 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
* Division of Biological and Biomedical Sciences, Nutrition and Health Sciences Program
Department of Biostatistics
Department of International Health, Rollins School of Public Health, Emory University, Atlanta, Georgia
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Methods. The study sample comprised 130 female singletons exposed to either Atole (53%, 91 kcal and 6.4 g protein/100 mL) or Fresco (47%, 33 kcal/100 mL, no protein) during the prenatal period and the first 2 years of life. EA was assessed at the ages of 22 to 29 years by knowledge, numeracy, and several reading tests. A summary measure of EA was computed based on 5 tests, and outcome variables were categorized into quintiles. Analysis was based on a proportional odds model. Generalized estimating equations were used to account for sibling clustering.
Results. Overall, 36.2% of women completed primary school. Women exposed to Atole had better EA than those exposed to Fresco (odds ratio [OR]: 2.8; 95% confidence interval [CI], 1.4, 5.4), with a significant treatment-by-schooling interaction. Atole was not associated with EA (OR: 1.5; 95% CI: 0.7, 3.2) among women who did not complete primary school, whereas among those who completed primary school, Atole was associated with improved EA (OR: 13.7; 95% CI: 3.7, 50.8).
Conclusions. We conclude that better nutrition during early childhood improved adult EA, but only among children who completed primary school.
Key Words: early childhood nutrition • supplementation • schooling • adulthood • educational achievement • longitudinal study • Guatemala
Abbreviations: EA, educational achievement • INCAP, Institute of Nutrition of Central America and Panama • SES, socioeconomic status • TRT, treatment • GRADE, schooling • CI, confidence interval • OR, odds ratio • GA, gestational age
Child malnutrition remains a major public health problem in developing countries.1,2 Early childhood, including the prenatal and early postnatal (birth to 2 years) periods, is considered the most sensitive period for development, during which the developing brain is especially vulnerable to nutritional insult.3–5 Early childhood undernutrition is associated with delayed cognitive development and poor school performance.6–9 In communities where undernutrition is endemic, supplementary feeding that meets nutritional requirements of children during the first 2 years of life may help prevent cognitive delays.10–12
Intelligence and educational achievement predict occupation and job performance and may be related to later success and lifetime earnings.13–15 Therefore, promotion of adequate nutrition during early childhood, in that it may improve intelligence and academic achievement, may help build human capital and have long-term economic implications for society.16,17 However, few studies have been conducted to understand the long-term consequences of early childhood undernutrition on adult intelligence and educational achievement.
Previous results from our study in Guatemala showed that supplementation with a nutritious supplement during the prenatal and the first 2 years of life significantly improved educational achievement, and, to a lesser extent, intelligence and information processing, at ages 13 to 19 years in both boys and girls.18,19 Some of these children were still in school at the time of the follow-up, 1988–1989. We conducted a second follow-up of the women in this cohort in 1996–1999, nearly a decade later, and repeated the assessment of educational achievement. The main objective of this analysis was to assess the benefits of childhood supplementation on adult educational achievement. Furthermore, given the well-established relationship between schooling and educational performance,15,20 we examined whether the effect of nutritional supplementation on educational achievement was modified by schooling.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The first study was a longitudinal, community-based, food supplement trial conducted by the Institute of Nutrition of Central America and Panama (INCAP) between 1969 and 1977. Four villages, stratified by size (2 large and 2 small), were randomized to receive either Atole or Fresco supplement. Atole was a high-energy, high-protein supplement, which contained 91 kcal of energy and 6.4 g of protein/100 mL, whereas Fresco contained 33 kcal of energy/100 mL and no protein. Both Atole and Fresco were fortified with vitamins and minerals. The target population of the intervention was pregnant or lactating women and children up to 7 years old. Both supplements were available twice daily, year-round, and consumption was ad libitum. Additional details about the original intervention study are described elsewhere.21
During 1988–1989, a follow-up study was conducted in the same 4 villages on subjects 13 to 19 years of age. Effects of early childhood supplementation on EA based on this follow-up study have been reported elsewhere.18,19 From 1996 to 1999, a second longitudinal study was conducted by Emory University and INCAP; its purpose was to assess effects of early childhood malnutrition in females on birth weight and postnatal growth and development of their children. Appropriate clearance for the ethical conduct of our research was obtained from the Institutional Review Boards of Emory University and INCAP. Men (fathers) were not included in this study of intergenerational effects and data collection was restricted to village residents (ie, migrants were not studied). Pregnancies were ascertained through intensive surveillance and were followed through delivery. Children then participated in a longitudinal study of growth and development until they reached 36 months. Considerable data were also collected on mothers, including schooling and educational achievement, which was assessed by a battery of educational tests consisting of a preliteracy test, tests of general knowledge, numeracy, reading, and two standard educational achievement tests. Identical procedures were used as in the 1988–1989 follow-up study.
We restricted our analysis to women who took the educational tests in the 1996–1999 study and who were exposed to nutrition supplements during the prenatal period and the first 2 years of life in the original longitudinal study between 1969 and 1977. This group, born between 1970 and 1974, was selected because the 1988–1989 study showed greater and more consistent effects of treatment on the cohort of children exposed during pregnancy and the first 2 years of life than on the cohort exposed after 2 years of life.18 Because the 1996–1999 study did not include out-migrants and the educational tests were administered only to those resident women who had at least a child (<3 years) during its duration, of 419 female singletons born between 1970 and 1974 in the original cohort, it was possible to include only 143 (34%) subjects. Comparisons of participants (n = 143) and nonparticipants (n = 276) indicated no differences (data not shown) in several childhood characteristics, including childhood socioeconomic status (SES), type of supplement received, average supplement intake during the first 2 years, level of stunting at 2 years, percent time ill during 0 to 1 year and 1 to 2 years, home dietary intake at 2 years, and paternal literacy. The proportion of subjects from Atole villages was also similar in participants and nonparticipants (52% vs 55%, respectively). Of 143 eligible cohort members, 13 subjects who had missing data on EA (n = 1), schooling (n = 1), or SES (n = 11) were excluded. The final sample for analyses was 130 women (67 from Atole villages and 63 from Fresco villages).
Data Collection and Variables
Treatment (TRT)
All subjects were categorized into 1 of the 2 TRT groups, Atole or Fresco.
The Psychological Test Battery
An educational battery, used previously in the 1988–1989 follow-up study, was administered. The Raven’s test of intelligence and information processing tests, used in the 1988–1989 follow-up study, were not repeated in the 1996–1999 study. The data collection team consisted of 1 supervisor and 5 field workers, using procedures similar to those described by Pollitt et al.18 The educational battery included tests of general knowledge related to daily life experiences, numeracy (numbers and simple calculations), a preliteracy test (letters, words, and phrases), a reading test of an article from a local newspaper, and 2 educational achievement tests, which were part of the Interamerican Series originally designed to assess reading abilities of Spanish-speaking children in Texas.22 All women took the tests on general knowledge and numeracy, regardless of their schooling status. Subjects who reported having achieved fewer than 4 years of schooling or those who reported 4 to 6 years of schooling but could not read the headline of a local newspaper article correctly were given a preliteracy test. Subjects who reported >6 years of schooling were presumed to be literate. The reading test and the two educational achievement tests on levels of comprehension and vocabulary were administered to women who were considered to be literate or who passed the preliteracy test (with <5 errors). Further details of the tests are provided elsewhere.18
A brief description of the tests and the variables are shown in Table 1. Five outcome variables, each representing a single test, were selected for analysis. First, for each test, a score was assigned to each question according to the nature and difficulty of the question, and the question scores were summed. Specifically, 1 point was assigned to each question in the tests on general knowledge, reading, and educational achievement. A score of "zero" was assigned to subjects who did not pass the preliteracy test and, hence, did not take the reading and the educational achievement tests. Thus, our analyses differ from those of Pollitt et al18 in that we include illiterate subjects. We also differ in terms of the coding of the numeracy test and in the generation of a summary measure of educational achievement. Because the level of difficulty of the questions gradually increased in the test of numeracy, factor analysis was performed to guide the computation of an overall numeracy score. One factor explained 61% of the variance (whereas the second and third factors each contributed <12% of the variance), with equal weights on all 6 subsections of the numeracy test (factor loadings ranged from 0.76 to 0.80; data not shown). We therefore weighted these subsections equally by assigning 10 points to each subsection, which had anywhere from 5 to 11 questions, and calculated a summary score for the numeracy test based on the sum of these 6 subsections. Thus, the maximum score for the numeracy test was 60 points. We found that scores on the 5 tests were strongly correlated. This led us to compute a summary measure of EA based on the sum of the 5 tests (general knowledge, numeracy, reading, level of comprehension, and vocabulary), where a score with a maximum of 20 points was assigned to each test (ie, equal weighting to each test). Factor analysis indicated that a single factor with equal loadings on these 5 tests (factor loadings ranged from 0.70 to 0.84) captured 63% of the variance (whereas the second and third factors each contributed <14% of the variance). The summary measure of EA, derived empirically, proved useful as a reliable and robust global measure of educational achievement. Because of the skewness of the data, all outcome variables were categorized into quintiles (ie, 
26 subjects per quintile).
|
|
Socioeconomic Indicators
A composite score for socioeconomic status during early childhood (SES) was constructed based on characteristics of home and possessions, mother’s education, and father’s occupational status,18 with low scores representing lower SES.
Schooling
All women had finished school when the 1996–1999 longitudinal study was conducted. The maximum grade achieved (GRADE) was used to indicate schooling, which was categorized as "completed primary school" (GRADE 
6 years) or "not completed" (GRADE <6 years). Examination of the effect of GRADE on educational achievement indicated a strong threshold effect at 6 years of schooling. Approximately one third of the subjects completed primary school.
Analytical Strategies
All analyses were conducted using SAS software (Version 8.2; SAS Institute Inc, Cary, NC). Analyses based on a cumulative logit model with proportional odds assumption, an extension of logistic regression for ordinal outcomes,23 were conducted to investigate the effects of early childhood nutritional supplementation on adult EA. All outcome variables have 5 ordinal levels based on quintiles, with "5" representing the highest level. The proportional odds model provides estimates of the covariant effects on EA based on the cumulative odds of "5" versus "1, 2, 3, or 4," "4 or 5" versus "1, 2, or 3," "3, 4, or 5" versus "1 or 2," or "2, 3, 4, or 5" versus "1." Because some (n = 15) subjects were siblings, a generalized estimating equations approach24 was used to account for clustering effects by means of the GENMOD procedure of SAS.
Four models were analyzed. We first controlled for age (model 1), then age and SES (model 2). Schooling (GRADE, "completed primary school" or "not completed") was also added (model 3). Finally, 2 interaction terms, TRT x SES and TRT x GRADE, were entered additionally (model 4) to assess whether the TRT effects were modified by SES or schooling. If the interaction was significant (P < .15), the treatment effects were compared within strata of SES or schooling. Dummy variables representing 4 of the 5 testers were added to the models to assess tester effects on relationships with treatment, SES, and schooling.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Characteristics of the study participants are summarized in Table 3. Women were 22 to 29 years of age when they took the educational tests. Age at follow-up and SES were not significantly different by treatment or village size. Overall, paternal literacy was 60.2%, with the highest level reported for the small Fresco village. The small Fresco village also had the greatest proportion of women who completed primary school, suggesting an engrained village tradition. Overall, 36.2% of all women completed primary school. The percentage of subjects in the highest 2 quintiles is also shown for each psychoeducational test and for the overall EA.
|
|
|
6 years vs <6 years) and the differential effects of schooling on EA by treatment (Fig 1). Atole was not associated with EA (cumulative OR: 1.5; 95% CI: 0.7, 3.2) among women who did not complete primary school. However, among those who completed primary school, Atole was associated with improved EA (OR: 13.7; 95% CI: 3.7, 50.8). We also present the effects of schooling on adult EA by treatment. Women who completed primary school (GRADE 6 years) performed significantly better than those who did not complete primary school (GRADE <6 years) in both Atole and Fresco villages. However, the schooling effects on EA were magnified among women who received Atole during early childhood. The cumulative ORs for schooling effects were 9.0 (95% CI: 3.7, 22.2) in Fresco villages and 85.5 (95% CI: 22.3, 327.1) in Atole villages.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Furthermore, schooling was a strong predictor of EA. Our analyses showed that both supplementation and schooling were independent predictors of women’s educational achievement. However, the effects of schooling were of greater magnitude. Furthermore, we found significant interactions between treatment and schooling, indicating that supplementation modified the impact of schooling and vice versa, which is consistent with the previous study.18,19 However, the schooling effects observed in the current study (1996–1999) seem to be more robust. Results from the previous follow-up demonstrated significant positive effects of schooling on test performance among subjects exposed to Atole but not among Fresco subjects.18,19 The current analyses, on the other hand, showed that women who completed primary school had significantly better educational achievement than those who did not, regardless of supplement type. The more consistent effects of schooling observed could be attributable to the fact that the subjects were adults who had already finished school when they were tested in 1996–1999; previously, in 1988–1989, they were adolescents and some were still in school. We also included subjects with no schooling in the analyses, increasing variability.
The 1988–1989 follow-up study also showed a significant interaction between treatment and SES. Among those with lower SES, subjects exposed to Atole performed significantly better than those exposed to Fresco, whereas no differences in educational achievement were found between Atole and Fresco subjects with higher SES. However, this TRT-by-SES interaction was not significant (P > .15) in the current analyses. We have no explanation for the lack of concordance with the previous analyses. Participants and nonparticipants in the analysis had similar SES scores and were similar in other characteristics examined, except EA in 1988–1989, which was lower among participants. The median EA was 62.6 among participants and 66.4 among nonparticipants (P < .02). These differences are small relative to the effects of treatment and schooling (Table 5) and seem unlikely to bias our results. SES effects were less robust than those of education in our analyses than in the previous ones. Perhaps, SES effects on educational achievement waned from adolescence to adulthood whereas those of education increased. Finally, low statistical power, because of the reduced sample sizes attributable to the exclusion of men, migrants, and women without children <3 years of age in 1996–1999 might be considered as an explanation. However, this is not likely to be the sole explanation, because we observed clear main effects and interactions for education with our sample size of 130 women.
Our study was conducted in Guatemala, a developing country, with a high prevalence of growth retardation among young children, indicating high levels of malnutrition.25 To our knowledge, ours is the first prospective study to examine the long-term effects of early childhood nutritional supplementation on women’s educational achievement. We used a very reliable battery of educational achievement tests. The test-retest correlation over a decade was 0.86 for the summary measure of educational attainment.
In addition to extending the follow-up to adulthood and having completed schooling information, we also improved on the analytic methods used by Pollitt and colleagues.18,19 Several analytical approaches were implemented. First, we used the generalized estimating equations approach to account for sibling clustering. Second, we categorized the outcome variables to 5 ordinal levels to take into consideration the skewness of the data. Then, a cumulative logit model with proportional odds assumption was implemented. Finally, in addition to assessing the effects of early childhood supplementation on each dimension of EA by each individual test, we further computed a summary measure of EA based on the 5 psychoeducational tests administered between 1996 and 1999. EA was calculated by weighting all tests equally, which was supported by factor analysis.
Pollitt and colleagues18,19 excluded subjects who failed the preliteracy test from the analyses of several of the tests: reading, level of comprehension, and vocabulary. We included these subjects in the lowest quintile by assigning them a score of "zero" to the test that they failed to take. This was an appropriate step because all subjects were given an opportunity to take all tests by qualifying through a preliteracy test. The preliteracy test assessed whether a person could recognize words, phrases, or sentences, which are the basis for the reading and achievement tests. Because all women took the tests of general knowledge and numeracy, regardless of literacy, the performance on these two tests were contrasted for those who did and did not take the reading and/or the achievement tests. We found that most of the women who did not take the reading and/or the achievement tests were categorized in the lowest quintile for tests on general knowledge and numeracy, which confirmed our decision to assign them to the lowest quintile for the overall EA. Therefore, it is unlikely that our decision would result in misclassification bias.
The previous and current analyses controlled for schooling, which varied by village and reflected differences already existing in 1969, when the first longitudinal study began. Similar to the findings of a previous study,26 the small Fresco village, which had the highest level of paternal literacy in 1969, 80%, had the highest percentage of women who completed primary school by 1996–1999, 67% (Table 3); the large Atole village had the lowest levels for these 2 indicators, 54% and 15%, respectively; and the other 2 villages were intermediate in terms of the paternal literacy in 1969 and the education attainment in 1996–1999. Thus, schooling needs to be adjusted for in examining the effects of supplementation on educational achievement because of the existence of village-level differences in levels of schooling that predate the supplementation experiment and which have persisted through time.
The key limitation of our analysis is that we included only a third of the cohort of females because the 1996–1999 study was restricted to village female residents, who had to have at least one child <3 years of age in 1996–1999. However, there is no evidence that our sample is biased. There were no differences found between participants and nonparticipants in childhood characteristics, including SES, supplement type and intake, home diet, stunting status at 2 years old, and paternal literacy. However, as noted earlier, participants had lower educational achievement, but these differences were small. Although men and migrants were not studied, we would not expect conclusions to be different because the previous analyses indicated that, although males outperformed females on some tests (numeracy and reading), supplement or schooling effects did not differ by gender or immigration status.18,19
Another limitation is that we studied only educational achievement. The earlier analyses18,19 included 2 additional tests: the Raven’s Progressive Matrices (a test of intelligence) and tests of information processing (simple, choice, and memory reaction time). These tests were not included, in part, for budgetary reasons and also because the previous study found that the educational tests we used were more sensitive to nutritional effects than the Raven’s or information processing tests.
Results from the current study indicate that better nutrition during early childhood improves women’s EA. Thus, investment in interventions aimed at improving early childhood nutrition in children can be expected not only to decrease the prevalence of stunting, as the Atole intervention was shown to have done,27,28 but also to prevent its negative functional consequences throughout the life cycle. In addition, we provide strong evidence that completing primary school has a pronounced effect on women’s educational achievement, stronger than that of a nutritious supplement. However, because supplementation magnified considerably the effect of schooling, the key implication for policymakers is that both early childhood nutrition interventions and schooling opportunities are necessary for optimal enhancement of EA, an important dimension of human capital.
|
ACKNOWLEDGMENTS |
|---|
We thank the Guatemalan participants in this study for their cooperation and Morgen Hickey for her contribution to the data management.
|
FOOTNOTES |
|---|
Reprint requests to (R.M.) Department of International Health, Rollins School of Public Health, Emory University, 1518 Clifton Rd, NE, Atlanta, GA 30322. E-mail: rmart77{at}sph.emory.edu
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  A. D. Stein, M. Wang, A. DiGirolamo, R. Grajeda, U. Ramakrishnan, M. Ramirez-Zea, K. Yount, and R. Martorell Nutritional Supplementation in Early Childhood, Schooling, and Intellectual Functioning in Adulthood: A Prospective Study in Guatemala Arch Pediatr Adolesc Med, July 1, 2008; 162(7): 612 - 618. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Watanabe, R. Flores, J. Fujiwara, and L. T. H. Tran Early Childhood Development Interventions and Cognitive Development of Young Children in Rural Vietnam J. Nutr., August 1, 2005; 135(8): 1918 - 1925. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. D. Stein, H. X. Barnhart, M. Wang, M. B. Hoshen, K. Ologoudou, U. Ramakrishnan, R. Grajeda, M. Ramirez-Zea, and R. Martorell Comparison of Linear Growth Patterns in the First Three Years of Life Across Two Generations in Guatemala Pediatrics, March 1, 2004; 113(3): e270 - 275. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2003 American Academy of Pediatrics. All rights reserved. |
||
|
||
