![[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}
|
|
|
|
|
||
Food Consumption and Nutrition Division, International Food Policy Research Institute, Washington, DC and Departments of
* Biostatistics and
Nutrition, University of North Carolina at Chapel Hill, Chapel Hill, NC
2To whom correspondence should be addressed. E-mail: c.eckhardt{at}cgiar.org.
 |
ABSTRACT |
|---|
 TOP ABSTRACT SUBJECTS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
KEY WORDS: • diet diversity • energy intake • height • linear growth • developing country
Diet is arguably the most influential determinant of linear growth because it is through diet that the influences of other determinants of linear growth such as socioeconomic status (SES)3 and infection are largely played out (1–4). Understanding the association between diet and linear growth is particularly salient in developing countries. In these areas, linear growth retardation affects >30% of children, with much higher proportions affected in some areas (5); it is also accompanied by increased morbidity and mortality (6,7) and a variety of other poor outcomes (8–13).
Research on the association between diet and linear growth in developing countries thus far has been constrained largely to the effects seen during infancy and early childhood for several reasons (14–22). Linear growth retardation tends to manifest itself during the 1st 2 y of life because infancy marks the period of fastest linear growth velocity and is particularly vulnerable (23). Thus, research has focused on this important period, and has questioned whether growth trajectories thereafter are capable of improvement and whether determinants of height can have further effects that are independent of their influence during infancy. In addition, due to the difficulties of collecting longitudinal diet and other data in developing countries, there are no studies to date that have had sufficient data to comprehensively examine diet effects over the entirety of the postinfancy period through the attainment of adult stature.
We examined the association between diet and height in the postinfancy period using data from the Cebu Longitudinal Health and Nutrition Survey (CLHNS), which provided longitudinal data on a variety of anthropometric and other measures from birth through adult stature. Most studies of diet and linear growth have focused on macronutrient intakes or on specific micronutrients or foods despite the fact that intakes of macronutrients and micronutrients are associated and work together to influence stature (24). We examined the effects of both energy intake and diet variety (DV), a broad indicator of micronutrient adequacy and a measure that allows for comparisons to be made across the entirety of the postinfancy period given the types of diet data collected at the various CLHNS survey points.
Using longitudinal data analysis methods, we determined the magnitude of the mean independent effects of diet on height across the postinfancy period; second, we examined whether these effects were expressed differently depending on SES and accompanying levels of need; and last, we determined how the observed effects of diet in the postinfancy period differed with age.
|
SUBJECTS AND METHODS |
|---|
|
TOPABSTRACTSUBJECTS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Data come from the CLHNS, a community-based cohort study of children born in 1983–1984 in urban and rural communities of Metro Cebu, the most rapidly growing metropolitan area of the Philippines. The CLHNS data provide detailed longitudinal anthropometric, diet, and household SES data along with measures of maturational timing and other factors relating to maternal and child health.
Data collection for the first round of the CLHNS began during the last trimester of pregnancy for all participating mothers. All pregnant women living in the selected communities during the enrollment period were invited to participate, and >95% enrolled. In the first round of the survey (1983–1986), 3080 singleton live births were included. Child data were collected at birth and then bimonthly for 2 y. Follow-up surveys took place in 1991 (mean age 8.5 y) with 2264 children (74% of the original sample), in 1994 (mean age 11.5 y) with 2192 children (71% of the original sample), in 1998 (mean age 15.5 y) with 2089 children (68% of the original sample), and in 2002 (mean age 18.5 y) with 2029 children (66% of the original sample). Data were collected in accordance with protocols approved by the Institutional Review Board of School of Public Health at the University of North Carolina at Chapel Hill. Of the 572 subjects who were lost to follow-up during the initial survey, 155 were lost due to death (27%), whereas the rest were not found or were no longer living in Metro Cebu. Of the 244 subjects lost between the end of the initial survey and the 1991 follow-up, 55 died (23%). The remaining losses in this and subsequent periods were attributable primarily to moving outside of the Metro Cebu study area. In the longitudinal models described below, these data contribute 4047 observations in boys and 3656 observations in girls.
Anthropometric measures.
During the initial study period (from birth through 2 y of age), recumbent length was measured to the nearest millimeter using a custom-designed length measuring board. During the follow-up surveys, standing height was measured using a portable stadiometer. All measurements were performed in the children’s homes by trained project personnel using standard techniques. Maternal height was measured at baseline using the same techniques. Height-for-age Z-scores (HAZ) were calculated using the lambda, mu, and sigma parameters (the power transformation, median, and CV, respectively, from the Box-Cox transformation) from the 2000 CDC growth reference data (25,26). The new references were chosen because they are more appropriate for breast-fed children and because they provide reference data past 18 y of age, which is necessary for calculating HAZ from the 2002 CLHNS follow-up.
Maturational timing.
Timing of sexual maturation was assessed by questionnaire using age at menarche in girls and pubic hair stage in boys. Girls were asked their age at menarche at each follow-up survey from age 11.5 y on. Pubic hair development was self-assessed by having boys compare themselves to line drawings depicting the 5 Tanner stages of development (27) at both 15.5 and 18.5 y of age and was validated against physician assessment in a separate but similar small sample that was recruited for that purpose. For ease of analysis, and because there were few subjects at the extremes, pubic hair development was categorized as early (Tanner stages 4 and 5), average (Tanner stage 3), and late (Tanner stages 1 and 2) based on the 1998 survey data at mean age 15.5 y.
SES index.
The SES index used was developed for the Cebu data, and reflected 3 facets of SES, i.e., income, assets, and maternal education. One point was awarded for ownership of each of 4 assets: air-conditioning, television (color or black and white), refrigerator, or vehicle (car, bus, truck, or motorized tricycle). Additionally 1, 2, or 3 points were awarded in concordance with falling into the lowest, middle, or highest tertile of per capita income, respectively. Last, 1 point was awarded if the mother had completed primary school, and 2 points were awarded if the mother had any schooling beyond primary school. Thus, the SES index has a potential range of 0–9 points. The outcome data in the longitudinal models used and described below are height measures at 8.5, 11.5, 15.5, and 18.5 y of age. We used the mean SES index from survey points 2 and 8.5, 8.5 and 11.5, 11.5 and 15.5, and 15.5 and 18.5 y of age, respectively, as predictor variables relating to these outcomes to better describe the mean SES influences experienced by the subjects between surveys and leading up to the outcome measures included.
Diet variety.
We include a DV score in our models because many micronutrients are important for supporting linear growth, yet intakes of micronutrients are highly correlated, making it difficult to examine their independent effects. The literature shows that DV is associated with overall nutrient adequacy, making it a useful measure of diet quality (28–30).
Because several nutrients important for growth (such as zinc) cannot be isolated from the Cebu data because they are not included in the Philippine food composition tables (31), we used a food group–based DV index, a common approach (28–30,32–34), rather than a nutrient-based index. We based our score on a DV score created previously for the Cebu population (35) in which 1 point was awarded for having eaten at least 1 food from each of several categories: fish, animal source foods, staple cereals, other starches, vegetables, fruits, beans and nuts, and dairy.
A variety of dietary assessment tools were used during the Cebu study. An FFQ covering a recall period of 1 y was administered at 8.5 y of age, one 24-h recall was administered at 11.5 y of age, and two 24-h recalls were administered on nonconsecutive days at each of the survey points thereafter. Portion sizes for both the FFQ and the 24-h recalls were ascertained using the same set of field-tested food models at each survey point. Mothers or guardians were the respondents for the children through age 11.5 y, whereas the subjects responded for themselves thereafter. The differences between the dietary assessment tools used imposed some difficulties in creating comparable measures of dietary variety over time. The FFQ is preferable for representing usual intake (36), but because it reports intakes based on a comprehensive list of 77 foods, it is likely to show greater dietary variety than the 24-h recalls. For the 24-h recall data, greater variety would likely be reflected through the use of data from two 24-h periods (at ages 25.5 and 18.5 y) than from one 24-h period (at age 11.5 y). In addition, 24-h recalls may not represent usual intake patterns (37).
An additional independent set of questions was administered in the same way at each time point, listing what foods were usually eaten for breakfast, lunch, dinner, and snacks. Thus, we used the responses from these "usual intake" questions to construct a comparable DV score at 8.5, 11.5, 15.5, and 18.5 y of age with a possible range of 0 to 8 points. Mendez (35) demonstrated the validity of this method within the Cebu data by validating the DV score constructed from these additional "usual intake" questions against the nutrient intake data from the more standard 24-h recall data at 11.5 y of age, and found that the DV score was strongly associated with nutrient adequacy for a variety micronutrients such as iron, calcium, and vitamin A (35).
Energy intake.
We use kilocalories (kcal)4 per day as calculated from the FFQ at 8.5 y of age and from the 24-h recalls at the later survey points, recognizing that the energy intakes are likely to be somewhat inflated at 8.5 y of age but that the association between increases in intake and increases in height is assumed to be the same. The use of energy tertiles (to overcome the differences due to the discrepancies in methods) did not substantively change our findings; thus we used the continuous measure energy intake from 8.5, 11.5, 15.5, and 18.5 y of age.
Statistical analysis.
Frequencies and percentages for maturational timing in boys, and means and SD for height, HAZ, SES index, DV score, and energy intake for each survey point from which the height outcome measures are taken (ages 8.5, 11.5, 15.5, and 18.5 y of age) were determined. Means and SD for height and HAZ at 2 y of age were also calculated because height at 2 y is included in our models as a control. Although the HAZ data were not used in our models, they are provided to show the extent of linear growth retardation in the Cebu population.
We fit a longitudinal model using Generalized Estimating Equations (GEE) predicting height (using outcome data from 8.5, 11.5, 15.5, and 18.5 y of age). We selected this type of "population average" regression model because we were not interested in estimating subject-specific parameters. GEE models also account for correlations of repeated measures within individuals, allow for unequally spaced longitudinal measures, and maximize sample size and power by accommodating missing observations (38).
We modeled the mean effects of several time-varying and non-time-varying predictors on height across the postinfancy growth period. Thus, the outcome measure of height comes from 8.5, 11.5, 15.5, and 18.5 y of age. The time-varying predictors included our main variables of interest, i.e., DV and energy intake. The time-varying continuous SES index variables were also included to control for confounding in the diet/growth association because SES likely affects food availability and quality, and also has effects on growth via factors such as morbidity and access to health care. The diet variables were taken from the same survey from which the outcome was drawn rather than using mean values between survey points, as was done with the SES data. Because the surveys were spaced several years apart and energy needs and diet change significantly during childhood, concurrent diet was considered to be a more accurate representation of recent diet affecting height than a mean measure. Age and age-squared variables were also included to account for the nonlinear association between age and height.
The non-time-varying predictors included mother’s height as a proxy for genetic influences and generational effects, maturational timing, and height at 2 y of age. Maturational timing variables (age at menarche for girls and pubic hair development stage at age 15.5 for boys) were included because although these variables refer specifically to events that occur in puberty, they reflect developmental trajectories that begin much earlier and which may influence the tempo of growth-attained stature. Height at 2 y of age was included to control for height at the onset of the postinfancy period and to control for the effects of diet and other predictors of growth during infancy, thus allowing our models to focus on independent postinfancy effects.
Lowess smoothers (39) were used to identify any gross violations of the linearity assumption between the dependent and independent variables for each survey point. Three boys with energy intakes > 6000 kcal/d (1 at 15.5 y of age and 2 at 18.5 y of age) were excluded from the analysis because their reported intakes were considered implausible and because they were identified as outliers in the height/energy intake association.
Interactions between the SES index and both the DV and the energy intake variables were explored to determine whether individuals of low SES might be more responsive to dietary improvements. Interactions of both the DV and the energy intake variables with the age and age-squared terms were tested to determine whether the associations between the diet variables and linear growth varied with age.
Coefficients with P < 0.05 were considered significant for main effects, whereas P < 0.15 was considered significant for interaction terms. All analyses were performed in Stata for Windows (Release 8.0, Stata Corporation).
|
RESULTS |
|---|
|
TOPABSTRACTSUBJECTS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
The mean HAZ was below the level of stunting (less than –2.0) at ages 2 and 8.5 y in both boys and girls with some recovery from stunting thereafter (Table 1). The mean HAZ measures at 18.5 y, which likely reflect final stature for the majority of the population, were above, although still near the cutoff point indicating stunting. These data thus indicate widespread linear growth retardation, with some improvements with age, in the Cebu population. The SES indices improved in both boys and girls over time, reflecting the rapid modernization and economic development occurring in the Philippines. The DV scores remained relatively stable over time in both boys and girls, whereas energy intake data showed inflated intakes at age 8.5 y (due to the use of a FFQ at this survey point as opposed to the 24-h recalls at the other survey points) with increasing intakes in both boys and girls from 11.5 to 18.5 y of age.
|
The coefficients and SE from the main effects models, without interaction terms included, are shown in Table 2. Energy intake and mother’s height were positively associated with height in the postinfancy period in both boys and girls. The coefficient for DV was positive and significant in boys, whereas there was virtually no DV effect seen in girls. Early maturation was associated with greater height in boys and girls as seen by the positive significant coefficient for the early maturation variable in boys and by the significant negative association between age at menarche and height in girls.
|
Although the mean daily energy intake tended to rise with SES index level at each age (as would be expected), there was sufficient overlap of energy/intake among SES levels to merit testing interactions between energy intake and SES. The correlation between kcal/d and the SES index was 
0.40 at ages 8.5 and 11.5 and weakened thereafter. Diet variety also tended to rise with SES at young ages, with a correlation of 0.35 at age 8.5, but the correlation was very weak by age 18.5 (<0.10). Interactions of both DV and energy intake with the continuous variable SES index were tested. In both boys and girls, only the interactions with energy intake were significant (P < 0.15) and were retained in the models with results shown in Table 3. To better understand the potential magnitude of this effect in the Cebu population, we used the model coefficients to calculate the change in height/100 kcal increase in energy intake at a high value vs. a low value of SES, stratified by sex (Fig. 1). In this example, the high and low values of SES were calculated by averaging the SES index across age, and then taking the mean + 1 SD (high value of SES) and the mean – 1 SD (low value of SES) in boys and girls separately. In both boys and girls, each 100 kcal increase in energy intake was associated with a 0.08 cm increase in height at the low values of SES, whereas it was associated with no difference in height (0.03 and –0.02 cm in boys and girls, respectively) at the high values of SES.
|
|
For the interactions between the age variables and the diet variables (both DV and energy intake), only the interactions with energy intake were significant. The results from the models including the age interactions are shown in Table 4. We used the model coefficients to calculate the effect of a 100 kcal increase in energy intake at the mean age from each of the survey points (Fig. 2). The pattern indicated that a 100 kcal increase in energy intake was positively associated with height at ages 11.5 and 15.5 y, with effects of 0.25 cm, and was negatively associated with height at 8.5 and 18.5 y of age in boys. In girls, the association was positive at 8.5, 11.5, and 15.5 y of age with effects decreasing in magnitude, and was negative at 18.5 y of age. The negative effects seen in both boys and girls were very small, ranging from 0.02 cm at age 8.5 y in boys to 0.10 and 0.06 cm in boys and girls at 18.5 y of age, respectively.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTSUBJECTS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
The main effect models, without interactions included, generally showed a positive association of energy intake with height during the postinfancy period. Although the coefficient for the energy intake variable was small, it should be remembered throughout the examination of the results that the coefficients for energy intake reflect the effect of only a 100 kcal increase in energy intake in a population in which 1 SD in energy intake at all the survey points included in the analysis was >500 kcal. In other words, variation from the mean by only 1 SD of energy intake for the population would be associated with a difference in height on a scale 5 times that shown for a 100 kcal difference. DV was significant in boys but not girls.
The coefficient for mother’s height was positive and significant in both boys and girls. Although mother’s height was included in the models as our best available proxy for genetic potential, it is not a pure marker for the role of genetics. In a developing country context such as the Philippines, the variability in height in all subjects is a product of SES, nutrition, and infection, in addition to genetics. Thus, although the significance of the mother’s height may suggest the importance of genetics for growth in the postinfancy period, it should be considered more as a control for generational effects that may include similarities in household socioeconomic conditions and other influences affecting both mother and child.
The significance and direction of the coefficients for the maturational timing variables demonstrate their influence on stature, indicating that earlier maturation was generally associated with increased height. Although very early maturation in some populations has been associated with short adult stature, it should be noted that these cases refer to menarche as early as 11.6 y of age in girls, for example, which truncates the normal duration of the childhood growth phase and the growth period overall (40). "Early" maturation in this paper, however, indicates maturation with regard to the mean age of the Cebu population. For example, mean age at menarche for the Cebu girls is 13.1 y of age, 1 y later than for girls in the United States (41). Thus "early" maturation in the context of the Cebu population is not early in the extreme, and would not be expected to be accompanied by abbreviated childhood growth.
It should be noted that although diet and maturation clearly had effects on height in the postinfancy period, height at 2 y of age was significant in both boys and girls, showing that each additional centimeter of height at age 2 y was associated with almost 1 cm of additional height (0.99 cm in boys and 0.88 cm in girls) at each survey point thereafter. This finding emphasizes the well-known importance of growth in the infancy period for determining subsequent height.
The interactions of the SES index with energy intake, significant in both boys and girls, showed an association of increased energy intake with larger increases in height at low levels of SES than at high SES. This indicates that a population likely characterized by low access to resources and poor health has more to gain from improvements in energy intake, whereas an affluent population has likely already attained stature close to the maximum of their genetic capacity and is thus less responsive to dietary improvements.
In girls, the interactions of the age terms with energy intake showed a pattern in which energy intake was positively associated with height at 8.5, 11.5, and 15.5 y of age, with effects decreasing in magnitude, and was slightly negatively associated with height at 18.5 y of age. Thus, energy intake had the greatest effect early in the postinfancy period when growth velocity was still relatively fast and height more easily affected. In boys, the pattern showed small negative associations with height at 8.5 and 11.5 y of age and large positive effects at 15.5 and 18.5 y of age. When a multivariate linear regression (not shown) modeling the data at age 8.5 was run (in other words, modeling height at age 8.5 as a function of the other variables included in the GEE models at the same age and controlling for height at 2 y), energy intake was not significant in boys (P = 0.08). It did not gain significance with the removal of the DV variable (P = 0.07), but did gain significance with the removal of the SES index variable (P = 0.03). Thus, the negative effect seen at age 8.5 in boys in the longitudinal model was really an absence of effect due to an enhanced association of energy intake variable with both the SES index and DV at this young age. As discussed previously, the diet data at 8.5 y of age was collected differently than it was at the other survey points (FFQ vs. 24-h recalls, respectively). We know that the energy intake values from the FFQ are somewhat inflated compared with those from the 24-h recalls. It could also be that the use of the FFQ elicited overreports of energy intake from the mothers of shorter individuals (perhaps to try and cover inability to adequately provide?) and/or underreports of energy intake from taller individuals, thus misclassifying intakes and veiling the true effects; however, it is difficult to envision how this type of error would be so sex specific.
We also further investigated the small negative association between energy intake and height at age 18.5 in both boys and girls using multivariate linear regressions (not shown). Energy intake was not significant in either boys or girls (P = 0.43 and P = 0.08, respectively), nor did it gain significance with the removal of DV (P = 0.49 and P = 0.09, respectively) or the SES index (P = 0.46 and P = 0.10, respectively). Thus, the small negative effects seen at 18.5 y of age were negligible and indicate no association between energy intake at 18.5 y of age, an age by which most subjects had attained their final adult stature.
As mentioned earlier, a limitation of this study was the lack of a good measure of morbidity to account for its effects on stature and to control for potential confounding of the diet/linear growth association. Although some morbidity measures were collected at each survey point, the types of questions asked were different from survey to survey, and thus a comparable morbidity measure (comparability at each survey point being a requirement for GEE model specification) was not possible. Even if a comparable measure could have been constructed from the questions asked, the questions did not adequately reflect the total morbidity influences experienced in the several years between surveys and thus would not have been valid measures.
The difference in the type of diet data collected at 8.5 y vs. the other survey points (FFQ vs. 24-h recalls, respectively) was another limitation. As discussed previously, the energy intake values derived from the FFQ were inflated compared with those from the 24-h recalls, as can be seen by examining the trend of energy intake with age in Table 1. This inflation likely resulted from attempting to derive daily intakes from the comprehensive list of foods included in the FFQ rather than from a strict 24-h recall. The unexpected lack of association between energy intake and height at 8.5 y of age may be due in some way to the quality of that energy intake data.
Although the DV variety score on which our score was based was used previously and validated at 11.5 y of age, a further investigation of the association between the DV measure and several nutrients revealed that the DV score, although constructed in the same way at each survey point, does not describe nutrient adequacy comparably across the surveys. The correlation of the DV score and various nutrient intakes, as calculated from the FFQ at 8.5 y of age and the 24-h recalls thereafter, revealed a diminishing association of DV with nutrient intakes with age. For example, the correlations of DV with energy, protein, and iron intake were all 0.33 at age 8.5 y in boys. By age 18.5 y, these correlations had decreased to 0.11, 0.11, and 0.12, respectively. In girls the correlations for energy, protein, and iron were 0.39, 0.40, and 0.42 at age 8.5 y, respectively, and were 0.07, 0.07, and 0.09 by age 18.5 y. This diminishing association of DV with nutrient adequacy with age may account for the lack of a significant association across the surveys in girls (Table 2). However, even the interactions of the age variables with DV were not significant (P > 0.15), indicating that DV, even when its correlation with nutrient adequacy was highest at the younger ages, was not a good predictor of height. To test whether DV was not significant due to the inclusion of SES and energy intake in the models, which may be capturing the DV effects through their potential association with DV, we ran the models again, first removing SES, and then removing energy intake. In both cases, the coefficients for DV remained virtually the same. Thus, it seems that the ability of the DV score to capture a distribution of nutrient adequacy decreases with age, perhaps due to diets that grow increasingly varied and difficult to distinguish through only 8 broad food groups. The Cebu DV score was shown to be associated with linear growth in children at 2 y of age (35), controlling for sex and other confounders. In this study, it was also associated with height in boys, but not in girls. This difference by sex may be because linear growth slows and stops sooner in girls than in boys, leaving less potential for growth effects in girls.
As described earlier, attrition of the CLHNS sample occurred over time. About 25% of loss to follow-up between birth and the 1991 survey (mean age 8.5 y) was due to death; the rest was due to migration out of the study area. Attrition after 8.5 y of age was due almost entirely to outmigration. Those that died tended to be of very low SES and grew poorly up until their deaths, whereas those that moved tended to be of higher SES and from urban areas. Although loss of subjects at either end of the SES spectrum could diminish the generalizability of our findings to the population as a whole, any effects are likely to be minimal because only a proportion of those at the ends of the SES distribution were lost.
The strengths of this research are many. In particular, the CLHNS study offers the most comprehensive data to date on both diet and height, the expression of linear growth, in a developing country context over the entirety of the postinfancy period, a phase of growth that is seldom studied. The longitudinal methods we employed maximized power in the data and made it possible to accommodate missing data and attrition. The ability of interventions to improve linear growth after infancy has ended is often questioned. Here we show that diet can continue to have independent positive effects on stature even beyond infancy. Specifically, nutrition interventions may be effective independently of socioeconomic changes, but should be provided as early as possible, with special attention given to the prepubertal growth spurt and to populations at the lower end of the SES distribution.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
3 Abbreviations used: CLHNS, Cebu Longitudinal Health and Nutrition Survey; DV, diet variety; GEE, generalized estimating equation; HAZ, height-for-age Z-score; SES, socioeconomic status.
Manuscript received 13 November 2004. Initial review completed 2 January 2005. Revision accepted 7 July 2005.
|
LITERATURE CITED |
|---|
|
TOPABSTRACTSUBJECTS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
1. Frongillo, E. A. (1999) Introduction. Symposium: Causes and Etiology of Stunting. J. Nutr. 129:529S-530S.
2. Stephensen, C. B. (1999) Burden of infection on growth failure. J. Nutr. 129:534S-538S.
3. Waterlow, J. C. (1994) Summary of causes and mechanisms of linear growth retardation. Eur. J. Clin. Nutr. 48:S210.
4. Keller, W. (1988) The epidemiology of stunting. Waterlow, J. C. eds. Linear Growth Retardation in Less Developed Countries, Nestle Nutrition Workshop Series Vol. 14:17-39 Raven Press New York, NY.
5. UNICEF (2004) The epidemiology of stunting. The State of the World’s Children 2004 UNICEF House New York, NY.
6. Pelletier, D. L., Frongillo, E. A., Schroeder, D. G. & Habicht, J. P. (1995) The effects of malnutrition on child mortality in developing countries. Bull. WHO 73:443-448.[Medline]
7. Bairagi, R., Chowdury, M. K., Kim, Y. G. & Curlin, G. T. (1985) Alternative anthropometric indicators of mortality. Am. J. Clin. Nutr. 42:296-306.
8. Mendez, M. A. & Adair, L. S. (1999) Severity and timing of stunting in the first two years of life affect performance on cognitive tests in late childhood. J. Nutr. 129:1555-1562.
9. Brown, J. L. & Pollitt, E. (1996) Malnutrition, poverty, and intellectual development. Sci. Am. 274:38-43.[Medline]
10. Haas, J. D., Murdoch, S., Rivera, J. & Martorell, R. (1996) Early nutrition and later physical work capacity. Nutr. Rev. 54:S41-S48.[Medline]
11. Pollitt, E., Gorman, K. S., Engle, P. L., Rivera, J. A. & Martorell, R. (1995) Nutrition in early life and the fulfillment of intellectual potential. J. Nutr. 125:111S-1118S.
12. The World Bank (1994) Annex B: Life cycle of women’s health. Tinker, A. Daly, P. Green, C. Saxena, G. Lasminarayan, R. Gill, K. eds. Women’s Health and Nutrition—Making a Difference Vol. 256:59-74 The World Bank Washington, DC.
13. Spurr, G. B. (1988) Body size, physical work capacity, and productivity in hard work: is bigger better?. Waterlow, J. C. eds. Linear Growth Retardation in Less Developed Countries, Nestle Nutrition Workshop Series Vol. 14:215-243 Raven Press New York, NY.
14. Obatolu, V. A. (2003) Growth pattern of infants fed with a mixture of extruded malted maize and cowpea. Nutrition 19:174-178.[Medline]
15. Simondon, K. B., Simondon, F., Costes, R., Delaunay, V. & Diallo, A. (2001) Breast-feeding is associated with improved growth in length, but not weight, in rural Senegalese toddlers. Am. J. Clin. Nutr. 73:959-967.
16. Rosado, J. L. (1999) Separate and joint effects of micronutrient deficiencies on linear growth. J. Nutr. 129:531S-533S.
17. Onyango, A., Koski, K. G. & Tucker, K. L. (1998) Food diversity versus breastfeeding choice in determining anthropometric status in rural Kenyan toddlers. Int. J. Epidemiol. 278:484-489.
18. Fawzi, W. W., Herrera, M. G., Willett, W. C., Nestel, P., el Amin, A. & Mahamed, K. A. (1987) Dietary vitamin A intake in relation to child growth. Epidemiology 8:402-407.
19. Marquis, G. S., Habicht, J. P., Lanata, C. F., Black, R. E. & Rasmussen, K. M. (1997) Breast milk or animal-product foods improve linear growth of Peruvian toddlers consuming marginal diets. Am. J. Clin. Nutr. 66:1102-1109.
20. Rogers, I. S., Emmett, P. M. & Golding, J. (1997) The growth and nutritional status of the breast-fed infant. Early Hum. Dev. 49:S157-S174.
21. Schroeder, D. G., Martorell, R., Rivera, J. A., Ruel, M. T. & Habicht, J. P. (1995) Age differences in the impact of nutritional supplementation on growth. J. Nutr. 125:1051S-1059S.
22. Neumann, C. G. & Harrison, G. G. (1994) Onset and evolution of stunting in infants and children. Examples from the Human Nutrition Collaborative Research Support Program. Kenya and Egypt studies. Eur. J. Clin. Nutr. 48:S90-S102.
23. Martorell, R., Kettel Khan, L. & Schroeder, D. G. (1994) Reversibility of stunting: epidemiological findings in children from developing countries. Eur. J. Clin. Nutr. 48:S45-S57.
24. Allen, L. H. (1994) Nutritional influences on linear growth: a general review. Eur. J. Clin. Nutr. 48:S75-S89.
25. CDC (Centers for Disease Control and Prevention) () Nutritional influences on linear growth: a general review. http://www.cdc.gov/nchs/about/major/nhanes/growthcharts/datafiles.htm [accessed October 15, 2001].
26. Cole, T. (1990) The LMS method for constructing normalised growth standards. Eur. J. Clin. Nutr. 44:45-60.[Medline]
27. Tanner, J. M. (1962) The LMS method for constructing normalised growth standards. Growth at Adolescence 2nd ed. Blackwell Scientific Oxford, UK.
28. Ruel, M. T. (2003) Operationalizing dietary diversity: a review of measurement issues and research priorities. J. Nutr. 133:3911S-3926S.
29. Kant, A. (1996) Indexes of overall diet quality: a review. J. Am. Diet. Assoc. 96:785-791.[Medline]
30. Kant, A. K., Schatzkin, A., Block, G., Ziegler, R. G. & Nestle, M. (1991) Food group intake patterns and associated nutrient profiles of the US population. J. Am. Diet. Assoc. 91:1532-1537.[Medline]
31. Food and Nutrition Research Institute (1990) Food group intake patterns and associated nutrient profiles of the US population. Food Composition Tables Recommended for Use in the Philippines, Handbook 1 6th rev. Department of Science and Technology Manila, Philippines.
32. Haines, P. S., Siega-Riz, A. M. & Popkin, B. M. (1999) The Diet Quality Index revised: a measurement instrument for populations. J. Am. Diet. Assoc. 99:697-704.[Medline]
33. McCrory, M. A., Fuss, P. J., McCallum, J. E., Yao, M., Vinken, A. G., Hays, N. P. & Roberts, S. B. (1999) Dietary variety within food groups: association with energy intake and body fatness in men and women. Am. J. Clin. Nutr. 69:440-447.
34. Krebs-Smith, S. M., Smiciklas-Wright, H., Guthrie, H. A. & Krebs-Smith, J. (1987) The effects of variety in food choices on dietary quality. J. Am. Diet. Assoc. 87:897-903.[Medline]
35. Mendez, M. A. (2000) The effects of variety in food choices on dietary quality. Resilient Growth and Intellectual Development in Filipino Children. Doctoral thesis University of North Carolina Chapel Hill, NC.
36. Willett, W. (1998) Food frequency methods. Willett, W. eds. Nutritional Epidemiology :74-100 Oxford University Press New York, NY.
37. Buzzard, M. (1998) 24-hour dietary recall and food record methods. Willett, W. eds. Nutritional Epidemiology :50-73 Oxford University Press New York, NY.
38. Zeger, S. L. & Liang, K. Y. (1986) Longitudinal data analysis for discrete and continuous outcomes. Biometrics 42:121-130.[Medline]
39. Cleveland, W. S. (1979) Robust locally weighted regression and smoothing scatterplots. J. Am. Stat. Assoc. 74:829-836.
40. Proos, L. A., Hofvander, Y. & Tuvemo, T. (1991) Menarcheal age and growth pattern of Indian girls adopted in Sweden. Acta Paediatr. Scand. 80:852-858.[Medline]
41. Parent, A. S., Teilmann, G., Juul, A., Skakkebaek, N. E., Toppari, J. & Bourguignon, J. P. (2003) The timing of normal puberty and the age limits of sexual precocity: variations around the world, secular trends, and changes after migration. Endocr. Rev. 24:668-693.
This article has been cited by other articles:
|
|
 |
|
  N. Mpontshane, J. Van den Broeck, M. Chhagan, K. K. A. Luabeya, A. Johnson, and M. L. Bennish HIV Infection Is Associated with Decreased Dietary Diversity in South African Children J. Nutr., September 1, 2008; 138(9): 1705 - 1711. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
