Introduction

Obesity is emerging as a serious public health issue in the Caribbean as in other developing nations (Forrester et al, 1996). Chronic cardiovascular disease has become a major determinant of mortality in the region and the prevalence of obesity has increased (Bjorntorp, 1998; Burrows, 2000). Mapping of the patterns of distribution and identification of individuals or groups at risk for obesity has clearly become a public health priority. Many countries undergoing the epidemiologic transition have scarce resources for health care and are now in the paradoxical position of dealing with the traditional maladies of early childhood undernutrition and emerging adult obesity simultaneously (Aguirre, 2000).

The limited success of adult obesity intervention programmes (McManus et al, 2001; Freedman et al, 2001) has led to increased interest in prevention programs aimed at children (Burrows, 2000). Limited availability of good convenient screening tools has been a problem in devising programmes for obesity prevention and management in children. Many children in the developing world have less than adequate nutrition in early childhood, although this may not be immediately obvious because of the phenomenon of catch-up growth (Walker et al, 1996). Obesity screening tools therefore need to be able to operate usefully in populations of mixed nutrition history. However most research in this area has been conducted in developed countries where childhood undernutrition is rare.

Body mass index (BMI) reflects adiposity well in adult groups, and has long been an affordable and useful method of assessment in adult studies of obesity (WHO, 1995). However because of differences in the rates of maturation and its effect on body composition, assessment of adiposity by BMI has been more challenging in children (Widhalm & Schonegger, 1999). Of particular interest is the association of low BMI with higher than expected body fat in several ethnic groups (Deurenberg-Yap et al, 2000; Yajnik, 2001) and evidence that episodes of early undernutrition may serve to potentiate adult obesity related disease (Eriksson et al, 1999; Ong et al, 2000). In addition recent work in American children suggests that visceral fat grows at a faster rate than fat from other depots (Huang et al, 2001) and in children from Spain a trend towards a central pattern of fat distribution was observed (Moreno et al, 2001). Added to this are mixed findings on the stability of BMI and fat distribution over time (Campbell et al, 2001; Guo et al, 1997). Clearly several factors may have an impact on the correct inter-pretation of BMI in risk assessment. The validity of adiposity assessment tools in different groups is therefore of importance.

The International Obesity Task Force (IOTF) recently published a very welcome set of overweight and obesity cut-off points for children, corresponding to the adult BMI cut-off points (Cole et al, 2000). Many studies in children have used measures such as skinfolds and body circumferences in assessing fatness. It is therefore important to compare these traditional methods of obesity screening to the IOTF BMI cut-off points and to examine their application in groups not included in the analyses from which they were derived. Reilly et al (2000) have shown that the IOTF obesity cut points had high specificity but relatively low sensitivity in obesity screening in childhood in the UK and that the sensitivity differed between boys and girls. This pattern is also seen in a variety of adolescent groups in the USA and Europe (Malina & Katzmarzyk, 1999).

We have followed a cohort of children some of whom were undernourished (as measured by linear growth retardation) in early childhood (Grantham-McGregor et al, 1991). Among the undernourished children there was substantial catch up growth and they had more central distribution of fat than the adequately nourished children (Walker et al, 2002). These children form an ideal group on which to test the robustness of BMI for both obesity tracking and as a screening tool in children with varying nutritional histories.

The purpose of our study was to examine the relationships of BMI to obesity indices derived from anthropometric measurements in a group of Jamaican children measured at age 7–8 y and again at 11–12 y and to determine tracking of obesity between late childhood and early adolescence. In addition we wished to compare identification of overweight children using the IOTF BMI cut-off points with identification by skinfold measurements.

Methods

The subjects in this study were participants of a prospective study conducted on the functional effects of linear growth retardation in inner city Jamaican children (Grantham-McGregor et al, 1991; Walker et al, 1991). In the original study all children aged 9–24 months with heights below -2 s.d. of the NCHS references (Hamill et al, 1977) were identified from a house-to-house survey. These stunted subjects (n=127) were compared to 32 non-stunted children (height for age >-1 s.d.) matched for age, sex and neighbourhood to every fourth stunted child. One hundred and twenty-two of the stunted children, the 32 non-stunted children, and an additional 175 subjects who had been identified at the time of the original survey but were not stunted in early childhood were examined during a follow up study at age 7–8 y (Gaskin et al, 2000). For the purposes of this study we re-examined 306 of the children at 11–12 y (116 stunted and 190 non-stunted, 91.6% of those originally identified). Most of the children who were stunted in early childhood were no longer stunted at age 7–8 y due to catch up growth (Walker et al, 1996). The stunted children remained smaller and thinner than the non-stunted group by every measure except the subscapular/triceps skinfold ratio (SSF/TSF) (Walker et al, 2002). The mean heights of the stunted children remained lower than that of the non-stunted children at age 11–12 y (stunted children—mean height=41.46.0 cm; mean height for age=-0.010.82: non-stunted children—mean height=153.46.0 cm, mean height for age=0.520.81, P<0.001). However, the children's heights were normally distributed when the groups were combined.

The Ethics Committee of the University of the West Indies gave approval for this study and the parents and guardians gave written consent for their children to participate. All measurements were conducted at our research unit.

Triceps (TSF) and subscapular (SSF) skinfolds were measured in triplicate and the average used. In addition, height, weight and mid-upper arm circumference (MUAC) were recorded. The skinfolds and MUAC were measured on the left side of each subject. All measurements were conducted by two trained observers, according to standard procedures (Lohman et al, 1988). Inter-observer reliability (intraclass correlation coefficients) was greater than 0.99 for all the measurements. The same observers measured the children at ages 7–8 y and 11–12 y. The sum of the skinfolds (sum SF) and BMI were calculated. Equations using the skinfold measurements were used to estimate percentage fat (%fat). Fat mass (FM) was determined from weight and %fat (Slaughter et al, 1988). These equations have been shown to produce good results in estimating body fat (Slaughter et al, 1988) and compare well to body fat estimated by near-infrared interactance (NIR) and foot-to-foot bioimpedance (BIA) in estimating body fat content (Sampei et al, 2001). At age 11–12 y the children were classified as pre-pubertal or pubertal as assessed by WHO recommended cut-off points, breast stage 2 for girls and genitalia stage 3 for boys (WHO, 1995).

Statistical analyses

Descriptive statistics by sex, and age are presented. To examine the extent of the tracking over the follow up period, Pearson product–moment correlation coefficients were calculated between each index of adiposity at 7–8 y and the same index at 11–12 y. Tracking of overweight status using IOTF BMI cut-off points was assessed. We also calculated the percentage of children who remained in the same BMI tertile at 7–8 and 11–12 y. Correlation analysis was conducted between BMI and the other indices of adiposity to discern the simple relationships of BMI to the other indices. Multivariate linear regression analyses were conducted using each index of obesity in turn as the dependent variable to further compare the relationships of BMI to the other indices of adiposity at 7–8 y and in early adolescence. Analyses were conducted separately for boys and girls and the children's age and height were entered in the model. At age 11–12 y pubertal status was also entered.

Children were categorized as overweight or not, based on the IOTF cut-off points for BMI equivalent to adult BMI of 25 (Cole et al, 2000) and as having high body fat by the 85th percentile of the 'National Health and Nutrition Examination Surveys I and II' references (NHANES I and II) for black children (Frisancho, 1990) for TSF, SSF and sum SF. Each measure was used in turn to calculate sensitivity and specificity of the BMI cut-off points to detect body fatness compared with skinfolds at 7–8 y and 11–12 y.

Results

The sample comprised 169 boys and 137 girls with mean age in late childhood of 7.89 (0.40) y and 7.92 (0.41) y and in early adolescence 11.81 (0.35) and 11.82 (0.34) y respectively. Table 1 shows descriptive statistics for the obesity indices and indicates the significant differences between the boys and girls. Two subjects were not measured at age 7–8 y and one subject did not have skinfolds measured at age 11–12 y. At age 11–12 y pubertal stage was not assessed in five children and therefore %fat was not calculated for those subjects, as there are separate prediction equations for prepubertal and pubertal children. Twenty-three percent of the boys and 88% of girls had reached puberty.

Table 1 - Obesity indices in girls and boys at age 7–8 y and 11–12 y (mean and s.d.).

Full table (24K)

Differences between boys and girls

At age 7–8 y mean BMI among girls was 14.931.72 kg/m² and among boys 15.111.37 kg/m² and there was no significant difference in mean BMI between the sexes. At age 11–12 y mean BMI among girls was 18.013.34 kg/m² and was significantly higher than that among boys, which was 17.15 (2.55) kg/m² (P<0.01). All measures of adiposity were higher in girls than boys at both ages (P<0.01).

All anthropometric measurements increased between 7 and 11 y with girls having larger increases as would be expected given the larger number of girls who had reached puberty. For both boys and girls the means of the skinfold measurements moved to a higher percentile of the 'NHANES Anthopometric Tables for Blacks' (Frisancho 1990) as the children aged. For example, the mean TSF for girls was below the 50th percentile at age 7–8 y and greater than the 50th percentile at 11–12 y. The group means for SSF were at higher NHANES percentiles than those for TSF or the sum of SF, eg the mean TSF among boys 11–12 y was just above the 50th percentile whereas the mean SSF was above the 75th percentile.

Tracking

Using the IOTF BMI cut points as the index of overweight, six girls were overweight and five boys at age 7–8 y. At 11–12 y 15 girls were overweight or obese compared with 14 boys. The BMI overweight cut-off point for girls was 18.03 kg/m² and for boys was 18.16 kg/m² at 7–8 y. At 11–12 y the cut-off points for overweight were 21.20 kg/m² and 20.89 kg/m² for girls and boys, respectively (Cole et al, 2000). All of the children classified as overweight at 7–8 y were still overweight at 11–12 y and the BMI of four of the five boys, and five of the six girls who were overweight at 7–8 y remained in the top five and six respectively of BMI ranking at age 11–12.

The cross-classification of the children by BMI tertile at age 7–8 y and 11–12 y is shown in Table 2. Overall 66% of the children remained in the same tertile. Very few (2.2 and 5.4% boys and girls, respectively) of the children who were in the lowest tertile at 7–8 y moved to the top tertile, and none of these were overweight as assessed by the IOTF cut-off points. Most (71.4 and 77.3% boys and girls, respectively) of the children in the top tertile of BMI at 7–8 y were also in the top tertile at age 11–12 y. We further examined BMI tracking by original stunting status most of the children in the highest tertile of BMI were non-stunted. However children in the highest BMI tertile tended to remain there over follow up whether stunted of not. The Pearson product–moment correlation coefficients of all the indices with themselves over time were high and generally tended to be slightly higher in boys (Table 3). However for fat distribution (SSF/TSF) the size of the coefficient was smaller than for the other measures.

Table 2 - Percentage of children who moved to a higher or lower tertile of BMI between 7–8 y and 11–12 y by sex.

Full table (13K)

Table 3 - Pearson product–moment correlation coefficients of the obesity indices at 7–8 y with corresponding indices at 11–12 y.

Full table (11K)

Relationship of BMI to the adiposity indices

The correlation coefficients between BMI and the other indices of adiposity were also high and significant. All tended to be larger in girls than boys at both ages and the relationships all tended to be stronger as the children got older. Within each age and sex group there was little variation in the magnitude of the correlations between BMI and the other indices.

Height was normally distributed (151.38.45 and 146.87.98 cm in girls and boys, respectively, at age 11–12 y); however, because the children had been selected according to their height at entry to the study, we always included height in the multivariate analyses. In addition, because the simple correlation coefficients between BMI and age were significant except among boys at 11–12 y, we included this variable along with height in the further analyses. Pubertal status was also included in the equations at age 11–12 y. Regression models were calculated predicting each of the following obesity indices, SSF, TSF, Sum SF, %fat and FM, from age, height and BMI at 7–8 y. Separate analyses were conducted for boys and girls because of the well documented differences in body composition.

BMI, was strongly related to all of the indices of obesity. The relationship with FM was strongest, BMI explaining 52 and 61% of the variance in FM among boys and girls, respectively, at age 7–8 y. BMI explained 42–47% among boys and 55–60% among girls, of the variance in the other indices of adiposity at 7–8 y. The trends were similar at age 11–12 y, with the strongest relationship again between FM and BMI; BMI explained 69% of the variance in FM for both sexes. The strength of the relationship of all of the indices with BMI increased at 11–12 y, with the increases being more marked among boys (Table 4). BMI was more strongly associated with the indices in girls at age 7–8 y, but by age 11–12 y the ability of BMI to predict the indices was similar in both sexes except for TSF among boys.

Table 4 - Regression coefficients for separate multiple regressions of the obesity indicies on BMI at ages 7 y and 12 y in both sexes controlling for age, height and puberty.

Full table (24K)

Sensitivity and specificity of IOTF cut-offs for BMI for identification of fat subjects

Table 5 shows the calculations of sensitivity and specificity for the international BMI cut-off points equivalent to an adult BMI of 25 kg/m² compared with the 85th percentile of the NHANES cut-off points for black children at 7.0–7.9 and 11.0–11.9 y for TSF, SSF and sum SF.

Table 5 - Sensitivity and specificity of BMI^a to identify children classified as fat by skinfolds^b at ages 7–8 y and 11–12 y.

Full table (15K)

In boys the BMI had very low sensitivity at 7–8 y to identify children with skinfolds above the 85th percentile, particularly for the SSF. The sensitivity of BMI to identify children classified by TSF and sum of skin folds was higher among 7–8-y-old girls but about 40% of the children were still misclassified. Sensitivity for SSF was also low among girls. The sensitivity of BMI for children classified by SSF, TSF and sum SF was higher at age 11–12 but nevertheless between 14% and 33% of children were misclassified dependent on the index and sex. The specificity of BMI was high at both ages for boys and girls.

Discussion

The children in this study were generally lean with only 11 classified as overweight at age 7–8 y increasing to 29 at age 11–12 y. All children came from low-income homes and our sample contained a number of children known to be undernourished in early childhood, as assessed by linear growth retardation. Substantial catch up growth had occurred and few children remained stunted (height for age less than -2 s.d. NCHS references (Hamill et al, 1977) at the time of this study. However it is possible that the children's households remained food insecure. Childhood undernutrition may increase susceptibility to obesity (Yajnik, 2001); however, although the cohort got fatter over the follow-up period, the children remained relatively thin compared with the reference populations. The increase in adiposity between mid-childhood and early adolescence follows the trends seen in a wide range of populations (Parizkova & Hills, 2001).

Girls were fatter than boys, by all indices, except for mean BMI at age 7–8 y. This was probably largely related to earlier sexual development among girls, the majority of whom were pubertal at age 11–12 y compared with only 23% of boys. Nonetheless, the trend in girls towards greater increases in adiposity has serious implications, as changes in childhood adiposity have been shown to be more closely related to adult adiposity in females (Guo et al, 2000; Parizkova & Hills, 2001). It is in keeping with the higher prevalence of obesity among adult Jamaican black women (Forrester et al, 1996). This finding points to childhood as the period where the gender differences in the rates of obesity may begin.

BMI in our study was highly correlated in childhood and early adolescence, which was different from a Guatemalan study in which correlations were moderate between age 5 and 14–17 y (Schroeder & Martorell, 1999). BMI tracking was also moderate between childhood and adolescence among Chinese children (Wang et al, 2000).

We examined stability of BMI rank as a means of estimating whether those children in the top of the fat distribution were likely to remain there over time. We found that this was indeed the case and it suggests that children identified as of higher adiposity rank are more likely to be at increased risk of adult obesity. We used BMI tertiles because the children in our study were generally lean, and evaluation of overweight tracking using BMI cut-off points was unlikely to fully describe the potential associations of childhood and adolescent fat status. Our findings suggest that adiposity ranking in Jamaican children is fairly stable. Tracking of adiposity as measured by skinfolds was also high. We found that a tendency for a central fat distribution tracked between 7 and 11 y in both sexes although the association was less strong than that for the other measures. This indicates that childhood screening may be useful, given the expected increases in the prevalence of obesity in populations in transition, such as ours.

Unlike the findings in other populations, in which there were appreciable differences in the magnitude of association of BMI with a variety of indices of obesity (Malina & Katzmarzyh, 1999; Schroeder & Martorell, 1999), in the present study the association between BMI and the other adiposity measures was very similar. In our study the strength of the relationships between BMI and the adiposity measures increased from 7–8 y to 11–12 y. This was similar to the findings in the Guatemalan study, however the associations were consistently higher among the Jamaican children. The close relationship between BMI and the body composition measures suggests that BMI reflects adiposity in these age groups.

The much higher adiposity percentile ranking resulting from the use of SSF as the index of obesity compared with the TSF suggests that this parameter may have a different relationship to fatness in this group compared to the reference group, which is surprising since we used references for black children. This finding may partially be explained by higher trunkal fat in the children who were previously undernourished (Schroeder & Mastorell, 1999; Walker et al, 2002), but nevertheless highlights the danger of use of indices derived in different populations.

This study confirms the high specificity of BMI in identifying children with high body fat reported elsewhere (Malina & Matzmarzyk, 1999; Reilly et al, 2000). This is important because it suggests that the trend towards wide application of BMI as a measure of adiposity among children is acceptable, since it is operating well in widely different groups and few children will be inappropriately identified for more in-depth follow-up.

At 11–12 y BMI was a relatively good measure of fatness for this group, with sensitivity ranging from 67 to 86%. However, several children who would have been classified as fat using the 85th percentile cut-offs for skinfolds were not identified. In contrast to the pattern at age 11–12 y, sensitivity was extremely low at age 7–8 y, particularly among boys. This suggests that the IOTF cut-offs for BMI may not be useful at this age in our population.

Our findings of low sensitivity and high specificity of the IOTF BMI cut-off points for overweight in our children at 7–8 y reflect a similar pattern for obesity in a British cohort of children at 7 y (Reilly et al, 2000). This is interesting since the children come from very different socio-cultural environments and the measures of fatness with which the BMI cut-offs were compared were different. This similarity coupled with the improvement in BMI for identification of overweight at 11–12 y emphasizes the point that BMI cut-offs have to be interpreted with caution in childhood.

In summary there was substantial tracking of BMI and the other indices of adiposity. However, although we can be fairly confident in this population that children identified as overweight by IOTF BMI cut-off points will in fact have relatively high body fatness, the BMI cut-off points will not identify many children with high body fatness. This is important since some studies among transitional populations have shown obesity associated with relatively low BMI (Deurenberg-Yap et al, 2000; Yajnik, 2000). Given the association between adiposity at these ages and adult adiposity (Williams et al, 1999), it would clearly be desirable to identify children with relatively high adiposity early to prevent further substantial increases.

References

1. Aguirre, P (2000). Socioanthropological aspects of obesity in poverty. In:Obesity and Poverty, a New Public Health Challenge, II, Pan American Health Organization
2. Bjorntorp, P (1998). Obesity: a chronic disease with alarming prevalence and consequences. J. Intern. Med., 244, 267–925. | PubMed |
3. Burrows, R (2000). [Prevention and treatment of obesity since childhood: strategy to decrease the non transmissible chronic diseases in adult.]. Rev Med Chil., 128, 105–110.
4. Campbell, PT, Katzmarzyk, PT, Malina, RM, Rao, DC, Perusse, L & Bouchard, C (2001). Stability of adiposity phenotypes from childhood and adolescence into young adulthood with contribution of parental measures. Obes. Res., 9, 394–400.
5. Cole, TJ, Bellizzi, MC, Flegal, KM & Dietz, WH (2000). Establishing a standard definition for child overweight and obesity worldwide: international survey. Br. Med. J., 320, 1240–1243. | Article | ISI | ChemPort |
6. Deurenberg-Yap, M, Schmidt, G, van Staveren, WA & Deurenberg, P (2000). The paradox of low body mass index and high body fat percentage among Chinese, Malays and Indians in Singapore. Int. J. Obes. Relat. Metab. Disord., 24, 1011–1017. | Article | PubMed | ChemPort |
7. Eriksson, JG, Forsen, T, Tuomilehto, J, Winter, PD, Osmond, C & Barker, DJP (1999). Catch-up growth in childhood and death from coronary heart disease: longitudinal study. Br. Med. J., 318, 427–431. | ISI | ChemPort |
8. Forrester, T, Wilks, R & Bennett, Fet al (1996). Obesity in the Caribbean. Ciba Found. Symp., 201, (17–26; discussion 31)2–6.
9. Freedman, MR, King, J & Kennedy, E (2001). Executive summary. Obes. Res., 9, 1S–5S. | PubMed | ISI | ChemPort |
10. Frisancho, AR (1990). Appendix A: Anthropometric tables for blacks. In:Anthropometric Standards for the Assessment of Growth and Nutritional Status, University of Michican Press
11. Gaskin, PS, Walker, SP, Forrester, TE & Grantham-McGregor, SM (2000). Early linear growth retardation and later blood pressure. Eur. J. Clin. Nutr., 54, 563–567. | Article | PubMed | ISI | ChemPort |
12. Grantham-McGregor, S, Powell, C, Walker, SP & Himes, JH (1991). Nutritional supplementation, psychosocial stimulation, and mental development of stunted children: the Jamaican Study. Lancet, 338, 1–5. | Article | PubMed | ChemPort |
13. Guo, SS, Chumlea, WC, Roche, AF & Siervogel, RM (1997). Age- and maturity-related changes in body composition during adolescence into adulthood: the Fels Longitudinal Study. Int. J. Obes. Relat. Metab. Disord., 21, 1167–1175. | Article | PubMed | ChemPort |
14. Guo, SS, Huang, C & Maynard, LMet al (2000). Body mass index during childhood, adolescence and young adulthood in relation to adult overweight and adiposity: the Fels Longitudinal Study. Int. J. Obes. Relat. Metab. Disord., 24, 1628–1635. | Article | PubMed | ChemPort |
15. Hamill, PVV, Driz, TA, Johnson, CL, Reed, RB & Roche, AF (1997). NCHS Growth Curves for Children, Birth–18 Years. DHEW publication no. (PHS) 78-1650 (Vital and Health Statistics Series 11, no. 165).Hyatsville MD: Department of Health and Human Services
16. Huang, TT, Johnson, MS, Figueroa-Colon, R, Dwyer, JH & Goran, MI (2000). Growth of visceral fat, subcutaneous abdominal fat, and total body fat in children. Obes. Res., 9, 283–289.
17. Lohman, RG, Roche, AF & Martorell, R (1998). Anthropometric Standardization Reference Manual. Champaign, IL: (Human Kinetics)
18. Malina, RM & Katzmarzyk, PT (1999). Validity of the body mass index as an indicator of the risk and presence of overweight in adolescents. Am. J. Clin. Nutr., 70, 131S–136S. | ChemPort |
19. McManus, K, Antinoro, L & Sacks, F (2000). A randomized controlled trial of a moderate-fat, low-energy diet compared with a low fat, low-energy diet for weight loss in overweight adults. Int. J. Obes. Relat. Metab. Disord., 25, 1503–1511.
20. Moreno, LA, Fleta, J, Sarria, A, Rodriguez, G, Gil, C & Bueno, M (2001). Secular changes in body fat patterning in children and adolescents of Zaragoza (Spain), 1980–1995. Int. J. Obes. Relat. Metab. Disord., 25, 1656–1660. | Article | PubMed | ChemPort |
21. Mueller, WH, Dai, S & Labarthe, DR (2001). Tracking body fat distribution during growth: using measurements at two occasions vs one. Int. J. Obes. Relat. Metab. Disord., 25, 1850–1855. | Article |
22. Ong, KKL, Ahmed, ML, Emmett, PM, Preece, MA & Dunger, DB (2000). Association between postnatal catch-up growth and obesity in childhood: prospective cohort study. Br. Med. J., 320, 967–971.
23. Parizkova, J & Hills, A (2001). Childhood Obesity, ed. IHJ Wolinsky.Boca Raton, FL: CRC press
24. Reilly, JJ, Dorosty, AR & Emmett, PM (2000). Identification of the obese child: adequacy of the body mass index for clinical practice and epidemiology. ALSPAC Study Team. Avon Longitudinal Study of Pregnancy and Childhood. Int. J. Obes. Relat. Metab. Disord., 24, 1623–1627. | Article | PubMed | ChemPort |
25. Sampei, MA, Novo, NF, Juliano, Y & Sigulem, DM (2001). Comparison of the body mass index to other methods of body fat evaluation in ethnic Japanese and Caucasian adolescent girls. Int. J. Obes. Relat. Metab. Disord., 25, 400–408. | Article | PubMed | ChemPort |
26. Schroeder, DG & Martorell, R (1999). Fatness and body mass index from birth to young adulthood in a rural Guatemalan population. Am. J. Clin. Nutr., 70, 137S–144S.
27. Slaughter, MH, Lohman, TG & Boileau, RAet al (1988). Skinfold equations for estimation of body fatness in children and youth. Hum. Biol., 60, 709–723. | PubMed | ISI | ChemPort |
28. Walker, SP, Powell, CA, Grantham-McGregor, SM, Himes, JH & Chang, SM (1991). Nutritional supplementation, psychosocial stimulation, and growth of stunted children: the Jamaican study. Am. J. Clin. Nutr., 54, 642–648. | PubMed | ChemPort |
29. Walker, SP, Grantham-McGregor, SM, Himes, JH, Powell, CA & Chang, SM (1996). Early childhood supplementation does not benefit the long-term growth of stunted children in Jamaica. J. Nutr., 126, 3017–3024. | PubMed | ChemPort |
30. Walker, SP, Gaskin, PS, Powell, CA & Bennett, FI (2002). The effects of birthweight and postnatal linear growth retardation on body mass index, fatness and fat distribution in mid and late childhood. Publ. Hlth Nutr., (in press)
31. Wang, Y, Ge, K & Popkin, BM (2000). Tracking of body mass index from childhood to adolescence: a 6-y follow-up study in China. Am. J. Clin. Nutr., 72, 1018–1024. | PubMed | ISI | ChemPort |
32. WHO (1995). Physical status: the use and interpretation of anthropometry, Report of a WHO Expert Committee. Technical Report Series no.Geneva: WHO
33. Widhalm, K & Schonegger, K (1999). BMI: does it really reflect body fat mass?. J. Pediatr., 134, 522–523.
34. Williams, S, Davie, G & Lam, F (1999). Predicting BMI in young adults from childhood data using two approaches to modelling adiposity rebound. Int. J. Obes. Relat. Metab. Disord., 23, 348–354. | Article | PubMed | ChemPort |
35. Yajnik, C (2000). Interactions of perturbations in intrauterine growth and growth during childhood on the risk of adult-onset disease. Proc. Nutr. Soc., 59, 257–265. | PubMed | ISI | ChemPort |
36. Yajnik, CS (2001). The insulin resistance epidemic in India: fetal origins, later lifestyle, or both?. Nutr. Rev., 59, 1–9. | PubMed | ISI | ChemPort |