![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
Department of Psychological Sciences, Purdue University, W. Lafayette, IN;
* Instituto de Investigacion Nutricional, Lima-18, Peru;
Grupo de Análisis para el Desarrollo (GRADE), Lima, Peru; and
** Nutrition Unit, Pan-American Health Organization, Washington, DC
3To whom correspondence should be addressed. E-mail: wachs{at}psych.purdue.edu.
 |
ABSTRACT |
|---|
 TOP ABSTRACT SUBJECTS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
KEY WORDS: • education • intelligence • diet • physical growth • iron
It has been estimated that in developing countries, more than one third of all children show signs indicative of severe malnutrition [e.g., weight-for-height 2 SD below the expected average (1)]. These figures do not take into account children who have clinical deficiencies in specific micronutrients such as iodine, vitamin A, or iron [e.g., 51% of children < 4 y old have iron deficiency anemia (1)]. Research has consistently shown how a variety of cognitive and noncognitive developmental outcomes are adversely affected by moderate-to-severe protein energy malnutrition (2), by specific micronutrient deficits such as prenatal iodine deficiency or iron deficiency anemia (3), and by chronic undernutrition, even when a child is not clinically malnourished (4,5).
Given that nutritional deficiencies are a major developmental risk factor, it is important to understand why so many children are consuming diets that are deficient in energy and/or specific nutrients. Traditionally, explanatory models of the etiology of childhood nutritional deficiencies have focused primarily on the scarcity of food and economic disadvantage as causal influences (6–8). However, the assumption that children’s nutritional deficiencies are essentially due to overall food scarcity or to a lack of family resources to purchase food has been increasingly questioned (9–11). Substantial numbers of poorly nourished children live in geographic regions in which food production is relatively high (12), or in families that are not economically disadvantaged (13). Meaningful variability in children’s nutritional status has been documented among low-income families living in the same household, neighborhood, or district (14–17).
What this literature suggests is that variability in the adequacy of children’s diets may reflect a variety of other influences in addition to economics and food availability. One such alternative influence is morbidity (1), which may act to reduce children’s appetite (3). A second alternative influence is cultural beliefs about feeding (18–21), particularly at times when children are ill. A third potential influence is individual parental characteristics (22). Two parental characteristics are the focus of the present study: maternal education level and maternal intelligence.
Maternal education is defined by the years of schooling achieved by the mother in the local school system. In a number of recent studies involving infants and preschool children from developing countries, higher levels of maternal education were linked to longer or more exclusive breast-feeding (23–27), better physical growth (28–30), and greater intake of both protein and developmentally critical micronutrients (31). However, little is known about the processes through which more schooling for mothers translates into better nutrition for their offspring. Available evidence supports a link between higher levels of schooling and higher levels of intellectual performance, both in developed (32) and in developing countries (33–35). Given that intelligence is traditionally conceptualized in terms of the individual’s ability to both modify and adapt to their environment (36,37), more education for women may promote higher levels of maternal intelligence. Higher levels of maternal intelligence may, in turn, facilitate increased usage of adaptive behavior strategies that could promote better offspring health and nutrition.
Currently, there are only a few studies from developing countries that have related maternal intelligence to indices of offspring nutrition. In studies done in Jamaica (38) and Nicaragua (33), better physical growth of offspring was found for children whose mothers were more intelligent. However, the results from Nicaragua were not significant when controlled for social class or maternal literacy, whereas the results from Jamaica were retrospective, leading to ambiguity about directionality. In a study done in Egypt, results indicated that toddlers whose mothers had higher levels of intelligence had higher-quality diets, even after controlling for family socioeconomic status (SES)4 (31). However, the sample size in the Egyptian study was relatively small (n = 76), leaving it unclear whether these results could be replicated with a larger sample.
In the present study, we examined the relation of maternal educational level and intelligence to measures of infant and toddler diet, anthropometry, and iron status in a low-income population in Lima, Peru. Infant diet and anthropometry were chosen as outcome variables based on the usage of these measures in previous research. We also chose to focus on iron status given evidence for both the high levels of iron deficiency in many developing countries (1) and repeated findings documenting adverse developmental consequences for iron-deficient infants (3,39). On the basis of previous research, we predicted more adequate nutrition for offspring of more educated women, even after controlling for social class. The critical issues to be addressed in our research were the following: 1) the relation of maternal intelligence to offspring nutrition; and 2) the degree to which maternal intelligence was uniquely related to the nutrition of offspring, over and above the contributions of maternal education and family social class.
|
SUBJECTS AND METHODS |
|---|
|
TOPABSTRACTSUBJECTS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
This study was conducted in the periurban shantytown of Canto Grande, which is an area within the district of San Juan de Lurigancho, which in turn is part of greater Lima, Peru. The rates of unemployment and underemployment for the overall population in Canto Grande were considered high according to national census criteria. Our initial sample consisted of 249 newborns (47.4% boys), only 2 of which were preterm. These 2 infants were used in our analyses, but were assessed at gestational rather then chronological age to correct for the degree of preterm birth. Neither was a statistical outlier with regard to iron status (online supporting material). At 3 mo of age, 246 children remained in the study; 245 remained at 6 mo, and 241 at 12 mo. Because our funding was depleted, we were able to assess only 104 of the children in our sample at 18 mo. Children assessed at 18 mo were those who reached that age while funding was still available.
Measures
Maternal and family measures. Pregnant women were enrolled in the study at the end of the 1st trimester of pregnancy. Within 1 mo after the mother was first enrolled, a project social worker and a nutritional field worker visited the family home. During this initial visit, the nutrition field worker measured maternal height and weight. The social worker assessed the level of maternal education based upon maternal report of number of school grades completed, and also administered a culturally relevant SES inventory developed at the Instituto de Investigación Nutricional (IIN) in Lima. Two measures of family SES were derived from this instrument: 1) the number and type of family household possessions, and 2) a factor score assessing the level of home quality (construction material of the house and home services, e.g., water, electricity, and sewage). During the 3rd trimester of pregnancy, mothers were administered the child form of the Raven Progressive Matrices and the Spanish version of the Vocabulary subtest of the Thurstone Primary Mental Aptitudes scale by a project psychologist. The child form was used because of significant floor effects detected in our pilot testing when the adult form was used. Because the Raven and the Vocabulary scores were significantly correlated with each other (r = 0.57, P < 0.01), we standardized and summed the 2 scores to derive a single score for maternal intelligence. Details on family demographics and maternal characteristics are available as Online Supporting Material with the online posting of this paper at www.nutrition.org (Supplemental Table 1).
|
Because of collinearity among our dietary intake variables and among our dietary factor scores, we chose to compute a cumulative dietary risk score for each infant (42). For each nutrient assessed, we determined whether the infant’s intake was <80% of current recommended intake (RI) from complementary foods for infants of a given age (43–45). One difficulty in assigning RIs for energy and nutrients from complementary foods is that these vary with the amount of breast milk consumed, especially at 6 mo when the variability in breast milk intake, and thus the amount from complementary food, is very large. Consequently, at 6 mo, we adjusted the RIs for energy and nutrients according to our estimation of the level of the infant’s intake of breast milk. Recommended intakes for energy and specific nutrients were estimated according to breast milk intake [below (–1 SD), above (+1 SD), or at the mean]; the degree to which the infant’s intake matched these was calculated (see online supporting material). Infants whose intakes fell below the 80% cutoff value for a given nutrient were given a score of 1 for that nutrient, whereas infants whose scores were above the cutoff value were given a score of 0. Scores were summed with a possible range from 0 to 11, with higher scores indicating less adequate nutritional intake (higher cumulative dietary risk). As a test of the validity of the dietary risk index, we also computed a measure of nutrient density adequacy for each nutrient assessed (43). The mean nutrient density adequacy for the 10 nutrients assessed (not including energy) was correlated with the child’s dietary risk score at each age. Results indicated that the correlations were significant and in the expected direction (for details, see online supporting material).
At 3, 6, 12, and 18 mo, child length was measured by trained IIN nutritional field staff with the child in a recumbent position using a Franklin Plane length board. Child weight was measured with a calibrated spring balance scale at 3 and 6 mo and a calibrated clock balance scale at 12 and 18 mo. Measurements of triceps and subscapular skinfold thickness were obtained using a Lange Skinfold Caliper (Beta Technology). When infants were 12 and 18 mo old, 4 mL of venous blood was drawn early in the morning or before the midday meal by trained pediatric nurses. Blood samples were transported to IIN on the day they were drawn; the samples were analyzed for hemoglobin, serum iron, total iron binding capacity (TIBC), and C-reactive protein. Hemoglobin was analyzed by the colorimetric method (HemogloWiener kit produced by Wiener Laboratories). Serum iron and TIBC were analyzed by the colorimetric method (total iron and unsaturated iron binding capacity kit produced by Sigma, Catalog # 565-B). Individual levels of transferrin saturation were derived using the standard formula of serum iron/TIBC x 100. C-Reactive protein was analyzed by the radial immunodifusion method (Human C-reactive Protein ’EL’ NANAORID kit produced by The Binding Site).
This study was approved by the Ethics Committee of the IIN, Lima, Peru and the Committee on the Use of Human Research Subjects, Purdue University. Guidelines used by these committees include the principles espoused in the Helsinki Declaration and the WHO operational guidelines. Families who participated were informed of the study protocol and signed consent was obtained.
Statistical methods
Although statistical significance was defined by the standard P < 0.05 level, to minimize the likelihood of a type II error, we also considered P-values > 0.05 and < 0.10 as worth interpreting. In our primary analyses, we followed a 2-stage analytic strategy. In the first stage, we computed canonical correlations between our multiple predictor and multiple outcome measures. A test of the null hypothesis that all correlations between multiple predictor and multiple criterion variables are zero order is equivalent to a nonsignificant canonical correlation between a set of predictor and a set of criterion variables (46). A significant canonical correlation allows rejection of the null hypothesis that all correlations between multiple predictor and multiple criterion variables are zero order, and thus offers protection against a type I error when multiple comparisons are computed (46). If the overall canonical correlation was significant, in the second stage of analysis, we utilized regression analyses to identify which maternal predictors were driving the relation to offspring diet. All significant regressions were recomputed, initially entering in our 2 measures of family SES, to determine whether the contributions of maternal education or intelligence would change after partialling out family SES levels. Secondary analyses involved the use of correlations to test for interrelations between the predictor and outcome variables, and the use of t tests when group comparisons were made. In addition, when testing for mediation, we applied the Freedman-Schatzkin testing procedure, as recommended by MacKinnon et al. (47).
|
RESULTS |
|---|
|
TOPABSTRACTSUBJECTS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Descriptive data on child nutritional measures are shown in Table 1. Although the infants in this study were not a population at risk for severe protein-energy malnutrition, they were at risk for inadequate intake of certain micronutrients, especially iron and zinc. The physiologic consequences of inadequate iron intake were confirmed by both the hemoglobin and transferrin saturation values, with mean hemoglobin values falling below the cutoff value for iron deficiency anemia (anemia cutoff, hemoglobin < 110 g/L), and mean transferrin saturation also falling below 16% transferrin saturation. In addition, 16% of our sample at 12 mo and 19.3% of our sample at 18 mo had elevated concentrations of C-reactive protein, which is a marker for acute inflammation (40). Because transferrin saturation values can be sensitive to inflammation, we examined the possibility of a relation between C-reactive protein concentrations and serum iron, TIBC, and transferrin saturation scores. None of these relations were significant.
Descriptive statistics for family characteristics, intercorrelations among our predictors, and intercorrelations among our outcome measures are found in the online supporting material. Two descriptive findings are of note. First, maternal height was significantly related to child length at 3 (r = 0.29, P < 0.01), 6 (r = 0.36, P < 0.01), 12 (r = 0.37, P < 0.01), and 18 mo (r = 0.33, P < 0.01), as well as to maternal years of education (r = 0.25, P < 0.01) and maternal intelligence (r = 0.17, P < 0.01). A situation in which both the predictor variable (maternal education, maternal intelligence) and the outcome variable (offspring length) are related to a 3rd intervening variable (maternal height) meets the criteria for mediation (47). In a mediation situation, the relation of predictor to outcome variable may be due to the common influence of the intervening variable on both predictor and outcome variables. Because of this observed pattern, we tested for mediation in all analyses involving offspring length.
Second, there were unexpected significant differences on 4 of our measures between the 104 infants who were tested at 12 and 18 mo compared with the 137 who were tested at 12 but not at 18 mo. Infants who were not tested at 18 mo lived in homes rated as lower quality (t = –2.98, P < 0.01), weighed more (t = 2.28, P < 0.05), had higher hemoglobin levels (t = 2.24, P < 0.05), and had lower dietary risk scores (t = 2.79, P < 0.01) than infants who were tested at both 12 and 18 mo.
Predictors of offspring nutrition
Infant and toddler diet.
In our first analysis, we computed the canonical correlation between our 3- and 6-mo feeding code scores and our 6- and 12-mo dietary risk score as outcome variables, and maternal intelligence and education level scores as predictors. The overall canonical correlation between maternal education and intelligence levels and our 3-, 6-, and 12-mo dietary intake measures was significant (Rc = 0.24, 
2 = 19.04, P < 0.025), indicating that at least some maternal predictors were significantly related to offspring dietary intake over and above chance levels.
In the second stage of our data analysis, breakdown of this significant canonical correlation indicated nonsignificant regressions between maternal intelligence and education level and 3-mo feeding code (R = 0.08) and 12-mo dietary risk scores (R = 0.13). However, the regression relating 6-mo feeding code scores to maternal intelligence and education approached the P < 0.05 level (R = 0.15, F [2,236] = 2.85, P = 0.06). The breakdown of this regression indicated that predictive variance was accounted for primarily by maternal education level, and less use of other liquids or foods to complement breast milk was associated with higher maternal education levels. We then recomputed this latter regression initially entering in both home quality and family possession SES measures as separate covariates. This analysis revealed that the relation between maternal education level and offspring feeding code scores continued to be predictive, at essentially the same magnitude and in the same direction, indicating a unique contribution of maternal education over and above family SES levels. Both the unadjusted and adjusted ß for maternal education are found in Table 2.
|
Because of reduced sample size and only 1 dietary outcome measure (dietary risk), we computed results separately for the 18-mo data using multiple regression. The regression relating maternal intelligence and education level to offspring dietary risk at 18 mo was significant (R = 0.32 [F 2,98] = 5.71, P < 0.01). Predictive power was again uniquely associated with maternal intelligence, with lower dietary risk scores for offspring of more intelligent mothers. When the regression was recomputed taking family SES into account, the contribution of maternal intelligence to offspring diet remained significant at essentially the same magnitude and in the same direction (Table 2).
Infant and toddler anthropometry.
The overall canonical correlation between maternal intelligence and education level as predictors and offspring length, weight, and skinfold thickness at 3, 6, and 12 mo was not significant (Rc = 0.30, 2 = 23.00). However, the age-specific canonical correlations involving this predictor and outcome set at 3 mo (Rc = 0.22, 2 = 12.06, P = 0.061) and at 12 mo (Rc = 0.23, 2 = 12.44, P = 0.053) did approach the P < 0.05 level (the 6-mo canonical correlation was not significant: Rc = 0.15, 2 = 5.88). To minimize the possibility of a type II error, we choose to break down the 3- and 12-mo canonical correlations.5 Neither of the regressions at 3 mo were significant (maternal education and anthropometry: R = 0.11, F [3,234] = 1.04; maternal intelligence and anthropometry, R = 0.14, F [3,244] = 1.73). At 12 mo of age, although the regression relating maternal intelligence to anthropometry was not significant (R = 0.16, F [3,228] = 2.02), the regression for maternal education level was significant (R = 0.19, F [3,238] = 4.18, P < 0.025). Children of more educated women were taller than offspring of less educated women. When this analysis was rerun entering in our 2 SES indices, the contribution of maternal education to offspring length remained significant at essentially the same magnitude and in the same direction as the results from the unadjusted analysis (Table 3).
|
Because of reduced sample size, we again computed results for the 18-mo data set separately. At 18 mo, the canonical correlation between maternal education and intelligence as predictors and offspring length, weight, and skinfold thickness was significant (Rc = 0.36, Wilks Lambda = 2.54, P < 0.025). The breakdown regression for maternal intelligence was not significant (0.20, F [3.97] = 1.37). However, the regression involving maternal education was significant (R = 0.34, F [3.101] = 4.53, P < 0.01). The offspring of more educated women were taller than offspring of less educated women. When this analysis was rerun entering in our 2 SES indices, the contribution of maternal education to offspring length remained significant at essentially the same magnitude and in the same direction (Table 3).
When mediational analysis procedures were applied to the 18-mo database, the prediction of infant length by maternal education level declined significantly after adjusting for maternal height [change in ß = –0.092, t (105,2) = 2.41, P < 0.05]. However, the prediction of infant length by maternal height also declined significantly after adjusting for maternal education level [change in ß = –0.096, t (105,2) = 2.52, P < 0.05].
Toddler biochemistry.
Because of the reduced sample size at 18 mo of age we computed separate canonical correlations between maternal education and intelligence, and offspring hemoglobin and transferrin saturation at 12 and 18 mo. Although the canonical correlation between this predictor and outcome variable set was not significant at 12 mo of age (Rc = 0.26, 2 = 6.12), the canonical correlation at 18 mo had a P-value < 0.10. (Rc = 0.31, 2 = 8.06, P = 0.089). Given the paucity of data relating maternal characteristics to offspring iron status, we chose to break down the 18-mo canonical correlation. Because the regression using transferrin saturation as a predictor had a P-value > 0.05 (R = 0.25, F [2,83] = 2.82, P = 0.070), we chose not to interpret this relation. The regression involving hemoglobin at 18 mo was significant (R = 0.28, F [2,87] = 3.82, P < 0.05). Although there was no significant predictive variance associated with maternal education level, the relation of maternal intelligence to offspring hemoglobin level did approach P < 0.05, with the pattern indicating higher hemoglobin levels for offspring of more intelligent mothers. When our analysis was rerun, taking into account the effect of our 2 measures of family SES, the contribution of maternal intelligence to offspring hemoglobin levels remained at the same magnitude and in the same direction as in the original analysis (Table 4).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTSUBJECTS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Although maternal education level predicted infant length, our results for length fit the pattern of conditions required for mediation of a predictor-outcome relation by an intervening variable, namely, maternal height. Analysis for mediation effects indicated that the relation between maternal education and infant length at 12 mo could be attributed to the mediating influence of maternal height. At 18 mo of age, dual mediation appeared to be occurring, i.e., the relation between maternal education and offspring length was mediated by maternal height, whereas the relation between maternal height and offspring length was mediated by maternal education. The lack of dual mediation at 12 mo does not appear to reflect reduced power given that differences appeared at 18 mo with a smaller sample. Similarly, the discrepancy between the 12- and 18-mo mediational results cannot be attributed to differences in education level, given that there were no differences in the education level of the mothers assessed at 12 and 18 mo. The mechanism linking maternal height and level of education is likely to be a component of poverty, such as inadequate maternal nutrition in early childhood. Early poor nutrition of women could result in both stunted maternal growth and an increased risk of maternal school failure, which in turn could adversely affect the physical growth of their offspring. In support of this hypothesis, studies from developing countries reported that smaller children are less likely to initially enroll in school on time and are more likely to drop out of school during or after grade school (49,50).
Why do higher levels of education for women in developing countries translate into better nutrition for offspring? The assumption that the relation of higher maternal education to better nutrition of offspring simply reflects higher family SES is not supported by either previous research (31,51) or by the results of the present study, indicating significant prediction from maternal education even after controlling for family economic factors. The alternative assumption that more schooling leads to greater exposure to nutrition principles does not appear to be valid in Peru, given the relatively small amount of time spent on teaching nutrition in Peruvian schools. It was also hypothesized that more educated women in developing countries are more receptive to or better able to understand new sources of information about childcare and nutrition (52), or are better able to maximize utilization of existing family and community resources (51–53). The characteristics describing more educated women parallel the characteristics ascribed to more intelligent individuals, who are often described as better able to modify, adapt to, and deal with complexities in their existing environment (36,37). In addition, there are clear links between higher levels of maternal education and intelligence in developing countries (34,35). Particularly in developing countries, where low family income limits food choice, parental characteristics that can influence food choice, and feeding strategies, such as maternal intelligence, may be more important influences on the nutrition of offspring than would be the case in situations in which families have sufficient resources to purchase whatever and how much food they desire.
A few studies from developing countries have reported better nutrition for offspring of more intelligent women (31,38). However, as discussed in the introduction, none of these earlier studies are conclusive. Results from the present study replicate this earlier research in terms of indicating that offspring of more intelligent women are receiving a higher quality diet (reduced dietary risk) than offspring of less intelligent women, even after taking into account family economic circumstances and maternal education. In addition, our results expand upon previous research in terms of our finding that lower maternal intelligence is related to lower offspring hemoglobin. However, before concluding that our results indicate a unique relation of maternal intelligence to offspring iron status, it is necessary to provide corroborative evidence that the low hemoglobin values in our sample are a marker for low iron status, given that hemoglobin levels can be influenced by a variety of other factors in addition to iron deficiency (40). Two sources of corroborative evidence are found in our data. First, our descriptive data (Table 1) indicate that a high proportion of our sample had diets that were low in iron, and that children at higher dietary risk had lower hemoglobin values then children at lower dietary risk (online supporting material). Second, if hemoglobin values reflect something other then iron status, we would expect a discordance between our 2 iron status measures, with substantial numbers of children in our sample having low hemoglobin values paired with normal levels of transferrin saturation. Comparing cross-tabulations, we find that this is not the case, with only 3.7% of our sample at 12 mo and 4.4% of our sample at 18 mo having low hemoglobin and normal transferrin saturation. These 2 lines of evidence support our interpretation that maternal intelligence is a unique predictor of offspring iron status. Given both the prevalence and adverse developmental consequences of early iron deficiency (1,3,39), identification of noneconomic caregiver characteristics that predict offspring iron status is an area of research that has implications for the development of primary prevention strategies to reduce levels of iron deficiency anemia in infancy.
In interpreting the present results, it could be argued that our findings may reflect more educated or more intelligent women being more accurate reporters of their feeding practices and offspring diet than are less educated and less intelligent women. However, the validity of this alternative interpretation is weakened given that these maternal report measures predicted objective measures of adequacy of offspring nutrition such as physical growth and iron status. It could also be argued that the relation between maternal education and maternal intelligence may not reflect contributions of the school environment to intelligence, but rather may simply mean that more intelligent women are likely to progress further in school. This interpretation is countered by reviews documenting that there are additional cognitive benefits of a longer time in school, over and above the contribution of initial level of intelligence (32).
In terms of other potential limitations on interpretation of our results, there were unexpected differences in 4 of 9 of our measures between infants who were tested at 12 and 18 mo, compared with those who were tested at 12 mo but not at 18 mo due to exhaustion of project funds. However, there were no group differences in our 2 major predictors, maternal education and intelligence. In addition, the group differences for home quality, weight, hemoglobin, and dietary risk, although significant, were primarily of intermediate to small effect size [mean effect size difference 0.32 of a SD, range 0.29–0.37 of a SD (54)]. The overall pattern of differences between those whom we tested at 12 and 18 mo vs. those whom we tested only at 12 mo does not suggest a pattern of systematic bias or that our 18-mo results would have been seriously discrepant if we had been able to test all infants at this age.
A second limitation on interpretation of our findings is the fact that significant relations were seen primarily when infants were 6 and 18 mo of age, with fewer significant relations found when infants were 3 or 12 mo of age. Variability in outcome as a function of age is not unique to our findings, as seen in the results from studies relating parental education and intelligence to the nutrition of offspring (31) and studies relating supplementary feeding to cognitive development (2,3). Our data (Table 1) do not support the hypothesis that the age differences reported are related to restricted outcome variability at 3 and 12 mo. Nor does the pattern of results support a model of cumulative influence, with contributions of parental characteristics increasing as children grow older. Although there is no obvious explanation for our age findings, one implication of our findings and those from earlier research (31,55) is the importance of systematically examining age differences as a potential moderating variable.
A third limitation on interpretation of our results involves the use of a 4-point scale for our 3- and 6-mo feeding code measure. When variables with reduced range are utilized, correlations will be lower than when measures with a less limited distribution are used, thus increasing the risk of a type II error. Thus, there remains the possibility that existing relations between our predictors and the 3-mo feeding code scores are being masked by the restricted distribution of feeding code scores. A final limitation of this study is the difficulty of assessing the dietary risk score from complementary foods, especially at 6 mo, which is a transitional period in the process of complementing breast milk with other foods. Nevertheless, the significant correlations of our dietary risk scores with both mean nutrient density adequacy and with measures of infant iron status, even though of moderate magnitude, support the assumption that the dietary risk score is a valid estimation in the absence of knowing the quantity of breast milk consumed.
Taken together, results from past research and our current findings support the validity of a proposed model incorporating both a direct influence of maternal education on offspring nutrition as well as an indirect influence, through the influence of maternal education on maternal intelligence (56). Overall, the fact that maternal intelligence uniquely predicts offspring dietary risk and iron status, whereas maternal education uniquely predicts early infant feeding practices and is mediated by maternal height when physical growth is the outcome, suggests that maternal education is not a proxy for maternal intelligence and that maternal intelligence is not a proxy for maternal education. Rather, both maternal education and intelligence appear to have unique influences on different aspects of the nutritional status and diet of offspring.
Several directions for future research are suggested from our findings. One such direction would be a follow-up study relating maternal intelligence to offspring iron status using measures of child iron stores, such as ferritin, that may be more sensitive to the nature of the child’s diet. Another direction would be a detailed study of the specific food-purchasing strategies used by more or less educated or intelligent women. A third direction would involve the question whether relations between maternal intelligence and offspring nutrition are due to variability in general intelligence (e.g., IQ), or to variability in specific components of intelligence [e.g., Sternberg’s distinction between "practical" vs. "analytical" intelligence (37)]. If relations of maternal intelligence to offspring nutrition were driven by specific components of intelligence, an important policy question would be whether school experiences and curriculum in developing countries are likely to help develop those components of intelligence, which could facilitate the ability of women to provide adequate nutrition for their infants, even when family income is restricted.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
2 Supplemental tables and descriptive data are available as Online Supporting Material with the online posting of this paper at www.nutrition.org.
4 Abbreviations used: IIN, Instituto de Investigación Nutricional; RI, recommended intake; SES, socioeconomic status; TIBC, total iron binding capacity.
5 Because we had multiple anthropometry outcome variables, we chose to separately regress education and intelligence with all anthropometry outcomes rather than separately regressing each anthropometry outcome against both education and intelligence. This allowed us to reduce the number of regressions computed for this data set, which we saw as an appropriate analytic strategy given that the canonical correlation for anthropometry only approached the traditional 0.05 level of statistical significance.
Manuscript received 4 January 2005. Initial review completed 7 February 2005. Revision accepted 19 June 2005.
|
LITERATURE CITED |
|---|
|
TOPABSTRACTSUBJECTS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
1. UNICEF (1998) The state of the world’s children 1998: malnutrition causes, consequences and solutions. Nutr. Rev. 56:115-123.[Medline]
2. Grantham-McGregor, S., Fernald, L. & Sethuraman, K. (1999) The effects of health and nutrition on cognitive and behavioral development in children in the first three years of life. Part 1: Low birth weight, breast feeding and protein energy malnutrition. Food Nutr. Bull. 20:53-75.
3. Grantham-McGregor, S., Fernald, L. & Sethuraman, K. (1999) Effects of health and nutrition on cognitive and behavioral development in children in the first three years of life. Part 2: Infections and micronutrient deficiencies: iodine, iron and zinc. Food Nutr. Bull. 20:76-99.
4. Sigman, M., Espinosa, M. & Whaley, S. (1998) Mild malnutrition and the cognitive development of Kenyan school children. Nutrition Health and Child Development. Pan American Health Organization Scientific Monograph Number 566 :91-103 PAHO Washington, DC.
5. Wachs, T. D. (2002) Nutritional deficiencies as a biological context for development. Hartup, W. Silbereisen, R. eds. Growing Points in Developmental Science :64-84 Psychology Press Hove, UK.
6. Bhargava, A. & Bouis, H (1992) Maximum likelihood estimation of between and within variations in energy and protein intakes from infancy to adolescence for the Philippines. Stat. Med. 11:533-545.[Medline]
7. Mwadime, R. & Baldwin, S. (1994) Relationship between household access to food and malnutrition in eastern and southern Africa. East Afr. Med. J. 71:571-579.[Medline]
8. Thomas, D., Lavy, V. & Strauss, J. (1996) Public policy and anthropometric outcomes in the Cote d’Ivoire. J. Public Econ. 61:155-192.
9. Miranda, C., Turecki, G., Mari, J., Andreoli, S., Marcolim, M., Goihman, S., Piccini, R., Strom, B. & Berlin, J. (1996) Mental health of the mothers of malnourished children. Int. J. Epidemiol. 25:128-133.
10. Tucker, K. & Young, F. (1989) Household structure and child nutrition. Soc. Indic. Res. 21:629-649.
11. Valenzuela, M. (1997) Maternal sensitivity in a developing society: the context of urban poverty and infant chronic undernutrition. Dev. Psychol. 33:845-855.[Medline]
12. Mulder-Sibanda, M., Sibanda-Mulder, F., d’Alois, L. & Verna, D. (2002) Malnutrition in food surplus areas. Food Nutr. Bull. 23:253-261.[Medline]
13. Rahman, A., Lovel, H., Bunn, J., Iqbal, Z. & Harrington, R (2004) Mother’s mental health and infant growth: a case-control study from Rawalpindi, Pakistan. Child Care Health Dev. 30:21-27.[Medline]
14. Cravioto, J. & DeLicardie, E. (1972) Environmental correlates of severe clinical malnutrition and language development in survivors from kwashiorkor or marasmus. Nutrition, the Nervous System and Behavior. Pan American Health Organization Scientific Publication # 251 :73-94 PAHO Washington, DC.
15. Gittelsohn, J., Thapa, M. & Landman, L. (1997) Cultural factors, caloric intake and micronutrient sufficiency in rural Nepali households. Soc. Sci. Med. 44:1739-1749.[Medline]
16. Kerr, M., Bogues, J. & Kerr, D. (1978) Psychosocial functioning of mothers of malnourished children. Pediatrics 62:778-784.
17. Powell, C. & Grantham-McGregor, S. (1985) The ecology of nutritional status and development of young children in Kingston, Jamaica. Am. J. Clin. Nutr. 31:1322-1331.
18. Engle, P., Zeitlin, M., Medrano, Y. & Garcia, L. (1996) Growth consequences of low income Nicaraguan mothers theories about feeding one year olds. Harkness, S. Super, C. eds. Parents Cultural Belief Systems :428-446 Guilford New York, NY.
19. Mull, D. (1991) Traditional perceptions of marasmus in Pakistan. Soc. Sci. Med. 32:175-191.[Medline]
20. Scheper-Hughes, N. (1987) A basic strangeness: Maternal estrangement and infant death. Super, C. eds. The Role of Culture in Developmental Disorder :131-153 Academic Press San Diego, CA.
21. Zeitlan, M. (1996) My child is my crown: Yoruba parental theories and practices in early childhood. Harkness, S. Super, C. eds. Parents Cultural Belief Systems :407-429 Guilford New York, NY.
22. Birch, L. & Fisher, J. (1998) Development of eating behaviors among children and adolescents. Pediatrics 101:539-549.
23. Aghaji, M. (2002) Exclusive breast-feeding practice and associated factors in Enugu Nigeria. West Afr. J. Med. 21:66-69.[Medline]
24. Giashuddin, M., Kabir, M., Rahman, A. & Hannan, M. (2003) Exclusive breastfeeding and nutritional status in Bangladesh. Indian J. Pediatr. 70:471-475.[Medline]
25. Li, L., Li, S., Ali, M. & Ushijima, H. (2003) Feeding practice of infants and their correlates in urban areas of Beijing, China. Pediatr. Int. 45:400-406.[Medline]
26. Nwankwo, B. & Brieger, W. (2002) Exclusive breastfeeding is undermined by use of other liquids in rural southwestern Nigeria. J. Trop. Pediatr. 48:109-112.
27. Vaahtera, M., Kulmala, T., Hietanen, A., Ndekha, M., Cullinan, T., Salin, M. & Ashorn, P. (2001) Breastfeeding and complementary feeding practices in rural Malawi. Acta Paediatr 90:328-332.[Medline]
28. Abidoye, R. & Nwachie, C. (2001) Comparative anthropometrics of 3 to 24-month-old children breastfed in both high and low socio-economic strata in Lagos, Nigeria. Public Health 115:157-162.[Medline]
29. Alderman, H., Hentschel, J. & Sabates, R (2003) With the help of one’s neighbors: externalities in the production of nutrition in Peru. Soc. Sci. Med. 56:2019-2031.[Medline]
30. Tumwine, J. & Barugahare, W. (2002) Nutrition status of children in Kasese district at the Uganda-Congo border. East Afr. Med. J. 79:427-434.[Medline]
31. Wachs, T. D. & McCabe, G. (2001) Relation of maternal intelligence and schooling to offspring nutritional intake. Int. J. Behav. Dev. 25:444-449.
32. Ceci, S. (1991) How much does schooling influence general intelligence and its cognitive components?. Dev. Psychol. 27:703-722.
33. Sandiford, P., Cassel, J., Sanchez, G. & Coldham, C. (1997) Does intelligence account for the link between maternal literacy and child survival?. Soc. Sci. Med. 45:1231-1239.[Medline]
34. Stevenson, H., Chen, C. & Booth, J. (1990) Effects of schooling and urban rural residents on gender differences in cognitive ability and academic achievement. Sex Roles 23:535-551.
35. Wachs, T. D., McCabe, G., Yunis, F., Kirksey, A., Harrison, G., Galal, O. & Jerome, N. (1996) Relation of nutritional intake and context to cognitive performance of Egyptian adults. Intelligence 22:129-154.
36. Gottfredson, L. (1997) Why g matters: The complexity of everyday life. Intelligence 24:79-132.
37. Sternberg, R. (1997) The concept of intelligence and its role in lifelong learning and success. Am. Psychol. 52:1030-1037.
38. Baker-Henningham, H., Powell, C., Walker, S. & Grantham-McGregor, S. (2003) Mothers of undernourished children have poorer psychosocial functioning and this is associated with stimulation provided in the home. Eur. J. Clin. Nutr. 57:786-792.[Medline]
39. Lozoff, B. (1998) Exploratory mechanism for poorer development in iron-deficient anemic infants. Nutrition, Health and Child Development. Pan American Health Organization Scientific Monograph, 566 :162-178 PAHO Washington, DC.
40. Lee, R. & Nieman, D. (1996) Exploratory mechanism for poorer development in iron-deficient anemic infants. Nutritional Assessment 2nd ed. Mosby St. Louis, MO.
41. Creed-Kanashiro, H. (1998) Exploratory mechanism for poorer development in iron-deficient anemic infants. Tabla de Composición de Alimentos Peruanos Instituto de Investigación Nutricional Lima, Peru.
42. Burchinal, M., Roberts, J., Hooper, S. & Zeisel, S (2000) Cumulative risk and early cognitive development: a comparison of statistical risk models. Dev. Psychol. 36:793-807.[Medline]
43. Dewey, K. & Brown, K (2003) Update on technical issues concerning complementary feeding of young children in developing countries and implications for intervention programs. Food Nutr. Bull. 24:5-28.[Medline]
44. Food and Agriculture Organization of the United Nations/World Health Organization (2002) Update on technical issues concerning complementary feeding of young children in developing countries and implications for intervention programs. Human Vitamin and Mineral Requirements Report of a joint FAO/WHO consultation Rome, Italy.
45. World Health Organization (1998) Complementary Feeding in Developing Countries: A Review of Current Scientific Knowledge. WHO/NUT/98.1 WHO Geneva Switzerland.
46. Johnson, R. & Wichern, D. (1982) Complementary Feeding in Developing Countries: A Review of Current Scientific Knowledge. Applied Multivariate Statistics 2nd ed. Prentice-Hall Englewood Cliffs, NJ.
47. Mackinnon, D., Lockwood, C., Hoffman, J., West, S. & Sheete, V. (2002) A comparison of methods to test mediation and other intervening variable effects. Psychol. Methods 7:83-104.[Medline]
48. Demographic and Health Survey (2001) A comparison of methods to test mediation and other intervening variable effects. Encuesta demografica y de salud familiar Instituto Nacional de Estadistica e Informatica. Lima, Peru.
49. Daniels, M. & Adair, L. (2004) Growth in young Filipino children predicts schooling trajectories through high school. J. Nutr. 134:1439-1446.
50. The Partnership for Child Development (1999) Short stature and the age of enrollment in primary school: studies in two African countries. Soc. Sci. Med. 48:675-682.[Medline]
51. Cleland, J. & van Ginneken, J. (1988) Maternal education and child survival in developing countries. Soc. Sci. Med. 27:1357-1368.[Medline]
52. Guldan, G., Zeitlin, M., Beiser, A., Super, C., Gershoff, S. & Datta, S. (1993) Maternal education and child feeding practices in rural Bangladesh. Soc. Sci. Med. 36:925-935.[Medline]
53. Thomas, D., Strauss, J. & Henriques, M. (1991) How does mother’s education affect child height?. J. Hum. Resour. 26:183-211.
54. Cohen, J. (1992) A power primer. Psychol. Bull. 112:155-159.[Medline]
55. Pollitt, E., Gorman, K., Engle, P., Martorell, R. & Rivera, J. (1993) Early supplementary feeding and cognition. Monogr. Soc. Child Dev. 58:1-99 discussion 111–118.
56. Wachs, T. D. (2005) Linking nutrition and education: a cross-generation model. Food Nutr. Bull. (suppl.) 26:S159-S167.[Medline]
This article has been cited by other articles:
|
|
 |
|
  C M Schooling, C Q Jiang, M Heys, W S Zhang, P. Adab, K K Cheng, T H Lam, and G M Leung Are height and leg length universal markers of childhood conditions? The Guangzhou Biobank cohort study J Epidemiol Community Health, July 1, 2008; 62(7): 607 - 614. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Giovannini, G. Radaelli, E. Riva, and G. Banderali Low Prepregnant Body Mass Index and Breastfeeding Practices J Hum Lact, February 1, 2007; 23(1): 44 - 51. [Abstract] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
