Contents - Previous - Next

4. The INCAP longitudinal study (1969-1977) and its follow-up (1988-1989): An overview of results

Introduction
The INCAP longitudinal study (1969-1977)
Guatemalan follow-up study (1987-1988)
Concluding remarks
Acknowledgments
References
Notes

Reynaldo Martorell1

Introduction

The Institute of Nutrition of Central America and Panama (INCAP) studies described in this chapter are unique in the international nutrition literature. The longitudinal study (1969-1977), now well-known, generated one of the best data sources on the effects of improved nutrition during pregnancy and the first few years of life on the physical growth and mental development of children. Its wealth of longitudinal information on growth, infection, diet, and development has also provided scientists with unparalleled opportunities for exploring diverse questions about child health and nutrition. What makes the situation unique is that a comprehensive, long-term follow-up of former participants in the study was carried out in 1988-1989. This has permitted, for the first time, the possibility of relating improved nutrition in the early stages of life to outcomes measured in the adolescent and young adult.

The results presented here are taken from previous publications. In a supplement to the Food and Nutrition Bulletin (Vol. 14, 1992), my colleagues and I reviewed the history, design, and methods of the INCAP longitudinal study (1969-1977), described the villages and their people, presented key findings from the longitudinal study, and reviewed the history, design, and methods of the follow-up study (1987-1988) carried out on former participants in the longitudinal study. The analysis of the follow-up data is ongoing, but enough results are available to comment on their general nature. A monograph dealing with behavioral outcomes in adolescents has been published (Pollitt et al., 1993) and a set of papers emphasizing the follow-up was published in 1995 as a supplement to the Journal of Nutrition. In addition to presenting an overview of results, emphasis is also placed in this chapter on both the scientific and the policy significance of the results from the longitudinal and follow-up studies.

The INCAP longitudinal study (1969-1977)

Read and Habicht (1992) have reviewed the history, and Habicht and Martorell (1992) the design, of the INCAP longitudinal study which was carried out by the then existing Division of Human Development with funding from National Institute of Child Health and Development (NICHD). The purpose of the study was to test the hypothesis that malnutrition impairs physical growth and mental development. The design selected called for a high-protein supplement (atole) to be provided to two villages and a no-protein drink (fresco)

The study took place in eastern Guatemala, an area populated by Ladinos (Spanish-speaking people of mixed Spanish-Indian ancestry) in villages where agriculture was the primary source of income. Great care was taken to select the villages to be as homogenous as possible (Habicht and Martorell, 1992), although some socioeconomic differences were later shown to exist among the villages (Bergeron, 1992; Engle et al., 1992). Two pairs of villages were eventually selected: one pair of large villages and one pair of small villages. Santo Domingo and Conacaste were the most similar large villages (about 900 inhabitants each), and Espíritu Santo and San Juan were the most similar small villages (about 500 inhabitants each). Ruel et al. (1992) have shown that the anthropometric characteristics of both children and adults were strikingly similar in the different villages before 1969, when the study began. As will be shown later, growth retardation was common and severe; by three years of age, most children in the communities were below the third percentile of the World Health Organization (WHO) reference curves. Diarrheal and respiratory infections were frequent causes of morbidity. Breast-feeding was on demand, the median duration being 18 months. The diet was monotonous; beans and corn were the principal staples. Houses were generally small (one or two rooms), with adobe walls, dirt floors, and tile or tin roofs. No homes had electricity in 1969, and few had latrines; only about 10% had water piped in the home, and most obtained water from wells or, in one village, from a nearby river. Additional information about the villages is given by Habicht and Martorell (1992) and by Bergeron (1992).

Selection of atole or fresco was randomized within pairs of villages. Conacaste and San Juan (large and small) were chosen to receive atole, and Santo Domingo and Espíritu Santo to receive fresco. Atole is a hot gruel containing Incaparina (a vegetable and protein mixture developed at INCAP), skim milk, and sugar. Fresco, which was served at room temperature, is similar to Kool-Aid and contains flavoring and sugar. This is the same as the INCAPARINA supplement used in the study described in Chapter 1. The ingredients of the two drinks per cup of supplement (180 ml) are given in Table 1.

The supplements were distributed and consumed in a centrally located feeding station for 2 to 3 hours at midmorning and midafternoon, induding weekends. These times were chosen because they were the easiest times for mothers and children to attend and because they did not interfere with usual meal times.

Consumption was recorded carefully for each child under 7 years and for all women who were pregnant or lactating. Subjects were initially given one cup of supplement (180 ml). One cup of atole provided 163 kcal and 1 1.5 g of protein, whereas one cup of fresco had only 5 9 kcal and no protein. Both supplements contained vitamins and minerals (Table 1). Additional cups were provided if requested, and leftovers were measured carefully. Thus, a careful record of daily consumption was obtained.

TABLE 1 Formulas and Nutrient Content of Supplement Beverages Quantities per 180mL Cup Serving

Values shown are as of October 1971.

Curative medical care was available free of charge to the villagers on weekdays throughout the period of the study at a clinic adjacent to the feeding center. The program relied on auxiliary nurses to provide care under the close supervision of a physician, to whom difficult cases were referred. Pregnant women were immunized against tetanus and children against tuberculosis, diphtheria, whooping cough, tetanus, measles, and poliomyelitis. Additional details are given in Habicht and Martorell (1992).

In addition to data on the intake of supplement by mothers and children, longitudinal information on diet, morbidity, growth, and mental development was collected at specific ages in children from birth to seven years of age, and dietary, morbidity, anthropometric, and clinical information on mothers during pregnancy and lactation. Diets were assessed by means of 24-hour dietary recalls, and morbidity data were collected through interviews of mothers at home every two weeks. Length, weight, and several circumferences and skinfolds were collected in children. Additional details are given in Habicht and Martorell ( 1992) and in Engle et al. (1992a,b).

Technically speaking, the longitudinal study design calls for "village" to be the unit of analysis. A drawback of this approach is that it can be followed only for certain outcomes (e.g., infant mortality, physical growth) for which baseline data exist. The low power of the village-level design, which contains only two villages per treatment, makes it unfeasible as the approach to be used routinely. Instead, most analyses have used the individual as the unit of analysis but have adjusted for relevant differences among and within villages, such as differences in socioeconomic status and education. Although the subject-level analysis cannot be said to address causality in the sense that the village-level analysis does, it can be viewed as enhancing plausibility or the internal consistency of results, and hence their persuasiveness (Habicht and Martorell, 1992). For example, subject-level analyses allow one to explore whether a dose-response relationship exists between supplement and outcomes of interest. The persuasiveness of the findings is enhanced to the degree that this and other expectations are met. However, factors associated with supplementation attendance and consumption need to be taken into account, as noted by Schroeder et al. (1992).

The Nature of the Intervention

What is the nature of the intervention that took place during the INCAP longitudinal study? This discussion can be grouped into intended and unintended consequences of the longitudinal study.

The major and intended changes in the villages were better medical care and improved nutrition. The medical care included prenatal services, pediatric examinations, and vaccinations, all at no cost to families. By design, all four villages received the same quality of medical care. On the other hand, the supplementation program was meant to produce strong nutritional contrasts between atole and fresco villages. To appreciate why the supplements were formulated as described in Habicht and Martorell ( 1992), readers must recall that protein was perceived as the main limiting nutrient in the diets of developing countries during the 1960s. Thus, the formulas were chosen so as to cause striking differences between village types in protein intakes; the small amount of energy in the fresco was not viewed with concern, and the similar concentrations in the supplements of vitamins and minerals were meant to further isolate protein as the key nutritional difference in intakes between subjects receiving atole and fresco drinks. Implicit in this expectation was the need for patterns of attendance and intake to be similar in atole and fresco villages. A desired result of similarity in attendance patterns is equivalence in the degree of socialization associated with interpersonal contact in the food supplementation centers, an aspect of great relevance for the interpretation of supplementation effects on cognitive outcomes. However, attendance and consumption patterns did not prove to be the same in mothers Johnson, 1991 ) or in children (Schroeder et al., 1992). Mothers given fresco consumed about three times as much of the supplement as mothers given atole, resulting in nearly equal intakes of supplementary energy in the fresco and atole villages Johnson, 1991). In children, attendance and consumption were much lower in fresco than in atole villages in the first three years of life, so that intake not only of protein but also of energy and other nutrients was considerably greater in stole villages (Schroeder et al., 1992).

The extent of dietary improvements caused by the supplementation program is illustrated in Figures 1 and 2 for energy and protein, respectively, based on information reported by Martorell et al. (1982). From 15 to 36 months, the differences in total energy intake between atole and fresco villages were about 90 to 100 calories (Figure 1). Total energy intakes in stole villages were 11% to 12% greater than the other. Most feeding programs aimed at young children are designed to provide more than 100 calories a day; however, most fail to achieve this level of dietary impact (Beaton and Ghassemi, 1982). The relative impact of the supplementation program on protein is greater than for energy; total intakes of protein in atole villages were almost 9 g or 40% greater than in fresco villages. Analysis of data for other nutrients also reveals large differences at this age for vitamins and minerals. Thus, in contrasting total intakes in atole and fresco children, differences are evident with respect to energy, protein, and other nutrients. There is no theoretical reason to suppose that just because the percentage increase in total intakes is less for one nutrient than for another (e.g., energy versus protein as shown in Figures 1 and 2, respectively), one difference is more important than the other. Importance depends more on what is limiting in the diet.

The results from the nutrition intervention are best viewed, in the case of children, as showing the value of improving food consumption in general rather than that of any particular nutrient. In women, however, energy, rather than protein or other nutrients in the supplement, best explains the relationship between supplement intake during pregnancy and improvements in birth weight (Lechtig et al., 1975). The analyses to support this claim are made possible by the overlapping ranges in supplement energy intakes during pregnancy in stole and fresco villages.

FIGURE 1 Energy Intake of Children 15 to 36 Months from Home Diet and Supplement

There are potentially many unintended effects of the design. The continued presence over eight years of a team of outsiders in small, traditional villages must have had an impact on attitudes and even practices about health, nutrition, and child rearing, among others. These effects are probably important but remain unmeasurable. Because data collection activities were similar in all four villages and because INCAP personnel were routinely rotated among all villages, there is the expectation that the effects, whatever their nature, were similar in atole and fresco villages. There was no deliberate attempt at social stimulation, but the children in all four villages were exposed to the increased activities of the program unlike those in other similar Guatemalan communities.

Effects in Early Childhood

The nutrition intervention benefited children in many ways. Supplementation during pregnancy improved birth weights; women who ingested more than 20,000 kcal from the supplements during pregnancy had half the risk of delivering a low birth weight baby (<2,500 g) compared to those who ingested less than 20,000 kcal (Lechtig et al., 1975). Infant mortality rates were markedly reduced; compared to 1949-1968 rates, infant mortality from 1969 to 1977 declined by 66% in atole villages compared to 24% and 19% in fresco and nonintervened (i.e., control) villages, respectively (Rose et al., 1992). Although the number of days children were ill with diarrhea was not reduced by the nutrition intervention (Rose and Martorell, 1992), the atole did protect against the negative effects of diarrhea on growth (Martorell et al., 1990b). The atole was also found to promote speedy recovery from wasting (Rivera et al., 1991). Although the effects of the atole on growth in children were found to be important (Martorell et al., 1982), these effects were confined to the first three years of age (Martorell and Klein, 1980; Schroeder et al., 1995). Specifically, atole intake was not related to growth from three to seven years of age. Another improvement associated with the nutrition intervention was enhanced motor development (Lasky et al., 1981). Finally, the atole had only a minor effect on mental development, certainly much less than anticipated (Engle et al., 1992b) although statistically significant.

FIGURE 2 Protein Intake of Children 15 to 36 Months from Home Diet and Suplement

Martorell et al. (1982) have presented the results for the impact of the nutrition intervention on physical growth in three-year-old children. These results are given in Figure 3, which shows the percentage of children who were severely stunted, defined as less than three standard deviations (SD) below the NCHS mean, by supplement type and calendar time. Values for height at three years of age from the WHO/NCHS reference population were used. These correspond to the National Center for Health Statistics component of the reference data (22 years) and not to the Fels component (<3 years). One centimeter was added to the cutoff point corresponding to 3 SD to adjust for the fact that length, and not height, was measured in Guatemala. The adjusted values used as cutoff points were 84.5 cm for boys and 83.8 cm for girls. The prevalence of severe stunting was extremely high, around 45%, but similar in atole and fresco villages when the study began in 1969. Over the course of the study, the prevalence was reduced by half in atole villages but stayed at about the same level in fresco villages.

A logistic regression model, with the village as the unit of analysis, was used to test the time trends shown in Figure 3. The model that was found to best fit the data included severe stunting as the dependent variable (1 = severely stunted, 0 = others) and the following independent variables: treatment, sex, year and treatment * sex, treatment * year and sex * year interaction terms (Table 2). The analysis indicates that atole villages had less severe stunting than fresco villages; the significant interaction terms reveal that there was a significant decline in stunting in the atole villages and that this decline was most pronounced in girls.

TABLE 2 Logistic Regression Results of Trends in Severe Stunting^a by Treatment Type, Sex and Year (model chi-square=49.96, df=6)

^a Defined as <3 S.D.s below the NCHS median (see text)
^b Treatment=Atole=1, Fresco=0
^c Sex=males= 1, females=0
^d Year is expressed continuously as 5 values: 69,70.5,72.5,74.5 and 76.5. The values are the average for each category.

FIGURE 3 Change Over Time in Percent of 3-Year-Olds With Severe Growth Failure: Sexes Combinated* (* x=3.70, p=.06)

The important improvement shown above occurred in the absence of changes in diarrhea! diseases (Rose and Martorell, 1992). Had both infection and diet been altered, the expected effect would likely have been greater. Interestingly, growth rates from 5 years of age until adulthood in the Guatemalan study population were not very different from those observed in well-to-do populations (Martorell et al., 1990a). Certainly, later childhood and adolescence cannot be said to be stages in life when marked growth failure occurs. Rather, the first three years are when growth retardation in Guatemala is intense and when nutritional interventions can have the greatest impact on growth.

Guatemalan follow-up study (1987-1988)

The follow-up study is unique (Martorell and Rivera, 1992). There have been some follow-up studies in developing countries, but these have focused on very specific groups, such as survivors of severe malnutrition; long-term follow-up studies of nutrition interventions have never been carried out, to our knowledge. Another distinguishing feature of the follow-up study is its comprehensive nature: no study in developing countries has included the range of measures of human function that were included in the follow-up.

The follow-up study permits one to ask whether the benefits found in early childhood persist into adulthood. A novel contribution of the study is that it allows examination of functional effects that can only be measured later in life, extending, in effect, the horizon for evaluating nutrition interventions. Its central hypothesis is as follows: Better nutrition during early childhood leads to adults with a greater potential for leading healthy productive lives.

However, most subjects were adolescents when the data were collected in 1988. The issue of when it was best to carry out the study was debated intensely among the research members, and though it was agreed that it would be better to measure the subjects as adults, it was felt that the opportunity to carry out the study might not present itself later. It was also recognized that much would be learned about adolescence per se by going ahead with the study. This was seen as a significant contribution, since much less is known from nutrition studies in developing countries about adolescents than about young children.

The use of the word "potential" in the hypothesis was deliberate. It was realized that productivity in an economic sense could not be adequately assessed in a young population. This could only come later when the subjects formed independent households and settled into their occupations. Similarly, other functions, such as parenting, could only be measured later when they formed families.

Overview of Key Results from the Follow-up Study

Data were collected on many functional domains (Rivera et al., 1992), but in this brief overview of results, only three aspects are emphasized: body size and composition, work capacity, and intellectual performance. All three areas are very important, and improvements in one or all would be seen as contributing to human capital formation.

Three aspects of the body size and composition results stand out (Rivera et al., 1995). First, adolescents who were exposed to the atole during the first three years of life were taller and had greater fat-free masses than those who received fresco. However, there was some attenuation of the effects observed at age three. In other words, there was a small degree of catching up in fresco villages with respect to atole villages. It is interesting that the anthropometric effects were greatest in females. The cutoff point of less than 149 cm, equivalent to a height of 4 ft 11 in, is often used as a criterion of obstetric risk in women (Krasovec, 1991). Of those females over 16 years of age who had been exposed to the supplements from birth to three years of age, 49% of the fresco subjects had very short stature compared to 34% of atole subjects (Figure 4). Little or no growth in height occurs after 16 years (Martorell et al., 1995). Similar findings are obtained when the analysis is restricted to women more than 18 years old (fresco, 12/27 or 44%; atole, 9/32 or 28%).

Differences in fat-free mass also stand out. Females from atole villages had 2.1 kg more fat-free mass than females from fresco villages. These differences are equivalent to an effect size of about 0.5, that is, equal to a positive shift of 0.5 SD. Cohen (1977), author of a popular book on statistical power, labels an effect of this magnitude as medium.

Work capacity was significantly improved in subjects exposed to the supplements in their first three years of life, but only in males (Haas et al., 1995). This effect was large. Atole males had maximal oxygen consumptions (VO2max) that were 0.3l/min greater than those of fresco males. The difference is equivalent to about 0.7 SD, approaching what Cohen (1977) would call a large effect size. Another interesting finding is that the larger working capacity of atole males could not be explained by differences in fat-free mass (i.e., VO2max/kg fat-free mass was still greater in atole villages). The basis for these qualitative tissue differences between atole and fresco subjects is unclear.

A feature of all analyses carried out to date with respect to measures of intellectual performance is that they control for schooling variables (Pollitt et al., 1993) because the villages had differed in patterns of school attendance since before the beginning of the study (Habicht and Martorell, 1992; Bergeron, 1992). One of the fascinating discoveries is that intellectual performance was more affected in adolescence among subjects exposed to the supplements during pregnancy and the first two years of life than among those exposed to a supplement later in life. The behavioral analyses have focused on a different sample than the biological analyses (exposure during pregnancy and the first two years of life vs. exposure during the first three years of life, respectively). This is due to different perceptions about vulnerability in each area and to the fact that effects of atole on growth were found only in children less than three years of age (Martorell and Klein, 1980). The atole-fresco differences found in children were less than 0.2 SD, compared to differences of around 0.6 SD found in adolescence using a summary variable of intellectual performance (i.e., a factor score that combines literacy, numeracy, general knowledge, Raven's Progressive Matrices, reading, and vocabulary). To use Cohen's labels, the effects found in children can be described as small, while those found in adolescents can be called medium to large. There are also strong indications that the effects in adolescence are found only in those cohorts exposed to supplementation during pregnancy and the first two years of life. Examination of the subcomponents making up the summary variable shows that effects were found in four of six tests. Effects were found in both males and females.

Potential Significance of the Results of the Follow-up Study

First, the greater body size and increased fat-free mass found in females would be expected to impact positively on reproductive fitness (Krasovec and Anderson, 1991). Short stature is a risk factor for cephalopelvic disproportion, delivery complications, and maternal obstetric mortality. Also, greater fat-free mass has been found to lead to higher birth weights.

The improved working capacity in males might result in increased productivity in men engaged in hard physical labor. The literature clearly supports this expectation (Spurr, 1983).

Finally, sharp minds are valued by all societies and by parents everywhere in recognition that improvements in intellectual performance are bound to improve the capacity of individuals to function in a variety of settings. Consider two suggestions. One is that such improvements might lead to better employment opportunities and greater earnings. Another is that intellectually better-endowed adults will be better parents because they will be better providers as well as being able to meet the developmental needs of their children.

Contents - Previous - Next