点击显示 收起
1 From the Departments of International Health (ADS, MH, UR, DGS, and RM) and Biostatistics (HXB), Emory University, Atlanta.
2 Supported by the National Institutes of Health and the American Heart Association. 3 Address reprint requests to AD Stein, Department of International Health, Rollins School of Public Health of Emory University, 1518 Clifton Road NE, Atlanta, GA 30322. E-mail: astein2{at}sph.emory.edu.
ABSTRACT
Background: The secular increase in height is assumed to result from long-term improvements in nutritional intakes and reductions in infectious disease burdens. Nutritional supplementation in early life reduces stunting in chronically undernourished populations. It is unknown whether these improvements can be transmitted to subsequent generations.
Objective: Our objective was to estimate the intergenerational effect on offspring length of improved nutrition in the mother’s childhood.
Design: We studied 263 children born in 1996–1999 to 231 women who had received nutritional supplementation, ie, atole (high-protein, moderate-energy drink) or fresco (nonprotein, low-energy drink), prenatally and up to age 7 y as part of a community trial in Guatemala between 1969 and 1977. Child length was measured at different times to age 36 mo.
Results: Children born to women who received the enhanced supplement were taller (age-adjusted difference: 0.80 cm; 95% CI: 0.16, 1.44 cm) than were children whose mothers received the low-energy supplement. This increment was independent of the children’s birth weight or socioeconomic status but was substantially attenuated and no longer significant after adjustment for maternal height (adjusted difference: 0.43 cm; 95% CI: -0.10, 0.96 cm; P > 0.10). The effect of maternal nutritional supplementation was more pronounced in boys than in girls (P for interaction < 0.10) and in children born to women who received supplements at ages 3–7 y than in children born to women who received supplements at ages 0–3 y (P for interaction < 0.01).
Conclusion: Nutritional supplementation in childhood has positive effects on both the supplemented persons and on the subsequent generation.
Key Words: Children • energy • Guatemala • growth • intergenerational effects • nutrition • protein • supplementation
INTRODUCTION
In many countries there has been a secular trend of increased attained height (1). It has been recognized for many years that inadequate early nutrition and high burdens of infectious disease contribute to growth deficiency in children (2). Stunting, which has its origins in prenatal growth failure (3), increases the risk of adverse pregnancy outcomes, including low birth weight (4), thus perpetuating the cycle of undernutrition.
Increasing the growth rates of girls, if it translates to improved stature and to increased birth weights in their children, represents a powerful long-term strategy for improving public health in developing countries. Growth deficits attributable to undernutrition can sometimes be ameliorated with improved nutrition in the first few years of life (5). Whether the effects of such an intervention have persistent effects on growth in the next generation remains an unanswered question. We therefore evaluated the effect of nutritional supplementation of mothers during their early childhood on the growth of their offspring. The study, originally a randomized controlled trial of nutritional supplementation, was conducted in Guatemala.
SUBJECTS AND METHODS
The data for the present analysis were derived from research conducted between 1969 and 1999 among 3 generations of residents from 4 villages in eastern Guatemala. From 1969 to 1977, the Institute of Nutrition of Central America and Panama (INCAP) conducted a controlled intervention trial in 4 rural communities to study the effects of nutritional supplementation on the growth and development of children. The 4 villages were paired by size and allocated randomly within pairs to receive 1 of 2 supplements. One pair received atole, a nutritious gruel with a high-protein and moderate-energy content. The other pair received fresco, a nonprotein, fruit-flavored sweetened drink. Both supplements were fortified with selected vitamins and minerals. Supplements were freely available to all village residents twice each day in a central location in the village, although only the intakes of supplement by pregnant and lactating women and children aged < 7 y were recorded. Compared with the fresco group, the group of mothers and children who were provided with the atole supplement had lower infant mortality rates and improved growth rates (in those aged < 3 y), and birth weight was associated with maternal energy intake in both the atole and fresco groups (6).
From 1996 to 1999, we conducted a prospective study of pregnancy and child growth in the same 4 villages. All women of childbearing age, regardless of participation in the original supplementation study, were regularly screened for pregnancy, and pregnancies were followed through delivery. Mothers and children were then studied intensively until the children reached 36 mo of age. Birth weights were recorded, and repeated measures of child length were obtained by using a portable length board. All phases of the study were approved by the Institutional Review Boards at Emory and INCAP, and informed consent was obtained for all measurements.
Data management
We used data from 3 generations: grandmothers (G1; participated as adults in the 1969–1977 study), mothers (G2; participated as children in the 1969–1977 study), and children (G3; born during the 1996–1999 study). To be eligible for the present analysis, the G3 children had to have at least one measure of length, and the G2 mothers had to have been born in one of the study villages between 1962 and 1977, and thus had been exposed to atole or fresco for at least part of their own early life. Births to in-migrant women, who were followed and studied during 1996–1999 but for whom maternal early growth patterns are unknown, and to out-migrants, who were not followed-up, were excluded from this analysis. We tested the hypothesis that the type of supplementation provided to the G1 and G2 generations between 1969 and 1977 affected the growth rate of the G3 children, taking into account other determinants of the growth rate, including grandparental [eg, height, socioeconomic status, and education level], maternal (eg, height, body composition, socioeconomic status, and education level), and child (eg, birth length, birth weight, and gestational age) characteristics, which are potential confounders or mediators of any such effect.
Variable definitions
Child length was the primary outcome variable. Body length of the G3 children was measured at different times up to the age of 36 mo. When more than one measurement was made in a calendar month, the most recent measurement was used for the analysis. Because of the study closeout in 1999, few children reached 36 mo of age during the data collection period, and the child length data are right censored. Child age at the time of each body length measurement was rounded to the nearest month. Treatment group is a dichotomous variable indicating the type of nutrition intervention (atole or fresco). Socioeconomic status was derived from the mother’s education level, father’s occupation, and house quality factor score by using data collected during 1975 and 1996 and was rescaled to yield an index with a mean of 0 and an SD of 1 with the use of an approach described elsewhere (7). Maternal and grandparental education levels were categorized as the completion of primary school or not. Gestational age was retained as a continuous variable (range: 32–44 wk). Child birth order was categorized as 1 if the child was born as the first or second child of the family and as 0 if otherwise. Twin status was a dichotomous variable.
Missing covariate data were imputed as the variable mean plus or minus a random factor, generated on the basis of the distribution of the observed data for that variable, to ensure that neither the mean nor the variance of the variable was affected. Imputation of covariates enables inclusion of all records for which the primary independent and dependent variables (supplementation type and offspring length) are available and does not introduce bias if the assumption of missing at random holds (8). We generated 15 imputed data sets and repeated the statistical analysis on each set with the use of SAS PROC MIANALYZE to generate summary point estimates, CIs, and P values (SAS software, version 8; SAS Institute, Cary, NC).
Model development and statistical analysis
We computed summary statistics for all variables. Linear mixed regression models were then fitted. The model was fitted by using SAS PROC MIXED with REPEATED statement (SAS software, version 8). The likelihood ratio test, together with the Akaike and Schwarz information criteria, were used for model selection (9). Linearity and outliers were checked by scatter plots and residual plots. We developed a series of models. Because age is a major determinant of body length, we first evaluated the age-adjusted effect of supplement type. The significance of age-specific differences in length were tested by using a Bonferroni correction to account for the multiple ages at which children were measured. The effects of other characteristics of the children, mothers, and grandparents were screened univariately by fitting a series of models. Each potential confounder was entered individually into the model to explore its relation with child length and age at measurement. Main effects with P values > 0.10 were excluded from further model building. Maternal height was considered a potential mediator because of its likely presence in the causal pathway—in our study height was affected by supplementation in the first 3 y of life (7) and it is a known cause of variation in offspring height (3). Thus, models were developed with and without maternal height.
We then tested for an interaction of atole with the retained main effects by adding multiplicative interaction terms to the model without maternal height. Interaction terms were retained in the model at P ≤ 0.10. Finally, we tested whether the age at which supplementation was available affected the relations observed by testing a cohort by treatment interaction. We defined cohorts with different exposures to supplementation during prenatal, early postnatal (birth to 3 y of age), and later postnatal (3–7 y of age) periods, as described previously (6). Cohort 1 was born after 28 February 1974 and hence had only prenatal and early postnatal exposure, cohort 2 was born between 1 March 1969 and 28 February 1974 and had exposure in prenatal and early postnatal life and some later postnatal exposure. Cohort 3 was born between 1 January 1966 and 28 February 1969 and hence had no prenatal exposure, only limited early postnatal exposure, and considerable later postnatal exposure. Cohort 4 was born before 1 January 1966 and had no exposure in the first 3 y of life. We previously showed that supplementation in the first 3 y of postnatal life had the greatest effect on attained height in this population (5), and we hypothesized that the intergenerational effects of atole, if any, would be apparent in children born to this group of women. All tests were two-tailed.
RESULTS
Our data set included 263 children born to 231 mothers. The children had a mean of 9.6 measures of length (range: 1–17). Selected characteristics of the 3 generations, by treatment allocation in 1969–1977, are presented in Table 1. Annual growth rates for the children were slightly higher than those for their mothers in both the atole and fresco groups (Figure 1).
View this table:
TABLE 1 . Selected characteristics of 3 generations of participants in the Oriente Longitudinal study (Guatemala, 1969–1999), by type of supplementation received1
FIGURE 1. . Mean annual length increments for women in Guatemala who participated in the Oriente Longitudinal Study (1969–1977) and for their children (who were studied between 1996 and 1999), by type of supplementation received. Error bars indicate 95% CIs. Bars with different letters are significantly different, P < 0.05.
Bivariate, age-adjusted effects of child, maternal, and grandparental characteristics on child length are provided in Table 2. Children whose mothers received atole were longer than were children whose mothers received fresco at most ages (Figure 2). The age-adjusted effect of supplement type was 0.80 cm (95% CI: 0.16, 1.44 cm; P < 0.05). The age effect was highly significant (P < 0.0001).
View this table:
TABLE 2 . Effect of child, maternal, and grandparental characteristics on child linear growth from birth to 36 mo of age (mixed multivariate regression) among children born to women in Guatemala exposed early in life to nutritional supplementation1
FIGURE 2. . Mean length from birth to 30 mo of age in children born to women in Guatemala who participated as children in the Oriente Longitudinal Study (1969–1977), by type of supplementation received by the mother. Error bars indicate 95% CIs. *Significantly different from fresco after Bonferroni correction, P < 0.05.
In the multivariate model, maternal attained height and child birth length, weight, sex, and singleton status were each associated with child length (P < 0.05; Table 2). In the model with maternal height, the effect of treatment type (0.43 cm; 95% CI: -0.10, 0.96 cm) was no longer significant (P > 0.10).
We explored interactions between the remaining predictor variables and supplement type. The effect of supplement type was modified by both maternal education and child sex (Table 3). The effect of supplementation with atole was stronger among children whose mothers completed primary school than among children whose mothers did not complete primary school (P for interaction = 0.09) and was also stronger for boys than for girls (P for interaction = 0.08). The third-order interaction term was not significant (P > 0.95), and no other covariates showed significant interactions (P < 0.10) with supplementation type. Boys whose mothers received atole were 2.1 cm taller than were boys whose mothers received fresco, whereas for girls the effect was 0.2 cm. Similarly, the effect of atole was 1.8 cm among children of women who completed primary school and was 0.2 cm among children of women who did not complete high school.
View this table:
TABLE 3 . Joint effects of supplementation type, child’s sex, and maternal education on linear growth from birth to 36 mo of age, among children born to women in Guatemala who received nutritional supplementation early in life
The effect of supplementation type on G3 growth was apparent only in cohorts 3 and 4 (children of women born before the study launch), who received supplementation primarily in the later postnatal years (Table 4). The cohort-by-treatment interaction was highly significant in the age-adjusted model (P < 0.001, 3 df) but not after adjustment for covariates. For children of cohort 3 mothers, who received supplementation during part of the first 3 y of life but entirely during ages 3–7 y, the adjusted effect size was 2.07 cm (95% CI: 0.10, 4.05 cm; P < 0.001). Among children of cohort 4 mothers, who received no supplementation in the first 3 y of life, the adjusted effect size was 1.57 cm (95% CI: 0.004, 3.14 cm; P < 0.05). Both of these estimates were substantially attenuated and became nonsignificant after maternal attained height was added to the model.
View this table:
TABLE 4 . Effect of supplementation type on child growth by timing of supplement exposure1
DISCUSSION
Our study suggests a persistent but small intergenerational benefit of nutritional supplementation early in life among populations at risk of chronic undernutrition. Overall, children born to women who received nutritional supplementation early in life were almost 1 cm taller than were children born to women who received a less nutritious supplement. The difference between the 2 groups was established very early in life, persisted at least through age 2 y, and was independent of effects of supplement type on G3 birth length. The effect was substantially explained by differences in attained height between the 2 groups of mothers (itself attributable to the effects of supplementation), was more pronounced in boys and in children of women who completed primary school, and, surprisingly, was concentrated among children born to women who received supplementation primarily later in childhood.
Transgenerational effects on health have been documented since 1934, when Kermack et al (10) showed that improvements in infant mortality followed improvements in the mortality of women of reproductive age (10). Since then, a rich literature has documented that improvements in maternal health and education are important for child health. Because infant mortality and stunting are strongly associated, one might expect also that improvements in child length would result from improvements in maternal attained height. Our data are the first that we are aware of that document the effect of a specific intervention—nutritional intervention in childhood—on both the intervened population, in terms of attained height, strength, and potential productivity, and on the subsequent generation.
We observed 3 interactions in the data. We found that boys benefited more than did girls from maternal nutritional supplementation. Sex-specific effects on maternal nutrition on growth have been reported elsewhere, in the context of fetal programming, with boys generally being more sensitive to nutritional insults (11, 12). The benefit of maternal childhood supplementation on offspring growth was enhanced among the offspring of women who themselves completed primary school. We previously reported that supplementation with atole had a greater effect on cognitive function among individuals who completed primary school (13). It is likely that maternal education is capturing variance in several characteristics related to infant feeding behaviors, which might themselves contribute to offspring growth. Thus, these 2 interactions are reassuring.
Our finding that the effect of supplementation was stronger among offspring of women who received nutritional supplementation only later in childhood was unexpected, because most of the effect of supplementation on the women’s own growth was observed when supplementation was received prenatally and in the first 3 y of postnatal life (5). The data presented in Table 4 suggest that the maximal effect was observed among children who were exposed to supplementation during ages 3–7 y and during part of the first 3 y of life (cohort 3); a slightly attenuated effect was observed in cohort 4, who were exposed only after age 3 y, and no effect was observed among those born in cohorts 1 and 2, who had partial or full exposure to nutritional supplementation prenatally and during the first 3 y of life. It is possible that the effect in cohort 3 represents a reinforcement in later childhood of an effect in early life, such that among those exposed only early in life, the benefit of atole does not endure. If true, our data suggest that additional benefits to offspring growth might be obtained by supplementation later in life.
However, several alternative explanations need to be considered. First, it is possible that our sample represents a selected and biased subset of those in the original trial. We included in the analysis only those women who remained in the study villages and who gave birth to a child between 1996 and 1999. Out-migration rates differed by study village; hence, the allocation to treatment type became unbalanced. To the extent that out-migration differs also by attained height (14) and child growth rates (for which we have no data), bias may have been introduced.
It is possible that the effect of supplementation on G3 growth was due to consumption of the supplement, and hence increased growth, in adolescence. The oldest G2 women in the study, born between 1962 and 1966, would have had the opportunity to consume supplements through the ages of 11–15 y. We have no record of supplement consumption past the age of 7 y, because this was not the focus of the original study. It appears that there was little consumption of supplements by older children (H Delgado, INCAP, personal communication, 2002).
We conclude that improved nutrition in later childhood improved the growth pattern of the subsequent generation in the chronically undernourished population studied. If confirmed in other populations, these findings have implications for preschool and school-based feeding programs and may also have implications, as suggested by Barker (11), for the epidemic of metabolic disease caused by fetal undernutrition.
ACKNOWLEDGMENTS
We thank Humberto Méndez for expert data management and Sunan Zhang for data analysis.
ADS developed the analytic plan, supervised the data analysis, and wrote the first draft of the manuscript. HXB provided critical input about the statistical methods. MH implemented the data analysis under the guidance of ADS and HXB. UR, DGS, and RM designed the protocol for data collection and provided oversight during data collection. In addition, DGS contributed to the development of the analytic plan. RM designed the overall study under which these data were collected and formulated the generational hypothesis. All authors provided important intellectual contributions to successive drafts of the manuscript. None of the authors had any financial or personal conflict of interest with respect to the material reported in this paper.
REFERENCES