficient diet. It is generally an acute condition that tends to occur
at the end of the first or during the second year of life. Oedema,
hypopigmentation or dermatoses, dyspigmentation of the hair, and a significant
increment in the size of the liver are pathognomonic signs of protein
However, by the late 1950s and the early 1960s, it was becoming apparent that protein was not the sole growth-limiting factor among malnourished children. In 1959, Jelliffe coined the term "protein calorie deficiency" to encompass kwashiorkor, marasmus, and mild to moderate forms of protein calorie malnutrition. 2 In 1974, the head of the division of nutrition at the World Health Organization estimated that there were 98,470,000 children (0-4 years old) suffering from PEM in Asia, Africa, and Latin America (Bengoa, 1974).
2 Marasmus is characterized by severe
retardation in linear growth as the final stage in a long process of depletion
due to a diet poor in quality and quantity. It often occurs during the first
year of life, but it is also found among older children. Oedema is not present.
The term "protein calorie malnutrition" was subsequently changed to
"protein energy malnutrition" (PEM); we use the latter in this
PEM and Mental Development
The hypothesis linking PEM (protein energy malnutrition) to later cognitive deficit can be traced to the beginning of the twentieth century (see, e.g., Nicholls, 1923). Most of the research, however, was conducted during the mid-1950s (Geber & Dean, 1955, 1956) and the 1960s (Cabak & Najdanvic, 1965; Cravioto & Robles, 1965; Jelliffe, 1965; Pollitt & Granoff, 1967; Scrimshaw, 1969; Scrimshaw & Gordon, 1968; Stoch & Smythe, 1963). The published data extended clinical observations on malnourished children. Rich descriptions were provided of the behavioral apathy, listlessness, and reduced motor activity observed among severely undernourished children (Autret & Béhar, 1954; Jelliffe, 1965). Thus, it was only natural to suspect that undernutrition affected the brain since its effects on vital organs such as the liver had been well documented (McLaren, Faris, & Zekian, 1968).
Most, if not all, of the studies conducted during the 1960s on the effects of PEM on human behavior relied on correlational research designs. Investigators observed that infants and young children with a history of severe PEM (i.e., marasmus, kwashiorkor, or both) had poorer scores on developmental scales (e.g., the Bayley Scales), IQ tests, and tasks testing specific intellectual functions than comparison groups (for reviews, see Pollitt & Thomson, 1977; Ricciuti, 1981). Stunted children without clinical signs of severe malnutrition also performed comparatively poorly on tests of inter-sensory integration (Cravioto, DeLicardie, & Birch, 1966).
Inferences made at that time about the causal effects of PEM on cognition were thus based primarily on correlational data. Concerns regarding possible confounding and moderating effects of social and economic variables that covaried with nutritional status were dealt with by matched-group research designs. Subjects differing in anthropometry (e.g., above and below the tenth percentile for height distribution of the Harvard Growth Standards) were matched on an array of potential confounders (e.g., mother's education, father's occupation), and their performance on psychological tests (e.g., IQ) was compared (Cabak & Najdanvic, 1965; Hertzig, Birch, Richardson, & Tizard, 1972).
The consistency in findings among different studies did not compensate for the critical limitations of ex post facto research designs or for problems inherent to the nature of the research question. The social-psychological factors that characterized the immediate family environment of the PEM child were neither known nor controlled for. The factors that were controlled for were broad, graduated parameters of social structure such as mother's education or family income. Thus, conclusive inferences could not be drawn from such studies, no matter how carefully investigators matched the samples. In addition, operational definitions of nutritional status based on anthropometry are crude since body size is not determined solely by nutrient intake (Pollitt & Ricciuti, 1969; for an early, articulate discussion of this issue, see Suchman, 1968).
Although it was recognized at the time that laboratory studies conducted with experimental animals could make important contributions to understanding the problem, such studies also had serious limitations. Their focus was on severe protein deficiency, and the human conditions of mild to moderate malnutrition were not simulated in the laboratory. In addition, the animal models could not include social-environmental variables considered to be potential confounders or effect modifiers (Pollitt, 1969).
Recognition of the limitations in the designs of the studies conducted in this period led to an appreciation of the scientific merits of research protocols that included supplementary feeding as a key experimental variable. Shifting to this approach provided an opportunity to randomize study subjects and to control a substantive portion of a child's daily nutrient intake. It also allowed for the manipulation of the timing and duration of exposure to the experimental treatment and for a focus on mild to moderate PEM. During the 1960s, at least six large studies of the effects of early supplementary feeding on human development were implemented in different parts of the world, including the Guatemala study. Although supplementary feeding provided an alternative, and improved, research design to correlational studies, there are important limitations and considerations inherent in quasi-experimental designs as well. Specific issues relating to this study are discussed more fully in Chapter IV.
Developmental Psychobiology and the Effects of PEM on Cognition
Concerns with the putative effects of PEM on cognition that were prevalent in the 1960s were reinforced by innovative thinking in developmental psychobiology. Influential theoreticians postulated that the nature of early experiences resulted in permanent changes in the organization of neural cells in the cerebrum (Hebb, 1949; Hunt, 1961). Experimental research showed that brain parameters and constituents, such as brain weight, neuronal density, and nucleic acid, could be affected by the nature of the interactions that the organism had with its environment (Rosenzweig, 1966).
Similarly, the notion of critical periods in development was a compelling argument advanced in the late 1960s in support of a paradigm that related particular early experiences with later behavioral attributes. During its early development, the central nervous system was considered particularly vulnerable to teratogenic agents or to particular social stimuli (Bronfenbrenner, 1968; Scott, 1962). This line of thinking provided further justification for the hypothesis that early PEM results in permanent developmental changes (Pryor, 1974) - its face validity was obvious.
Research conducted during this period on the effects of early experience on development generally fit the main effect model and, in particular, the traditional biomedical model of disease causation, namely, that a given pathogen will lead to a specific disease. There was little regard for social and environmental factors as modifiers of the potential effects of early biological trauma; study designs focused on a linear cause-and-effect relation. The agent would be defined as "cultural" or "environmental deprivation," while the effect ("disease") was a comparatively low score on a mental test such as the Stanford-Binet. There were also attempts to define the mechanisms through which "deprivation" worked (Bernstein, 1961; Hess & Shipman, 1965, 1967).
Within this line of thinking, it seemed an urgent necessity to develop technologies to prevent or remediate the adverse effects of an impoverished environment through early intervention programs (Consortium for Longitudinal Studies, 1983). It also made sense that the War on Poverty would be launched and that it should include programs such as Head Start to prevent or remediate the effects of poverty and malnutrition. Work on the effects of early educational interventions on behavioral development was compatible with the notion that early supplementary feeding could prevent the emergence of effects of malnutrition among children living in severe poverty. The hypotheses formulated in both lines of endeavor were similar, and they called for a reductionistic approach that fit the biomedical model. This was attractive to many researchers, particularly in nutrition, where laboratory research and animal models were the preferred form of investigation.
In sum, theory and research in international nutrition and in developmental psychobiology during the 1960s provided strong justification for major experimental studies of PEM and its effects on mental development. Assumptions made at that time regarding the epidemiology of PEM, and the proposition that early experience during critical periods of brain growth could have lasting effects, provided strong support for the notion that protein was a major limiting nutritional factor among undernourished children and that PEM was a major biological trauma. Similarly, recognition of the limitations of observational studies conducted on humans and of experiments with laboratory animals supported the adoption of experimental designs that included early supplementary feeding as the key independent variable.
In the 1960s, the Institute of Nutrition of Central America and Panama (INCAP) was already a leading international nutrition institution. This experience of the institute's staff in field studies and the institute's attractiveness to prestigious scientists in nutrition epidemiology made it natural for INCAP to take a leadership role in this area of research by launching a quasi-experimental study that was intended to yield unique information on the effects of early nutrition on cognitive development.
The research design of the Guatemala study has included a period of longitudinal study spanning the years 1969-1977 and a cross-sectional follow-up during part of 1988-1989. In the next chapter, we describe the villages and the subjects chosen as well as the methods and results of the longitudinal study. Chapter III provides the conceptual underpinnings for the 1988 follow-up, including a review of the relevant literature as well as a discussion of how knowledge gained in the years since the inception of the Guatemala study might have led to a different conceptual and methodological approach. Chapter IV discusses the assumptions and deficiencies inherent in the original design that imposed limitations in the 1988 follow-up. Chapters V, VI, and VII present the methods and the results of the follow-up and a discussion of its findings. Chapter VII also presents what we believe to be the mechanisms underlying the observed relations between nutritional input and behavioral changes and the policy implications of these data. One of our main concerns is with the way in which the results that we obtained may be used in justifying nutrition intervention studies among nutritionally at-risk children.