SEARCH

SEARCH BY CITATION

Keywords:

  • stunting;
  • body composition;
  • infancy;
  • childhood;
  • BMI

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References

Objective: To determine whether African urban children who were stunted at 2 years of age demonstrated an altered body composition by the end of childhood, before entering puberty, at 9 years of age.

Research Methods and Procedures: This was a mixed-longitudinal study of 330 prepubertal African children (182 boys) from Soweto-Johannesburg, South Africa. Anthropometric data at 2 years of age were compared with anthropometric, DXA-determined body composition and fat patterning in late childhood (7 to 9 years).

Results: Children who had been stunted at 2 years were significantly shorter and lighter than non-stunted children at 7 to 9 years, but there were no differences in their BMI or centralization of body fat. Previously stunted status significantly predicted reduced weight and height at 7 to 9 years but did not predict BMI, body composition, or fat patterning after controlling for potential confounding factors. The odds ratio for stunting at 2 years as a predictor of overweight at 7 to 9 years was not significant at 1.09 (95% confidence limits: 0.30, 3.98).

Discussion: Greater BMI in stunted infants does not demonstrate a tendency toward overweight or obesity but is a reflection of the greater reduction in height rather than weight in stunted children. Stunted children may be programmed to accumulate greater body fat at central sites during adolescence, but we have been unable to show that these changes are evident before the initiation of pubertal development.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References

By definition, ∼2.5% of children in developed countries are shorter than −2 z scores and are, thus, classified as being stunted. In southern Africa, almost 20% of children in urban and rural areas are stunted by 2 years of age (1, 2). Research in the last decade, notably by Popkin et al. (3), provided evidence for concurrent stunting and overweight or obesity in some developing countries. These countries (South Africa, Russia, China, and Brazil) were linked by their classification as transitional countries in that they were transiting through both economic and social changes. Within child development, these changes were manifested by dietary patterns that had shifted from the traditional diets composed of high-fiber, low-fat foods to western diets of low-fiber, high-fat, high-energy, low-cost foods. It was subsequently hypothesized that the combination of nutritional stunting and exposure to cheap high-fat foods was a major contributing factor to concurrent stunting and overweight/obesity. Even though this scenario has gained general acceptance, few studies have been able to reproduce the concurrent stunting and obesity scenario in South Africa (2).

Other research set within the urban environments of developing countries has suggested that stunting in infancy may increase the future risk of greater fatness and overweight in later childhood (4). It is within the urban environment that transition is most evident. These environments are characterized by a rapidly increasing population density and concomitant changes from traditional low-fat to western high-fat diets, which may interact in poorer families, in which stunting is endemic, to increase the susceptibility to excess body fat gain in children who are stunted (5, 6). Sawaya et al. (5), for example, described a prevalence of 30% undernutrition (low height-for-age, low weight-for-age) among children in a shanty town in Sao Paulo, Brazil compared with concurrent prevalences of 21% and 8.6% for adolescent female and male overweight, respectively. There is some evidence to suggest that physiological mechanisms promote the accretion of body fat rather than protein after nutritional deprivation (7, 8, 9).

However, the majority of research into stunting and obesity in children is cross-sectional and, thus, does not follow the same children through childhood and adolescence. Changes in growth and body composition over time can, thus, only be inferred through implication and not measured within the same children. Only one longitudinal study has attempted to elucidate the effects of early growth on body composition in mid- to late childhood. Walker et al. (10) studied the effects of birth weight and postnatal linear growth retardation (i.e., stunting) on BMI, total body fat, and fat distribution in a longitudinal study of 116 stunted children and 190 normal children in Kingston, Jamaica. In contrast to previous cross-sectional analyses, they concluded that there was no increased risk of overweight at 7 to 11 years in previously stunted children. However, the study contains a number of methodological problems that bring these results into question. Walker estimated total body fat using the prediction equations for prepubertal and pubertal black children developed by Slaughter et al. (11), which have previously been tested on African prepubertal children and found to be unsuitable (12). In addition, Walker's results were, perhaps, confounded by the fact that 60% of the non-stunted group and 40% of the stunted group were already pubertal and were, thus, likely to be exhibiting consequent changes in the magnitude and distribution of body fat.

The current longitudinal study sought to determine the association between stunting in early childhood and overweight and fatness in a sample of prepubertal children 9 years old. Uniquely, we were able to use data from DXA to estimate body composition at 9 years in addition to standard body dimensions determined by anthropometry. Thus, we asked whether African urban children who were stunted at 2 years of age demonstrated an altered body composition by the end of childhood, before entering puberty, at 9 years of age.

Research Methods and Procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References

The Birth to Twenty (BT20)1 study is a longitudinal study tracking a cohort of 3273 children of all ethnic groups born between April and June 1990 living in Soweto and Johannesburg, South Africa and has been described in detail elsewhere (13, 14). Briefly, the mothers of singleton infants were recruited in the perinatal period, and the infants were followed from birth. They were assessed at approximate yearly intervals throughout childhood. Gestational age was determined by completed weeks from the last menstrual period, which was obtained in all cases from clinical sources (delivery records, etc.) compiled by midwifery or obstetric staff. A randomly selected subsample of children from this study was used to undertake a separate bone health study from the age of 9 years.

A mixed-longitudinal sample of 330 prepubertal African children (182 boys) was used for the analysis. All participants had weight data at birth, 2 years, and at some time between 7 and 9 years and height data at 2 and 7 to 9 years. Triceps and subscapular skinfolds were available on 126 children at 2 years and on all children at 7 to 9 years. DXA-determined body composition at 9 years was available on 136 of these children. There were no significant differences in anthropometric variables between those with and without skinfold and/or DXA data. All children were prepubescent, i.e., in Tanner's stage 1, for both breast/genitalia and pubic hair.

Anthropometric measurements were taken using standard techniques (15). These included weight from birth, length/height from 1 year, and skinfold measurements (using a Holtain Tanner/Whitehouse skinfold caliper) at the triceps and subscapular sites from 2 years of age. Within the bone health study, skinfolds at the biceps, suprailiac, midthigh, and medial-calf sites were added. Each skinfold measurement was taken three times, and a mean was calculated. DXA (Hologic 4500A) provided data on bone mineral content (grams), lean tissue mass (kilograms), fat mass (kilograms), and percentage body fat. Pubertal development was assessed by trained same-sex observers using the Tanner scaling technique on breasts/genitalia and pubic hair. Stunting was defined as a height-for-age status of less than −2 z scores at 2 years of age according to the Centers for Disease Control and Prevention (CDC) 2000 growth references.

Data were analyzed using logistic and linear regression models (SPSS, Inc., Chicago, IL) with stunting at 2 years as a predictor of weight, overweight, triceps and subscapular skinfolds, and centripetal fat ratio (CFR; =subscapular/subscapular + triceps) at 2 years and BMI, percentage lean, percentage fat mass, and CFR at 9 yrs. All models were controlled for sex and age at measurement, and for whether the child had been small for gestational age (SGA). The latter correction was used because SGA is known to be associated with stunting during infancy and childhood and, thus, if it were not controlled, would have a confounding effect on the analysis. The dichotomous variable of SGA rather than the continuous variable of birth weight was used because birth weight per se is not associated with stunting, but birth weights within the SGA category are associated. It seemed more reasonable, therefore, to control for SGA status rather than birth weight. SGA was defined as a birth weight less than the 10th centile of the CDC reference charts for gestational age in accordance with the common practice of classifying SGA as a birth weight inferior to a particular centile (16, 17) and, in particular, the practice of using the 10th centile (18).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References

Fifty-three (16.1%) of the children were stunted at 2 years of age and were not only significantly shorter but also significantly lighter than their non-stunted peers (p < 0.001) (Table 1). They were also significantly younger, but by only 0.03 years on average, which is equivalent to 12 days, and has not been regarded as biologically significant. Stunted children had significantly lower triceps skinfolds and lower, but not significantly lower, subscapular skinfolds. However, they presented with significantly greater BMI but no differences in their fat distribution or CFR.

Table 1.  Means for age, weight, height, BMI, CFR, and triceps and subscapular skinfolds at 2 and at 7 to 9 years for stunted and non-stunted children at age 2
 
 Stunted (n = 53) [mean (SD)]Non-Stunted (n = 277) [mean (SD)]Stunted (n = 53) [mean (SD)]Non-Stunted (n = 277) [mean (SD)]
  • *

    p < 0.05.

  • p < 0.001.

  • p < 0.01,

  • §

    At age 2, n = 14 for stunted and n = 112 for non-stunted.

Age (years)2.13 (0.09)2.10 (0.09)*8.45 (0.5)8.35 (0.5)
Weight (kg)10.4 (1.4)11.7 (1.7)23.4 (2.9)25.6 (4.1)
Height (cm)77.3 (2.9)84.0 (2.9)122.3 (5.9)126.3 (6.4)
BMI (kg/m2)17.6 (2.7)16.6 (2.2)15.6 (1.4)16.0 (1.8)
CFR§0.53 (0.05)0.52 (0.07)0.37 (0.03)0.37 (0.04)
Triceps skinfold (mm)§5.3 (0.7)6.2 (1.6)8.70 (1.92)9.42 (3.24)*
Subscapular skinfold (mm)§6.1 (1.2)6.7 (1.4)5.05 (1.04)5.59 (1.99)

Stunting at 2 years was a significant predictor of lower weight and greater BMI at 2 years but did not significantly predict subcutaneous fat values or fat distribution after controlling for the potential confounding factors (Table 2). Of the 53 stunted children at 2 years of age, 18 (5.5%) remained stunted at 7 to 9 years. The 53 previously stunted children continued to be significantly shorter and lighter than non-stunted children at 7 to 9 years of age, but there were no differences in their BMI or CFR (Table 1).

Table 2.  Stunting at 2 years as a predictor of body composition dimensions at age 2 and 7 to 9 years
 Triceps skinfold at 2 years* [β (SE)] (n = 126)Subscapular skinfold at 2 years* [β (SE)] (n = 126)CFR at 2 years* [β (SE)] (n = 126)Sum of skinfolds at 2 years* [β (SE)] (n = 126)Weight at 2 years* [β (SE)] (n = 330)BMI at 2 years* [β (SE)] (n = 330)
  • *

    All models control for age at 2 year measurement, sex, and SGA. Models triceps, subscapular, CFR and sum of skinfolds also control for weight at 2 year measurement.

  • All models control for age at 2 year measurement, sex, SGA, and age at 7 to 9 year measurement. Models triceps, subscapular, CFR, and sum of skinfolds also control for weight at 7 to 9 year measurement.

  • p < 0.001.

Stunted−0.524 (0.461)0.287 (0.366)0.027 (0.019)−0.238 (0.654)−1.10 (0.250)1.247 (0.341)
Triceps skinfold at 7 to 9 years [β (SE)] (n = 330)Subscapular skinfold at 7 to 9 years [β (SE)] (n = 330)CFR at 7 to 9 years [β (SE)] (n = 330)Sum of skinfolds at 7 to 9 years [β (SE)] (n = 330)Weight at 7 to 9 years [β (SE)] (n = 330)BMI at 7 to 9 years [β (SE)] (n = 330)Height at 7 to 9 years [β (SE)] (n = 330)
0.348 (0.354)0.039 (0.231)0.007 (0.006)0.388 (0.544)−2.157 (0.562)−0.191 (0.267)−0.046 (0.009)

Previously stunted status significantly predicted reduced weight and height at 7 to 9 years but did not predict BMI, subcutaneous fat, or centralization of fat after controlling for potential confounding factors (Table 2). The odds ratio for stunting at 2 years as a predictor of overweight at 7 to 9 years was not significant at 1.09 (95% confidence limits: 0.30, 3.98).

Within the subgroup of 136 children who had DXA data at 9 years of age, 15% were stunted at 2 years of age. That status did not significantly predict their DXA values for percentage lean or fat masses, BMI, or CFR status at 9 years of age.

Table 3 illustrates the DXA values of the stunted and non-stunted children. The stunted children had significantly less total and lean tissue than the non-stunted children.

Table 3.  DXA body composition values comparing stunted with non-stunted children
 
DXA ValueMeanSDMeanSDMeanSD
  • *

    p < 0.001 between stunted and non-stunted total lean values.

  • p < 0.05 between stunted and non-stunted total tissue values.

Total fat (g)6701.73407.48008.04331.27825.54227.7
Total lean (g)18,213.0*2724.220,747.22688.020,393.22824.1
Total tissue (g)26,485.64391.729,772.55835.929,313.35757.1
Body fat (%)24.38.325.88.425.68.4
Body lean (%)69.710.870.88.470.68.7

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References

In contrast to previous cross-sectional studies that predict a future risk of overweight in previously stunted children, the current longitudinal analysis found no significant relationship between stunted status at the end of infancy and body composition in late childhood. Thus, we support the findings of Walker et al. (10) that there is no association between stunting at the end of infancy and body composition in late childhood, although the more robust methodology used in this study now provides more compelling evidence for this argument.

The body fat prediction equations of Slaughter et al. (11), used by Walker et al. (10) and designed to predict fat in prepubertal children, must be used with care (12). Differences between Slaughter's and Walker's uses of Tanner's prepubertal and pubertal designations according to either pubic hair (Slaughter) or breasts or genitalia (Walker) must be viewed in relation to the well-documented independence of pubertal maturity indicators (19, 20). Slaughter et al. (11) also differed from Tanner (19, 20) in the designation of prepubertal and pubertal categories and, consequently, included pubertal adolescents in analyses of prepubertal samples. Because of pubertal changes in the quantity and distribution of body fat, the prepubertal sample is likely to reflect higher fat levels with greater centralization than would actually be the case. Walker et al. (10) did not find an association between stunting and body fat but did find a significant association between stunting and centralization of fat that was partially explained by birth weight and that was also seen to develop between 7 and 11 years.

The significantly greater BMI at 2 years of age in stunted children from the current study requires explanation. We believe that this does not demonstrate a tendency toward overweight or obesity but is a reflection of the greater reduction in height rather than weight in stunted children. Because the BMI equation uses height squared as a denominator, a proportionally greater reduction in height will tend to be exaggerated, giving shorter children greater values for BMI. In the current study, for instance, mean weight at 2 years approximates the CDC 2000 10th centile, whereas mean height is less than the fifth centile. This is further emphasized by the correlation between height and BMI, which was not significant in non-stunted children (r = −0.107) but significantly negative for stunted children (r = −0.505, p < 0.01).

Of course, it could be argued that the truly stunted child demonstrates proportional losses in both height and weight and that an increased weight in stunted children really does represent a degree of overweight. However, the absence of any supporting evidence from subcutaneous fat and the lack of any significant tracking of body fat measures from the end of infancy to late childhood lend support to an absence of relationship of stunting in infancy as a predictor of overweight in late childhood.

The finding of a significant relationship between stunting and concurrent and/or future overweight in cross-sectional studies can, perhaps, be explained by these arguments. Higher BMI in stunted infants does not necessarily imply a tendency toward obesity in later childhood because such children, when followed through childhood, do not maintain significantly greater BMI, certainly not within the transitional societies studied here in South Africa and by Walker et al. in Jamaica, nor do they demonstrate increased subcutaneous fat and/or significant centralization of that fat. The problems of interpreting changes over time from cross-sectional data are that the differences between cross-sectionally assessed children and adolescents cannot be guaranteed to be a result of natural change. Both are being assessed at the same point in time, and it cannot be guaranteed that the child will respond to current influences to produce the characteristics of the current adolescent.

These results do not imply that stunted children will not demonstrate a tendency to overweight and obesity during adolescence. Indeed, given the South American evidence from Sawaya et al. (5) and the feasible physiological mechanisms (7, 8, 9), it may well be that stunted children and, in particular, those recovering from protein-energy malnutrition, are programmed to accumulate greater body fat at central sites during adolescence. We have been unable to show, however, that these changes are evident before the initiation of pubertal development.

Acknowledgment

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References

The BT20 birth cohort study receives financial and logistic support from the Medical Research Council of South Africa, Anglo-American Chairman's Fund, Child, Youth and Family Development of the Human Sciences Research Council of South Africa, and the University of the Witwatersrand. The Bone Health study is financially supported by the Wellcome Trust (United Kingdom).

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References
  • Cameron, N., De Wet, T., Ellison, GTH, Bogin, B. (1998) Growth in height and weight of South African urban infants from birth to five years: the Birth To Ten Study. Am J Hum Biol. 10: 495504.
  • Jinabhai, C. C., Taylor, M., Sullivan, K. R. (2003) Implications of the prevalence of stunting, overweight and obesity amongst South African primary school children: a possible nutritional transition? Eur J Clin Nutr. 57: 358365.
  • Popkin, B. M., Richards, M. K., Montiero, C. A. (1996) Stunting is associated with overweight in children of four nations that are undergoing the nutrition transition. J Nutr. 126: 30093016.
  • Benefice, E., Garnier, D., Simondon, K. B., Malina, R. M. (2001) Relationship between stunting in infancy and growth and fat distribution during adolescence in Senegalese girls. Eur J Clin Nutr. 55: 5058.
  • Sawaya, A. L., Dallal, G., Solymos, G. M., et al. (1995) Obesity and malnutrition in a shanty town population in the city of Sao Paulo, Brazil. Obes Res. 3(suppl): 107115.
  • Sawaya, A. L., Grillo, L. P., Verreschi, I., da Silva, A. C., Roberts, S. B. (1998) Mild stunting is associated with higher susceptibility to the effects of high fat diets: studies in a shantytown population in Sao Paulo, Brazil. J Nutr. 128(Suppl 2): 415S420S.
  • Ashworth, A. (1969) Growth rates in children recovering from protein-calorie malnutrition. Br J Nut. 23: 835845.
  • Fjeld, C. R., Schoeller, D. A., Brown, K. H. (1989) Body composition of children recovering from severe protein-energy malnutrition at two rates of catch-up growth. Am J Clin Nutr. 50: 12661275.
  • Sawaya, A. L., Roberts, S. (2003) Stunting and future risk of obesity: principal physiological mechanisms. Cad Saude Publica. 19(Suppl 1): S21S28.
  • Walker, S. P., Gaskin, P. S., Powell, C. A., Bennett, F. I. (2002) The effects of birth weight and postnatal linear growth retardation on body mass index, fatness and fat distribution in mid and late childhood. Pub Hlth Nutr. 5: 391396.
  • Slaughter, M. N., Lohman, T. G., Boileau, R. A., et al. (1988) Skinfold equations for the estimation of body fatness in children and youth. Hum Biol. 60: 709723.
  • Cameron, N., Griffiths, P. L., Wright, M. M., et al. (2004) Regression equations to estimate percentage body fat in African prepubertal children aged 9 y. Am J Clin Nutr. 80: 7075.
  • Yach, D., Cameron, N., Padayachee, N., et al. (1991) Birth To Ten: child health in South Africa in the 1990's, rationale and methods of a birth cohort study. Paed Peri Epi. 5: 211233.
  • Richter, L. M., Yach, D., Cameron, N., Griesel, R. D., de Wet, T. (1995) Enrollment into Birth To Ten (BT20): population and sample characteristics. Paed Peri Epi. 9: 109120.
  • Cameron, N. (1984) The Measurement of Human Growth. Croom-Helm: London, United Kingdom.
  • Alexander, G. R., Himes, J. H., Kaufman, R. B., Mor, J., Kogan, M. A. (1996) United States national reference for fetal growth. Obstet Gynecol. 87: 163168.
  • Mamelle, N., Cochet, V., Claris, O. (2001) Definition of fetal growth restriction according to constitutional growth potential. Biol Neonate. 80: 277285.
  • Ogundipe, E. M., Wolff, C. D., Seed, P., Gamsu, H. R. (2000) Does the antenatal detection of small-for-gestational-age babies influence their two-year outcomes? Am J Perinatol. 17: 7381.
  • Marshall, W. A., Tanner, J. (1969) Variations in the pattern of pubertal changes in girls. Arch Dis Child. 44: 291303.
  • Marshall, W. A., Tanner, J. M. (1970) Variations in the pattern of pubertal changes in boys. Arch Dis Child. 45: 1323.