Contents - Previous - Next

10. Links between childhood and adult obesity

There is now a mass of evidence to support the concept that a heavy child is likely to grow into an overweight adult. It has been known for many years that obese children tend to become obese adults. For example, HAASE and HASENFELD (1958) found that 80% of overweight children remained markedly overweight when reexamined 20-30 years later; LLOYD, WOLF and WHELEN (1961) also found a majority of obese children entering adolescence aged 9-11 years remaining obese over the subsequent eight years. A larger study by ABRAHAM and NORDSEICK (1960) showed that 80% of obese 10- to 13-year olds were still overweight 26-35 years later. In this study, the normal-weight adolescents had a different outcome, with smaller fluctuations in weight and fewer becoming overweight in adult life.

Even longer-term follow-up studies over a period of 40 years have recently been reported from Sweden by MOSSBERG (1989). Childhood obesity persisted with 47% of the obese children remaining obese in later life. Most (85%) of these obese adults had been markedly overweight in childhood, i.e., weighing more than two standard deviations above the average for the age and sex of Swedish children. Parental and pubertal obesity seemed to amplify the poor health prognosis as did early excessive obesity. The last point was already evident from the substantial insurance data which emphasize the deleterious effects of obesity of early onset (BLAIR and HAINES, 1966).

These studies therefore strongly favour the concept that excessive weight, once established in childhood, tends to persist. MELBIN and VUILLE (1976) also found that rapid weight gain in infancy was significantly associated with overweight in adolescent boys, the same trend being observed in girls. On this basis the studies on infant growth rates and adult obesity may be linked to imply that infant growth patterns could identify those individuals with a greater propensity to adult obesity.

CHARNEY et al. (1976) showed just this phenomenon: 36% of infants below 6 months of age who exceeded the 90th centile by weight became overweight adults when studied 20-30 years later, and only 14% of average-weight or light-weight infants became overweight adults. The link between infant and adult weights were evident once the recognized influences of social class, educational level and parental weight on adult adiposity had been taken into account.

These findings could lead to the conclusion that excessive overfeeding in childhood leads to rapid weight gain and this then entrains a very high risk of adult obesity. An alternative explanation, however, is that the prevailing conditions conducive to weight gain operate in both childhood and adult life so that those susceptible members of society manifest their obesity both in infancy when they are overfed by their mothers and later in adult life when they fail to maintain their usual activity patterns or overeat for other social reasons.

If the second hypothesis is correct, then social conditions and personal approaches to weight stability may be more successful than a metabolic entraining of metabolism implied by the first explanation. KID (1970) and POSKITT and COLE (1977) showed that persisting infantile adiposity was not inevitable and that only a minority of heavy infants remained obese either at 6-8 years of age or even earlier at 4-5 years. The real question remains therefore one of assessing whether infant feeding can modify the degree of adiposity in adult life and whether by this or other mechanisms weight gain can be considered as a predictor of long-term health.

11. Experimental findings

It seems reasonable to consider some related studies on the long-term control of adiposity in primates. LEWIS et al. (1986) reported effects of feeding baboons three different formulations of milk for the first 4 months of life, i.e., until they were weaned (Table 1). Standardized volumes were fed, and the energy and other nutrient intakes were either increased or decreased by 40%. After normal feeding, underfeeding or overfeeding, all the infant baboons were weaned on to a freely available high-fat diet. Table 1 presents data recalculated to highlight the conclusion that adult adipocyte mass in all regions was affected by early dietary intake, but only in the external abdominal region was there a clear effect on adipocyte number. It would appear that hypercellularity did occur in infancy if the specific depots were still undeveloped at birth.

Table 1. Early feeding effects on adult baboon adipose tissue. Ratios of depot masses and adipocyte numbers in overfed and underfed animals

Fat depot



Total mass

Adipocyte numbers

Total mass

Adipocyte numbers


2.0 *


5.7 *



5.9 *


11.4 *





5.9 *


Posterior flank

5.7 *

2.5 *

24.3 *

3.5 *




11.8 *





3.3 *


Recalculated from data by LEWIS et al. (1986) where baboons are fed milk formula for 16 weeks providing 94.5 kcal/100 g or 40.5 kcal/100 g. Total adipose triglyceride mass and cellularity in each depot were assessed at 5 years after ad libitum feeding on the same high-fat diet.

* p < 0.05

In baboons, most adipocyte regions had developed, as in man, in late foetal life, so only the groin, flank and mesenteric depots were poorly developed at birth. Female animals also seemed much more susceptible to the influence of early feeding, but whether this reflected sex-specific changes after weaning in intake or activity was unknown. By 1 year of age, animals fed differently in infancy were of similar weight, and this similarity was maintained in males through to adulthood. Marked differences in the weights of females only reappeared at almost 3 years of age implying a role for pubertal events in allowing obesity to be revealed once more.

These findings lend support to the proposition that different levels of food intake in early life can indeed have a marked influence on adult adiposity but this is unlikely to be mediated by the induction of new adipocytes. How the effect is achieved remains unclear. Since the experimental findings manipulated total nutrient intake and not just energy intake, it is by no means certain that differences in energy intake were responsible.

12. The effects of early feeding practices on the programming of metabolism

The impact of specific nutrients, rather than of energy on long-term health, needs to be recognized because otherwise an intake of energy may be incorrectly identified as undesirable; it is difficult at present to exclude the possibility that the inadequate or excessive intake of some nutrients may be far more important than energy per se in programming facets of metabolism which are of key importance to long-term health. One very interesting example is the recent evidence that an adequate supply of absorbable dietary iron is crucial to tissue-dependent changes in brain development which determine the long-term ability to learn (HAAS and FAIRCHILD, 1989).

Crucial phases of brain development may depend on the adequate provision of specific nutrients at that time; when these nutrients are supplied later, functional recovery may not occur. One must therefore consider the concept of critical periods for specific nutrient supply as not just hypothetical but plausible. Crawford has recently emphasized this point in relation to the need for an appropriate supply and balance of v-3 and v-6 essential fatty acids in the processes of brain growth and myelination (CRAWFORD et al., 1989), but the quantitative aspects of this hypothesis have not been set out.

The ideal intake of a nutrient may also vary at different times of life. Thus, for example, a plentiful intake of absorbable iron is regarded universally as essential to health, whether one is considering children or adults (FAO, 1988). Yet, recently we have suggested that a crucial component of the atherosclerotic process is free radical damage to lipids promoted by intracellular iron and that intracellular iron stores, whilst useful as reserves in the bone marrow to cope with increased demands, may, when present in other tissues such as the endothelium, serve as activators of the atherosclerotic process, unless there is a plentiful supply of free radical scavengers such as vitamins C and E (JAMES, DUTHIE and WAHLE, 1989; DUTHIE, WAHLE and JAMES, 1989). Age-specific needs for nutrients could thus emerge.

Evidence for long-term programming of cholesterol metabolism comes from both human and primate research. Thus, MARMOT et al. (1980) have analysed the relationship between breast-feeding and subsequent serum cholesterol levels in the two longitudinal British studies, the 1946 Birth Cohort Study and the Whitehall Study of British Civil Servants. At the age of 32, men showed little difference in their serum cholesterol when those who had been breast-fed or bottle-fed from birth were distinguished, but in women who had been breast-fed, serum cholesterol was almost 0.5 mm lower than in those who had been bottle-fed despite consuming the same type of diet and amounts of fat. Adjusting for possible social class or body weight differences failed to eliminate the apparent effect of breast-feeding.

Cogent physiological evidence on the potential for the long-term programming of cholesterol metabolism again comes from primate studies by MOTT (1986). Infant baboons were breast-fed or provided with formulae containing specified and graded concentrations of cholesterol. After 14 weeks the baboons were weaned on to specified diets containing 40% energy as fat and either enriched with saturated fatty acids (P/S ratio 0.37) or unsaturated fatty acids (P/S ratio 2.1). The cholesterol content was also either 0.01 mg or 1.0 mg cholesterol/kcal so that four types of weaning diet could be tested with the effects of saturated fat and dietary cholesterol being distinguished. Three to six years later, cholesterol metabolism was assessed in detail.

The breast-fed baboons proved to have a higher concentration of total, VLDL and LDL cholesterol than those originally formula-fed, i.e., the effect was the opposite of that found by MARMOT et al. (1980) in adult women. Breast-feeding followed by a high saturated fatty acid intake also led to a lower HDL cholesterol concentration than in formula-fed animals, but the impact of polyunsaturated fatty acids on HDL cholesterol concentrations was very different in the breast-fed and bottle-fed baboons. Whereas a prolonged diet rich in unsaturated fatty acids increased HDL cholesterol levels, the same diet in animals initially formula-fed decreased cholesterol concentrations. Thus, the delayed effects of breast-feeding involve profound changes in the regulatory responses in cholesterol and/or lipoprotein metabolism to adult diets.

Table 2 shows data on cholesterol turnover from MOTT's (1986) review. Formerly breast-fed animals had a lower synthesis rate of cholesterol in association with the higher VLDL and LDL cholesterol concentrations. Cholesterol turnover was lower and the size of the cholesterol pool (presumably in tissues) was substantially lower in the breast-fed baboons. Despite these changes the fractional absorption of dietary cholesterol was increased by 15%. Again sex differences were observed with female baboons having a higher production rate of cholesterol. All the deferred effects of breast-feeding may well have reflected primary changes in lipoprotein synthesis, exchange or receptor uptake, but these facets of lipoprotein turnover were not studied.

Table 2. Deferred effects of breast-feeding in baboons: 4-14C cholesterol feeding at 3.5 years of age



Cholesterol absorption %



Transport in rapid exchange pool mg/kg/d



Cholesterol production rate mg/kg/d



Flux to slower pool mg/kg/d



Mass of slower pool mg/kg



All differences significant p < 0.05.
MOTT (1986).

It seems reasonable to conclude that the early overfeeding of babies could be harmful in that it increases the probability of a high rate of weight gain, of early obesity persisting into adult life with all its attendant sequelae of increased morbidity and premature mortality. On this basis, it might seem self-evident that the return to breast-feeding alone for the first 4-6 months of life with the increased likelihood of reasonable and not excessive growth rates should be welcome. Current data suggest that modern breast-fed children in affluent societies grow more slowly in length and weight, but that the divergence from the NCHS growth pattern is particularly evident in the second half of infancy, i.e., when they are being substantially supplemented with ordinary foods.

Whatever the link between modern breast-feeding and supplementary feeding practices, the net effect is a lower nutrient intake than that needed to achieve the NCHS standards. We thus might specify that the upper desirable intakes of energy are either above those usually consumed by the breast-fed child or that we need to revise the NCHS charts if they are to continue to be a standard rather than a reference set of measurements. Before doing so, however, one needs to examine evidence linking growth rates in infancy to adult morbidity and mortality since at present we have only been concerned with links inferred from studies on adiposity and those adult insurance data presenting evidence that early obesity is particularly hazardous.

13. Infant growth rates and long-term survival

Barker and his colleagues have rekindled interest in the long-term effects of foetal and postnatal growth rates. Barker originally noted a strong geographical relationship between mortality rates from ischaemic heart disease over a 10-year period around 1970 and infant mortality rates 50 years earlier. On this basis, BARKER and OSMOND (1986) postulated that poor nutrition in early life increases the child's susceptibility to an affluent diet. These findings echoed those from Scandinavia where a similar relationship between infant mortality rates and adult disease was found in Norway (FORSDAHL, 1973) and in Finland (NOTKOLA, 1985).

ROSE (1964) had earlier shown that twice as many siblings of patients with ischaemic heart disease had died in infancy or had been stillbirths. Prenatal influences may be important because BARKER et al. (1989a) have recently observed that systolic blood pressure at both 10 years and in adult life was inversely related to birth weight. Those children living in areas of England and Wales with high mortality rates from cardiovascular disease were shorter and had higher resting pulse rates than those living in other areas.

This may suggest that the hypothalamic 'setting' of both growth patterns and the autonomic system may be influenced by in utero feeding, a concept favoured by McCANCE, and Widdowson to explain the astonishing re-programming of final body size by early manipulation of food intake in pigs (WIDDOWSON, 1974).

Further, recent analyses by BARKER et al. (1989b) suggest that men with the lowest birth weight had the highest death rate from coronary heart disease, but crucial to the thrust of this paper was the observation that there was an interaction between birth weight and body weight at 1 year with the mortality risk of CHD later in life. The risk was substantially reduced if boys with a lower birth weight showed fast growth rates in the first year of life.

Two points must be remembered when assessing the significance of these suggestions. Firstly, it was not possible to eliminate the potential co-correlation with social class, but there was little to support this explanation. Secondly, 94% of the sample had been breast-fed so the strength of the evidence relates only to growth rates and not to breast-feeding per se. Barker found a convincing inverse relationship between standardized mortality rates from CHD and the individuals' body weights at 1 year only in those who were breast-fed, but given the small group of formerly bottle-fed men under study, one cannot exclude a similar relationship.

14. Conclusions

The simplest conclusion we can come to is that it is not possible to define an upper limit to the desirable intake of energy in childhood. A more refined approach to future recommendations can, however, be discerned. There would appear to be clear evidence from primates that there is a sex-linked programming not only of body weight but of cholesterol metabolism depending on (a) whether breast-feeding or bottle-feeding is used, (b) the amount fed, and (c) the type of fatty acids ingested.

Human studies also display a link between high infant growth rates and adult adiposity, but the proposed association between early adiposity and premature mortality may well be overwhelmed by the observed but unexplained link between high infant growth rates and a reduced risk of coronary heart disease. Since obesity only has an indirect influence on the risks of CHD, the impact of infant feeding on blood pressure and cholesterol metabolism is likely to be far more important than its effects on the long-term programming of adipose cell volume. Adipocyte hypercellularity does not appear to be involved in these effects, perhaps because in man enough adipocytes are preformed at birth and therefore simply expand rather than recruit new calls to accommodate the storage of fat..

The intriguing physiological question is how the effects of prenatal and postnatal nutrition are mediated to effect changes in the long-term programming of growth, pulse rate, cholesterol metabolism, adult adiposity and the risk of premature death. The public health implications are, however, clear: it would be very foolish to assume that estimates of desirable energy intakes should be reduced in line with new calculations of the intake of breast-fed children when the application of these observations to bottle-feeding practice is likely to reduce iron, energy and other nutrient intakes with a recognized effect in reducing growth rates. Rather than presupposing that lower growth rates in the breast-fed child throughout the first year are optimal, we must now recognize that the preliminary epidemiological evidence suggests that faster growth rates are more beneficial.

If obesity is established in infancy or childhood, then there is now substantial evidence that this increases the probability of obesity in adolescence and in adult life. Experimental findings on baboons support a long-term programming of adiposity not mediated by an adipocyte hypercellular mechanism. This evidence supports a lowering of recommended intake in childhood. However, some evidence now suggests that rapid growth in infancy is advantageous on a long-term basis, thus casting doubt on the wisdom of accepting slower growth rates in breast-fed children. Early feeding practices, e.g., in relation to iron, fatty acid and cholesterol metabolism, also program metabolism in the long term.

It is concluded that, given the uncertainty about the impact of different infant growth rates on adult risk of morbidity and mortality, a reduction in the recommended intakes for infants should only be undertaken when more reassuring data become available.

Contents - Previous - Next