Hostname: page-component-7c8c6479df-995ml Total loading time: 0 Render date: 2024-03-28T08:18:25.451Z Has data issue: false hasContentIssue false

Historical cohort studies and the early origins of disease hypothesis: making sense of the evidence

Workshop on ‘Nutritional models of the developmental origins of adult health and disease’

Published online by Cambridge University Press:  27 February 2009

Jonathan C. K. Wells*
Affiliation:
Childhood Nutrition Research Centre, UCL Institute of Child Health, 30 Guilford St, London WC1N 1EH, UK
*
Corresponding author: Dr Jonathan C. K. Wells, fax +44 207 831 9903, email J.Wells@ich.ucl.ac.uk
Rights & Permissions [Opens in a new window]

Abstract

The hypothesis that early-life growth patterns contribute to non-communicable diseases initially emerged from historical cohort studies, consistently associating low birth weight and infant weight gain with later disease risk. Cohort studies offer crucial life-course data on disease aetiology, but also suffer from important limitations, including the difficulty of adjusting for confounding factors and the challenge of interpreting data on early growth. Prospective randomised trials of infant diet appear to provide evidence in direct contradiction to cohort studies, associating faster early growth with disease risk. The present article attempts to resolve this contradiction on two grounds. First, insufficient attention has been directed to inconsistency of outcomes between cohort studies and prospective trials. Cohort studies can assess actual mortality, whereas prospective trials investigate proxies for disease risk. These proxies are often aspects of phenotype that reflect the ‘normalisation’ of metabolism in response to growth, and not all those displaying normalisation in adolescence and early adulthood may go on to develop disease. Second, a distinction is made between ‘metabolic capacity’, defined as organ development that occurs in early life, and ‘metabolic load’, which is imposed by subsequent growth. Disease risk is predicted to be greatest when there is extreme disparity between metabolic capacity and metabolic load. Whereas cohort studies link disease risk with poor metabolic capacity, prospective trials link it with increased metabolic load. Infancy is a developmental period in which nutrition can affect both metabolic capacity and metabolic load; this factor accounts for reported associations of both slow and fast infant growth with greater disease risk.

Type
Research Article
Copyright
Copyright © The Author 2009

Abbreviation:
RCT

randomised controlled trials

During the 20th century the burden of disease in industrialised countries shifted dramatically from infectious to non-infectious diseases as a result of changes in sanitation, hygiene, living conditions and nutrition(Reference Hicks and Allen1). In 1880 infectious, parasitic and respiratory diseases accounted for approximately 50% of all deaths in England and Wales, whereas cancers and diseases of the circulatory system accounted for <10%. By 1990 the corresponding percentages had altered to 17 and 70(Reference Hicks and Allen1). The modern killers became known as ‘lifestyle’ diseases, attributed to behavioural factors such as diet, sedentary behaviour and tobacco smoking, although both genetic factors and broader environmental factors such as pollution were also acknowledged to be important. Public health campaigns to reduce the prevalence of CVD and associated diseases such as type 2 diabetes, hypertension and stroke focused on dietary change (decreased consumption of fat, salt and refined carbohydrate, increased consumption of fruit, vegetables, vitamins and minerals), reductions in smoking and alcohol consumption and the promotion of leisure-time physical activity.

Within the last two decades a major shift in the understanding of the aetiology of these ‘lifestyle’ diseases has occurred, initiated by the analysis of several historical cohort studies conducted primarily in European populations. This work has led to the ‘developmental origins of adult disease’ hypothesis, proposing that patterns of growth and nutrition in early life are strongly predictive of later disease risk. Apparently-contradictory findings relating to this hypothesis now prompt re-evaluation of the strengths and limitations of historical cohort studies and the best way to interpret and utilise their findings in order to integrate them with findings from other study designs.

Historical cohort studies and the developmental origins of adult disease hypothesis

The first major published study deriving from a historical cohort emerged from the follow-up of young adult men, some of whom had been exposed to maternal famine during the Dutch ‘Hunger Winter’ at the end of the Second World War. A landmark publication showed that compared with adults who had not experienced maternal famine in utero, those who had been exposed had higher adult BMI if they had experienced the famine in the first trimester of pregnancy, but lower BMI if they had experienced the famine in the third trimester(Reference Ravelli, Stein and Susser2). These data were collectively suggested to comprise the induction of either adipocyte number or appetite regulation, depending on when the nutritional exposure occurred(Reference Ravelli, Stein and Susser2). A longer-term follow-up reported a similar association between early pregnancy exposure to famine and markers of obesity in middle-aged women, but did not reproduce the earlier findings in middle-aged men(Reference Ravelli, van der Meulen and Osmond3). Subsequent work in the same population, building on the ideas of Barker and colleagues(Reference Barker4Reference Hales, Barker and Clark6) (which are described later), linked early-pregnancy famine exposure with increased risk of CHD and atherogenic blood biochemistry, while also linking late-pregnancy exposure with increased risk of impaired glucose tolerance and hypertension(Reference Kyle and Pichaud7).

Beginning in the mid 1980s, groundbreaking work(Reference Barker4Reference Hales, Barker and Clark6) demonstrated inverse associations between birth weight and risk of the metabolic syndrome and CVD across the entire range of birth weight. The statistical significance of birth weight in these associations appeared to implicate fetal nutrition, and these studies in combination with the Dutch hunger winter data thus focused attention on poor nutritional supply in utero as a key factor predisposing to CVD and related diseases. This work inspired similar research in a variety of other populations. Over the last two decades, research into the developmental origins of adult diseases has been conducted in a variety of historical cohorts (summarised in Table 1). These historical cohorts are located in industrialised countries, and the data were originally collected because of concern with child health in the early 20th century(Reference Syddall, Aihie Sayer and Dennison8). Their recent exploitation has involved the follow-up of adults for whom data on maternal phenotype and on offspring growth and infant feeding were available, and in some cases tremendous effort has been required in order to re-establish a viable cohort of survivors for scientific study.

Table 1. Key historical and prospective cohort studies used in research on the developmental origins of adult disease

Approx, approximately.

* Not all those in the cohort were recruited at birth.

These historical cohort studies have consistently linked early-life experience with an increased risk of a variety of disease in adult life, including CVD, hypertension, stroke and type 2 diabetes(Reference Godfrey and Barker9). In 1991, capitalising on pioneering work that identified sensitive periods of development(Reference Widdowson and McCance10, Reference Davison and Dobbing11), the concept of ‘programming’ was proposed, whereby experience during early life exerts long-term effects on later phenotype(Reference Lucas12). Others have likewise referred to ‘metabolic imprinting’(Reference Waterland and Garza13), and there has been much interest in so-called ‘critical windows’ during which programming effects might be exerted. From an evolutionary perspective, the term programming has been criticised(Reference Bateson14), based on the argument that early-life experience does not contain specific ‘information’ about later disease. ‘Phenotypic induction’ was suggested as a more neutral term, in keeping with other areas of biology. ‘Programming’ has entered biomedical usage, and in relation to early life the two terms are to some extent interchangeable depending on the audience. However, phenotypic induction is a broader term with important benefits for integrating experience across the life-course and is used in the present article. One limitation of the ‘critical window’ approach is that physiological variability appears to involve different mechanisms at different life-course periods, and specific sensitive periods become increasingly difficult to define. The recent explosion of research into epigenetic modification of DNA expression has demonstrated one mechanism whereby nutrition in early life may generate long-term effects on phenotype(Reference Wolff, Kodell and Moore15), whereas later-life effects may be more attributable to incremental physiological changes.

There are a number of strengths characterising historical cohort studies. First, these large cohorts have in common data on birth weight, allowing certain hypotheses relating to associations between birth weight and later health or disease to be tested in different settings. Second, the participants are now middle-aged or elderly, hence analyses of their data offer robust evidence in relation to associations between early-life experience and actual morbidity or mortality in adult life. This factor represents a key strength because, although animal studies can generate life-course data over short time periods, the physiology of rodents may differ substantially from that of human subjects. Third, there is typically at least some data about the socio-economic status of the families, allowing some adjustment for this source of variability. The strengths of this approach have been demonstrated, for example, in the analyses of the Dutch Hunger Winter(Reference Ravelli, Stein and Susser2).

However, there are also a number of limitations to such studies. First, the data on early-life experience and confounders such as parental health and socio-economic status tend to be limited, and often incomplete. Second, observational studies of size or weight gain, with poor capacity to adjust statistically for confounders, are not adequate for demonstrating causation at the level of nutrition. As will be discussed, birth-weight data are especially hard to interpret and do not only index fetal experience. Finally, experience 60–90 years ago may not provide an appropriate model for contemporary populations(Reference Fewtrell, Kennedy and Singhal16). Given these limitations, researchers increasingly use other study designs to research the developmental origins of adult disease.

First, since the end of the Second World War onwards a number of large prospective birth cohorts have been established, as described in Table 1. These cohorts include national birth cohorts in the UK and more local cohorts in modernising countries. The early 1990s saw the initiation of the Avon Longitudinal Study of Parents and Children (ALSPAC), perhaps the most intensively studied cohort to date. Other cohorts have also been established recently in Brazil, the Philippines, South Africa and India. These newer cohorts, typically of several thousand individuals, have better data on early-life experience and potential confounding factors, more information on the continuous process of development and more frequent physiological measurements than do the historical cohorts. Broadly, they support the hypothesis that low birth weight is associated with increased adult disease risk(Reference Victora, Adair and Fall17).

Second, building on nutritional supplementation studies, researchers began applying the well-established pharmaceutical model to nutritional research. For example, between 1962 and 1977 all children within a group of Guatemalan villages were eligible for a food supplement, with two neighbouring villages given supplements differing in energy and protein content(Reference Martorell, Habicht and Rivera18). These children, now followed up as the ‘Human Capital Study’(Reference Grajeda, Behrman and Flores19), were however not randomised to the two diets and, although unlikely, it is technically possible for the villages themselves to have differed. Treating early-life nutrition from a pharmaceutical perspective, a series of randomised controlled trials (RCTs) were initiated whereby groups of preterm or term infants were allocated to different trial diets of known composition(Reference Lucas, Gore and Cole20Reference Fewtrell, Morley and Abbott22). These trials represent a better approach for adjusting for confounding factors, whether known or unknown(Reference Fewtrell, Kennedy and Singhal16), and are considered to produce the most robust evidence of causation. In particular, such trials are needed to demonstrate that early-life nutrition does indeed induce the later disease profile.

Nevertheless, randomised controlled trials (RCT) themselves suffer from certain limitations. First, they are difficult to perform at most ages (e.g. pregnancy or from early childhood onwards) and breast-feeding remains particularly difficult to study. These factors have resulted in widespread use of RCT with formula milks of differing composition but very little evidence for periods outside early infancy or for the kinds of total diets that individuals consume if not wholly formula-fed. Second, whilst they demonstrate the magnitude of the effect of the intervention, they cannot reveal the physiological mechanism or when during their administration they exert the effect.

Recently, it has become clear that the evidence from historical and prospective cohort studies and RCT contains major apparent contradictions. Whilst some researchers believe that fetal and infant undernutrition is key to the developmental induction of adult disease(Reference Barker23), others argue that overnutrition in the very same time periods is fundamental(Reference Singhal and Lucas24). This controversy risks polarising research into two opposing schools, detracting from the ability to develop a more holistic model of environmental impacts on disease risk. It also inhibits the formulation of public health policies, whereby the vast amount of research funding directed to the developmental origins hypothesis might generate tangible health benefits.

In the present article, it is argued that apparent controversy derives primarily from inconsistency in research outcomes, terminology, study designs and the interpretation of statistics, and not from either school being correct at the expense of the other. It is proposed that the different study designs emphasise different aspects of the total process of phenotypic induction, and that being more specific about what is induced during any particular developmental period aids the integration of the sum of data more successfully. This approach can therefore resolve the problem of how both under- and overnutrition during early life increase risk for a common set of diseases.

Controversies in the evidence

Birth weight is widely studied in part simply because the data are widely available(Reference Wilcox25). The interpretation of associations between low birth weight and later disease risk remains controversial, because birth weight is associated with postnatal growth rate as well as prenatal growth rate. As the statistical significance for these associations tends to depend on adjustment for adult weight, it has been argued that change in size between birth and adulthood is the primary determinant of risk, implicating postnatal growth rather than in utero development(Reference Lucas, Fewtrell and Cole26). For example, data from some RCT consistently demonstrate associations between increased infant growth rate and poorer subsequent metabolic profile(Reference Singhal, Cole and Lucas27Reference Singhal, Cole and Fewtrell29), paradoxically suggesting increased risk of the same diseases that have been associated in observational cohort studies with infant undernutrition (Table 2). According to the perspective discussed earlier(Reference Singhal and Lucas24), birth weight merely acts as a statistical marker of postnatal growth rate in historical cohort data and does not actually implicate fetal growth variability as the cause of adult disease variability.

Table 2. Associations between indices of reduced or greater infant growth with later disease

On the basis of RCT data the ‘growth acceleration hypothesis’ was proposed(Reference Singhal and Lucas24), which argues that rapid postnatal growth alone is the mechanism whereby early-life experience predicts subsequent risk. It is argued that because in each trial the groups tended to be very similar in terms of birth weight, there could be no contribution of fetal growth to the between-group differences in physiological markers of disease that were subsequently demonstrated. In support of the argument data are cited from animal studies associating early-life growth rate with later indices of function and health(Reference Metcalfe and Monaghan30). In these studies faster growth during development was shown to be associated with poorer outcome in a range of animal species. The Metcalfe–Monaghan model of ‘grow now, pay later’ was interpreted literally by arguing that the process of postnatal growth itself becomes harmful at higher rates(Reference Singhal and Lucas24).

This approach immediately raises the question as to why shorter adult stature and reduced leg length relative to total height are so widely associated with poorer health(Reference Smith, Greenwood and Gunnell31, Reference Schooling, Jiang and Heys32). From a broader perspective, animals vary profoundly in their rate of postnatal growth and this trait is a major feature of ‘life history’, which refers to the schedule of ontogenetic development, a key strategy whereby species adapt to different ecological niches. Ancestral humans (Homo erectus) are generally thought to have grown much more rapidly than modern humans(Reference Dean, Leakey and Reid33), and growth rate in Homo sapiens is considered to have become hormonally suppressed(Reference Sibly, Calow and Nichols34) to aid maternal reproductive energetics, as occurs in many social mammal species(Reference Bogin and Smith35). The idea that growth itself is inherently harmful is extremely simplistic, and the dynamics of the growth process require more careful evaluation.

Nevertheless, because both historical and prospective cohort and RCT data derive from large datasets and because each study design tends to generate findings consistent within itself, it is unlikely that either interpretation is wholly incorrect. Rather, it is necessary to be more specific about what exactly each study demonstrates and to understand how different statistical approaches emphasise different aspects of the biology of development.

Interpreting birth weight

Given the prominence of birth weight in many analyses, it is helpful to begin by exploring the notion that birth-weight variability reflects variability in the adequacy of fetal nutritional supply. Genetic factors undoubtedly influence fetal growth; however, studies are relatively consistent in attributing a minority of variability in size at birth to genetic factors(Reference Clausson, Lichtenstein and Cnattingius36, Reference Baird, Osmond and MacGregor37), indicating that most variability can be attributed to environmental factors. Logic suggests that low birth weight is indicative of undernutrition, and the Dutch famine data on those individuals exposed during the third trimester provide support for this hypothesis. Nevertheless, a recent study using ultrasound measurements to identify growth faltering in each trimester of pregnancy has revealed unexpected findings(Reference Hemachandra and Klebanoff38).

Whereas infants who faltered during the third trimester of pregnancy had lower weight and subscapular skinfold thickness relative to those who had not faltered during any trimester, those who faltered in the first trimester had higher values than all the other groups(Reference Hemachandra and Klebanoff38). This finding suggests that early faltering may result in a degree of catch-up growth in utero and that large birth weight need not necessarily indicate a lack of fetal growth faltering. This factor in turn may contribute to the paradoxical ‘U’-shaped association between birth weight and subsequent risk of obesity, as categorised by BMI(Reference Rogers39), and diabetes(Reference Kuh, Mishra and Black40).

Of importance for this debate are two issues: first, birth-weight variability is difficult to interpret as a marker of fetal growth faltering; second, fetal-growth variability (which is not the same as birth-weight variability) may nevertheless contribute to later disease risk. Confirming the latter hypothesis requires the effect of postnatal growth to be taken into account; however, this approach is challenging because although both early and later markers of growth have been associated with risk of same diseases, insufficient attention has been paid to the interpretation of different physiological markers of a given adult disease.

Research outcomes and different windows of induction

A second critical issue in unravelling apparent inconsistencies in the literature is the different outcomes assessed. Long-term cohort studies have typically been able to use mortality (e.g. death from CVD) or morbidity (e.g. diabetes, hypertension) as the primary outcome. Low birth weight consistently emerges as a strong predictor of such diseases, but in most cases only after adjustment for current weight. As these diseases emerge slowly throughout the life course, those conducting research on younger individuals are obliged to focus on proxy markers of disease, e.g. arterial distensibility rather than atherosclerosis, blood pressure rather than hypertension or fasting split pro-insulin levels (proposed to reflect insulin resistance) rather than diabetes(Reference Singhal, Cole and Lucas27Reference Singhal, Cole and Fewtrell29). For the youngest age-groups it is possible only to study physiological traits, such as size, body composition or hormonal concentrations, for which clear association with adult disease risk has yet to emerge.

This disparity in research outcomes is a major contributing factor to conceptual confusion, for two related reasons. First, it is difficult to integrate the different outcomes within a simple model of disease (see earlier). Second, study findings may on occasion conflict drastically, according to the age of the participants. This conflict can occur because not all those with apparent markers of disease at young ages go on to develop the disease. These issues may be clarified by elucidating how different periods of growth contribute to the induction of adult phenotype.

Growth may be broadly divided into two generic periods, one in which the offspring's immediate environment comprises maternal physiology, and a subsequent one in which the offspring is directly exposed to the external environment(Reference Wells41). Organogenesis occurs primarily within the first period, hence this critical period of development could be argued to be induced not by the external environment but by maternal phenotype(Reference Wells41, Reference Wells42). The duration of this maternal induction varies according to the trait in question. During pregnancy maternal and fetal haemodynamics interact, generating differential growth of peripheral v. central organs(Reference Barker, Bull and Osmond43). The offspring can respond adaptively, but its strategy is also open to maternal manipulation(Reference Wells41, Reference Haig44). This opportunity for direct haemodynamic interaction ceases at parturition. In contrast, offspring metabolism remains sensitive to maternal phenotype during the window of lactation, through hormonal effects on milk output and composition. Physiological traits such as nephron number(Reference Manalich, Reyes and Herrera45), cardiac structure(Reference Jiang, Godfrey and Martyn46) and pancreatic β-cell mass(Reference Hales and Barker47) are therefore strongly (although not necessarily exclusively) associated with fetal experience, whereas insulin metabolism is further sensitive to infant experience(Reference Plagemann, Harder and Franke48). It has been argued, from an evolutionary perspective, that the offspring adapts to the ‘niche’ of maternal metabolism and that the mother manipulates her offspring to optimise her own reproductive strategy(Reference Wells41, Reference Wells42). As many aspects of organ phenotype are essentially fixed from birth or early infancy onwards, as a result of the majority of rounds of cell division having been achieved(Reference Milner, Sharp, Fraser and Milner49), aspects of phenotype induced during this developmental period track subsequently into adulthood. Thus, reference can be made to the ‘maternal induction’ of offspring ‘metabolic capacity’, which is closely associated with certain aspects of organ phenotype.

The influence of maternal phenotype on offspring development must inevitably weaken with offspring age, but the schedule of this change depends on ecological factors. Until the development of agriculture, early postnatal growth would likewise have been strongly influenced by maternal phenotype, via lactation. Even after weaning, the tendency for human offspring to be provisioned by their mother would maintain a link between maternal phenotype and offspring growth rate(Reference Wells41). Such extended maternal care confers coherence on offspring development throughout the period of pregnancy, infancy and childhood(Reference Wells42). In recent millennia, most notably in industrialised populations, it has become possible for this maternal influence to be weakened or negated, exposing the offspring to stochastic growth patterns as early as 30 weeks post conception in those born preterm. It is this incoherence in nutritional experience that appears most strongly to predict subsequent disease.

Postnatal growth contrasts with prenatal growth by increasingly impacting on tissue size rather than fundamental structure, especially from late infancy. Postnatal growth, the sum of both size and somatic tissue (lean and fat), may thus be conceptualised primarily as generating ‘metabolic load’. The relative magnitude of growth, and the load it imposes, is associated with homeostatic adaptations in metabolic traits. For example, blood pressure increases systematically during the growth process(Reference Weder and Schork50), influenced independently by both somatic size (lean mass)(Reference Julius, Majahalme and Nesbitt51) and adiposity(Reference Voors, Webber and Frerichs52), with each contributing to the total metabolic load. Even weight gain in the first 3 months demonstrates such effects on blood pressure by 1 year of age(Reference Bansal, Ayoola and Gemmell53). These increases in blood pressure have been attributed to larger bodies imposing a greater load on the kidneys(Reference Weder and Schork50), invoking haemostatic changes in order to maintain renal homeostasis. In other words, variability in blood pressure during childhood and adulthood reflects the normalisation of metabolic load for a given metabolic capacity.

This process of normalisation evident for blood pressure is merely one example of a broader pattern whereby metabolic traits mediate the impact of growth-generated metabolic load on metabolic capacity. For example, insulin resistance and insulin secretion are closely related, such that in healthy individuals insulin resistance can be accommodated by increases in insulin secretion to maintain glycaemic control(Reference Bergman, Ader and Huecking54). However, the understanding of this normalisation perspective remains incomplete because of the tendency to target research primarily at those who are unwell, unduly emphasising, for example, the deleterious aspects of hypertension rather than the adaptive function that occurs over the entire range of blood pressure.

The notion that disease is the result of disparity between metabolic capacity and metabolic load was essentially described in the thrifty phenotype hypothesis(Reference Hales and Barker47), and this conceptual model remains capable of integrating the findings of diverse developmental origins research studies. What has not been sufficiently appreciated as yet is the shift in the target of growth, from metabolic capacity to metabolic load, and the effects of this shift on the induction of diseases.

Disentangling the contributions of fetal v. postnatal growth

Differentiating growth into periods targeting metabolic capacity or metabolic load clarifies in several ways the apparent controversy between the ‘low birth weight’ and ‘growth acceleration’ schools.

First, it is easier to appreciate apparent inconsistencies relating to outcomes. Low birth weight has been associated with risk of diabetes(Reference Barker, Hales and Fall55), while rapid infant growth has been associated with insulin resistance(Reference Singhal, Fewtrell and Cole28). However, since diabetes represents a ‘two hit’ phenomenon, in which insulin resistance is accompanied by β-cell defect preventing compensations in insulin secretion(Reference Bergman, Ader and Huecking54), the epidemiology of diabetes need not be identical to the epidemiology of insulin resistance. Put simply, the life-course induction of physiology (e.g. β-cell mass, nephron number) differs from that of physiological function (e.g. glycaemic control, blood pressure), which in turn differs from that of disease (e.g. diabetes, hypertension). Whether the insulin resistance induced by faster infant growth(Reference Singhal, Fewtrell and Cole28) actually leads to diabetes in later life is likely to be mediated by β-cell function, and hence by fetal developmental experience.

Second, it is possible to resolve apparent inconsistencies relating to statistical models. Studies of older adults with CVD show that, holding current weight constant, those with low birth weight have an increased disease risk(Reference Victora, Adair and Fall17, Reference Barker, Eriksson and Forsén56), which implicates the induction of metabolic capacity in early life through strong effects on traits such as nephron number or β-cell mass, as discussed earlier. Studies of younger adults show that rapid infant weight gain is associated with increased physiological markers of risk, regardless of birth weight(Reference Singhal, Cole and Lucas27, Reference Singhal, Fewtrell and Cole28), which implies the environmental induction of increased metabolic load. The greatest disparity between metabolic load and metabolic capacity occurs in those individuals who are born small and become large, for whom the risk of CVD and the metabolic syndrome is greatly increased(Reference Bavdekar, Yajnik and Fall57Reference Byrne and Phillips59). Recent studies further suggest that different periods of childhood growth impact differentially on the risk of specific diseases(Reference Barker, Osmond and Forsén60, Reference Forsén, Eriksson and Tuomilehto61). Each statistical approach may therefore be considered merely to emphasise one component of the dynamic process whereby metabolic load is superimposed on metabolic capacity.

Third, it is beneficial to differentiate between study designs and possible sensitive periods of growth. The findings of intervention studies conducted in both preterm and full-term infants during the immediate postnatal period have been used to argue that postnatal growth is implicated as the key ‘critical window’ in which susceptibility to CVD is induced(Reference Singhal and Lucas24). However, such studies only demonstrate the magnitude of effect of an intervention conducted during a specific period. As RCT typically begin in the immediate postnatal period, when (under natural conditions) breast-fed infants normally receive little nutritional intake and lose up to 10% of their birth weight(Reference Evans, Evans and Royal62), the interventions are likely to overload metabolism in this particularly sensitive period of development, and hence disproportionately target the induction of metabolic load rather than metabolic capacity. Where catch-up growth is distributed over longer time periods it is possible that it may more successfully target metabolic capacity, and given the benefits of catch-up growth for survival in many settings(Reference Victora, Barrosa and Hortaa63) this hypothesis requires further research. Metabolic load may be enhanced throughout postnatal life, not only by infant or childhood ‘growth acceleration’ as suggested, but also by adult obesity once growth has ceased.

This interpretation does not therefore support the model that CVD is essentially induced by postnatal experience alone, as suggested previously(Reference Singhal and Lucas24). Although it is claimed that the hypothesis of ‘growth acceleration’ being harmful is supported by data from a range of animal species, the vast majority of such animal studies describe catch-up growth following initial nutritional insult, while the remaining studies involve genetically variant animals(Reference Singhal, Cole and Lucas27) and so cannot attribute adult phenotypic variability to early-life environmental variability. Again, these animal studies implicate an inherent link between different growth periods in the aetiology of later disease, one earlier period exerting deleterious effects on metabolic capacity and the other imposing an increased metabolic load on that capacity.

Fourth, it is possible to differentiate alternative pathways whereby public health policies might beneficially impact on health. The close association between birth weight and subsequent lean mass(Reference Wells, Chomtho and Fewtrell64), and the fact that organogenesis is largely completed by birth, suggests that interventions on the mother, during her own development or during pregnancy, may represent the optimum approach for benefitting offspring metabolic capacity. Birth-weight supplementation studies tend to have limited efficacy(Reference Wells41, Reference Wells42), hence nutritional intervention earlier in the maternal life course may be most successful. Animal studies, for example, show that maternal effects continue to act on offspring phenotype across several generations(Reference Price, Hyde and Coe65). Alternatively, during pregnancy maternal metabolic control may represent a better candidate for intervention than maternal diet(Reference Wells41), as it is the nutrient concentration gradients between fetus and mother that determine fetal supply. In contrast, postnatal interventions may prove most beneficial in preventing the induction of excess metabolic load. Public health policies such as breast-feeding and the promotion of healthy childhood diets and activity levels all have the capacity to target this component of development.

It is well established that babies with low birth weight tend to experience catch-up growth in postnatal life(Reference Ong, Ahmed and Emmett66). In the Avon Longitudinal Study of Parents and Children cohort indices of in utero growth retardation were found to be strongly associated with rapid postnatal growth rate (Reference Ong, Ahmed and Emmett66), while researchers using animal models have likewise admitted that they are powerless to prevent such catch-up occurring under natural conditions (S Ozanne, personal communication). Although the magnitude of this catch-up may potentially be regulated through control of diet or feeding schedule, the capacity for such manipulation is limited in breast-fed infants in whom nutritional intake is essentially ‘invisible’. Associations between birth weight and adult disease risk thus partly reflect the fact that low birth weight is associated with both poor fetal growth and rapid infant growth. However, excess childhood, adolescent and adult weight gain all increase metabolic load, and the deleterious effect of low birth weight cannot therefore be attributed entirely to its inverse association with infant growth, as is emphasised by recent studies(Reference Barker, Osmond and Forsén60, Reference Forsén, Eriksson and Tuomilehto61, Reference Law, Shiell and Newsome67).

How different study designs merely emphasise different components of the process whereby nutritional experience contributes to adult disease risk can now be reconsidered. The historical cohort studies highlight how, holding adult weight (or metabolic load) constant, disease risk is greatest in those born small. The physiological evidence attributes this effect of low birth weight primarily to reduced metabolic capacity, deriving from traits such as compromised β-cell mass, nephron number and cardiac structure although low birth weight also contributes to disease risk by inducing catch-up growth. Postnatal RCT demonstrate how increasing metabolic load exacerbates disease risk. This outcome can be achieved in any individual, regardless of initial size, but studies consistently show that those of low birth weight tolerate the increased load least well. Fig. 1 illustrates the different aspects of a single model of the developmental origins of disease that are investigated by observational cohort studies v. RCT.

Fig. 1. Schematic diagram illustrating the different emphases of randomised controlled trials (RCT) v. historical cohort studies. Disease is attributed to a high metabolic load (deriving from postnatal growth) being exerted on metabolic capacity (organ phenotype determined primarily by fetal experience). RCT investigate the effects of growth on metabolic load and show that rapid weight gain exacerbates metabolic load independent of birth size (A). In contrast, historical cohort studies show that holding current size (i.e. metabolic load) constant, disease risk is greatest in those born small, which can be attributed to their reduced metabolic capacity (B). (), Metabolic capacity; (), metabolic load.

The conflicting results for infancy presented in Table 2 can be attributed to the fact that this period of development incorporates both the later stages of the development of metabolic capacity and the early stages of the generation of metabolic load. Cohort studies associating poor infant weight gain with later disease risk indicate the continued suppression of metabolic capacity. For example, although β-cell mass is strongly associated with birth weight, it continues to increase during infancy(Reference Bouwens and Rooman68) and is therefore predicted to be deleteriously influenced by poor infant weight gain. In contrast, RCT associating faster infant growth with increased adolescent blood pressure(Reference Singhal, Cole and Lucas27) or insulin resistance(Reference Singhal, Fewtrell and Cole28) imply the imposition of increased metabolic load, possibly through overloading metabolism during the sensitive period immediately after birth. To elucidate this issue further, RCT manipulating growth should be delayed until after the first weeks of life and should then generate physiologically-relevant differences in infant weight gain, rather than extremes. More broadly, the optimal magnitude of infant growth for later health remains a key topic for research, especially in those born small, with the optimal level likely to reflect a trade-off between counterbalancing risks. Fig. 2 illustrates the shifting target of growth across the life course and the contrasting effects of infant undernutrition v. overnutrition.

Fig. 2. Schematic diagram illustrating the change in the target of growth with increasing age through the life course. Early-life growth primarily promotes metabolic capacity, whereas later growth induces metabolic load. Infancy reflects a time period when both outcomes can be influenced, with the relative effect on metabolic capacity v. metabolic load varying according to current nutritional status and previous growth patterns. Such a model can explain why both poor and rapid growth in infancy have been associated in different studies with poorer health outcomes.

Conclusions

The present article has aimed to probe the value of historical cohort studies in research into the developmental origins of adult disease. Recently, apparent contradictions in the scientific evidence for such disease aetiology have led to some researchers questioning the value, or interpretation, of these observational studies. It has been argued that these contradictions can be attributed to inconsistencies between studies in their design, study populations, outcomes and statistical interpretation. Understanding the changing association between growth and metabolic capacity v. metabolic load during the process of ontogenetic development improves the integration of research findings and clarifies two generic complementary pathways whereby public health interventions may reduce the risk of non-communicable diseases. Optimal adult health is predicted to derive from minimising two counterbalancing risks, i.e. the development of poor metabolic capacity v. the development of excess metabolic load. Infancy appears a particularly sensitive component of development because both risks manifest during this period. Public health policies may address each of these risks, and a coherent approach across the life course will avoid the disparity of growth between different developmental periods that appears most strongly associated with disease.

Acknowledgements

This review article involved no specific grant from any funding agency in the public, commercial or not-for-profit sectors. The author declares no conflict of interest.

References

1. Hicks, J & Allen, G (1999) A century of change: trends in UK statistics since 1990. Research paper 99/111. http://www.parliament.uk/commons/lib/research/rp99/rp99–111.pdfGoogle Scholar
2. Ravelli, GP, Stein, ZA & Susser, MW (1976) Obesity in young men after famine exposure in utero and early infancy. N Engl J Med 295, 349353.CrossRefGoogle ScholarPubMed
3. Ravelli, AC, van der Meulen, JHP, Osmond, C et al. (1999) Obesity at the age of 50 y in men and women exposed to famine prenatally. Am J Clin Nutr 70, 811816.CrossRefGoogle ScholarPubMed
4. Barker, DJ (1991) The intrauterine environment and adult cardiovascular disease. Ciba Found Symp 156, 310.Google ScholarPubMed
5. Barker, DJ (1998) Mothers, Babies and Health in Later Life. Edinburgh: Churchill.Google Scholar
6. Hales, CN, Barker, DJ, Clark, PM et al. (1991) Fetal and infant growth and impaired glucose tolerance at age 64. Br Med J 303, 10191022.CrossRefGoogle ScholarPubMed
7. Kyle, UG & Pichaud, C (2006) The Dutch famine of 1944–1945: a pathophysiological model of long-term consequences of wasting disease. Curr Opin Clin Nutr Metab Care 9, 388394.CrossRefGoogle ScholarPubMed
8. Syddall, HE, Aihie Sayer, A, Dennison, EM et al. (2005) Cohort profile: the Hertfordshire cohort study. Int J Epidemiol 34, 12341242.CrossRefGoogle ScholarPubMed
9. Godfrey, KM & Barker, DJ (2000) Fetal nutrition and adult disease. Am J Clin Nutr 71, Suppl., 1344S1352S.CrossRefGoogle ScholarPubMed
10. Widdowson, EM & McCance, RA (1960) Some effects of accelerating growth. I. General somatic development. Proc R Soc Lond B Biol Sci 152, 188206.Google ScholarPubMed
11. Davison, AN & Dobbing, J (editors) (1968) The developing brain. In Applied Neurochemistry, pp. 253286. Oxford: Blackwell.Google Scholar
12. Lucas, A (1991) Programming by early nutrition in man. Ciba Found Symp 156, 3850.Google ScholarPubMed
13. Waterland, RA & Garza, C (1999) Potential mechanisms of metabolic imprinting that lead to chronic disease. Am J Clin Nutr 69, 179197.CrossRefGoogle ScholarPubMed
14. Bateson, P (2001) Fetal experience and good adult design. Int J Epidemiol 30, 928934.CrossRefGoogle ScholarPubMed
15. Wolff, GL, Kodell, RL, Moore, SR et al. (1998) Maternal epigenetics and methyl supplements affect agouti gene expression in Avy/a mice. FASEB J 12, 949957.CrossRefGoogle ScholarPubMed
16. Fewtrell, MS, Kennedy, K, Singhal, A et al. (2008) How much loss to follow-up is acceptable in long-term randomised trials and prospective studies? Arch Dis Child 93, 458461.CrossRefGoogle ScholarPubMed
17. Victora, CG, Adair, L, Fall, C et al. (2008) Maternal and child undernutrition: consequences for adult health and human capital. Lancet 371, 340357.CrossRefGoogle ScholarPubMed
18. Martorell, R, Habicht, JP & Rivera, JA (1995) History and design of the INCAP longitudinal study (1969–77) and its follow-up (1988–89). J Nutr 125, Suppl., 1027S1041S.Google ScholarPubMed
19. Grajeda, R, Behrman, JR, Flores, R et al. (2005) The human capital study 2002–04: tracking, data collection, coverage, and attrition. Food Nutr Bull 26, Suppl. 1, S15S24.CrossRefGoogle ScholarPubMed
20. Lucas, A, Gore, SM, Cole, TJ et al. (1984) Multicentre trial on feeding low birthweight infants: effects of diet on early growth. Arch Dis Child 59, 722730.CrossRefGoogle ScholarPubMed
21. Lucas, A, Fewtrell, MS, Morley, R et al. (2001) Randomized trial of nutrient-enriched formula versus standard formula for postdischarge preterm infants. Pediatrics 108, 703711.CrossRefGoogle ScholarPubMed
22. Fewtrell, MS, Morley, R, Abbott, RA et al. (2001) Catch-up growth in small-for-gestational-age term infants: a randomized trial. Am J Clin Nutr 74, 516523.CrossRefGoogle ScholarPubMed
23. Barker, DJ (2007) The origins of the developmental origins theory. J Intern Med 261, 412417.CrossRefGoogle ScholarPubMed
24. Singhal, A & Lucas, A (2004) Early origins of cardiovascular disease: is there a unifying hypothesis? Lancet 363, 16421645.CrossRefGoogle Scholar
25. Wilcox, AJ (2001) On the importance – and the unimportance – of birthweight. Int J Epidemiol 30, 12331241.CrossRefGoogle ScholarPubMed
26. Lucas, A, Fewtrell, MS & Cole, TJ (1999) Fetal origins of adult disease – the hypothesis revisited. Br Med J 319, 245249.CrossRefGoogle ScholarPubMed
27. Singhal, A, Cole, TJ & Lucas, A (2001) Early nutrition in preterm infants and later blood pressure: two cohorts after randomised trials. Lancet 357, 413419.CrossRefGoogle ScholarPubMed
28. Singhal, A, Fewtrell, M, Cole, TJ et al. (2003) Low nutrient intake and early growth for later insulin resistance in adolescents born preterm. Lancet 361, 10891097.CrossRefGoogle ScholarPubMed
29. Singhal, A, Cole, TJ, Fewtrell, M et al. (2004) Is slower early growth beneficial for long-term cardiovascular health? Circulation 109, 11081113.CrossRefGoogle ScholarPubMed
30. Metcalfe, NB & Monaghan, P (2001) Compensation for a bad start: grow now, pay later? Trends Ecol Evol 16, 254260.CrossRefGoogle ScholarPubMed
31. Smith, GD, Greenwood, R, Gunnell, D et al. (2001) Leg length, insulin resistance, and coronary heart disease risk: the Caerphilly Study. J Epidemiol Community Health 55, 867872.CrossRefGoogle ScholarPubMed
32. Schooling, CM, Jiang, CQ, Heys, M et al. (2007) Height, its components, and cardiovascular risk among older Chinese: a cross-sectional analysis of the Guangzhou Biobank Cohort Study. Am J Public Health 97, 18341841.CrossRefGoogle ScholarPubMed
33. Dean, C, Leakey, MG, Reid, D et al. (2001) Growth processes in teeth distinguish modern humans from Homo erectus and earlier hominins. Nature 414, 628631.CrossRefGoogle ScholarPubMed
34. Sibly, R, Calow, P & Nichols, N (1985) Are patterns of growth adaptive? J Theor Biol 112, 553574.CrossRefGoogle ScholarPubMed
35. Bogin, B & Smith, BH (1996) Evolution of the human life cycle. Am J Hum Biol 8, 703716.3.0.CO;2-U>CrossRefGoogle ScholarPubMed
36. Clausson, B, Lichtenstein, P & Cnattingius, S (2000) Genetic influence on birthweight and gestational length determined by studies in offspring of twins. Br J Obstet Gynaecol 107, 375381.CrossRefGoogle ScholarPubMed
37. Baird, J, Osmond, C, MacGregor, A et al. (2001) Testing the fetal origins hypothesis in twins: the Birmingham twin study. Diabetologia 44, 3339.CrossRefGoogle ScholarPubMed
38. Hemachandra, AH & Klebanoff, MA (2006) Use of serial ultrasound to identify periods of fetal growth restriction in relation to neonatal anthropometry. Am J Hum Biol 18, 791797.CrossRefGoogle ScholarPubMed
39. Rogers, I (2003) The influence of birthweight and intrauterine environment on adiposity and fat distribution in later life. Int J Obes Relat Metab Disord 27, 755777.CrossRefGoogle ScholarPubMed
40. Kuh, D, Mishra, GD, Black, S et al. (2008) Offspring birth weight, gestational age and maternal characteristics in relation to glucose status at age 53 years: evidence from a national birth cohort. Diabet Med 25, 530535.CrossRefGoogle ScholarPubMed
41. Wells, JC (2007) The thrifty phenotype as an adaptive maternal effect. Biol Rev Camb Philos Soc 82, 143172.CrossRefGoogle ScholarPubMed
42. Wells, JC (2003) The thrifty phenotype hypothesis: thrifty offspring or thrifty mother? J Theor Biol 221, 143161.CrossRefGoogle ScholarPubMed
43. Barker, DJ, Bull, AR, Osmond, C et al. (1990) Fetal and placental size and risk of hypertension in adult life. Br Med J 301, 259262.CrossRefGoogle ScholarPubMed
44. Haig, D (1993) Genetic conflicts in human pregnancy. Q Rev Biol 68, 495532.CrossRefGoogle ScholarPubMed
45. Manalich, R, Reyes, L, Herrera, M et al. (2000) Relationship between weight at birth and the number and size of renal glomeruli in humans: a histomorphometric study. Kidney Int 58, 770773.CrossRefGoogle ScholarPubMed
46. Jiang, B, Godfrey, KM, Martyn, CN et al. (2006) Birth weight and cardiac structure in children. Pediatrics 117, e257e261.CrossRefGoogle ScholarPubMed
47. Hales, CN & Barker, DJ (1992) Type 2 (non-insulin-dependent) diabetes mellitus: the thrifty phenotype hypothesis. Diabetologia 35, 595601.CrossRefGoogle ScholarPubMed
48. Plagemann, A, Harder, T, Franke, K et al. (2002) Long-term impact of neonatal breast-feeding on body weight and glucose tolerance in children of diabetic mothers. Diabetes Care 25, 1622.CrossRefGoogle ScholarPubMed
49. Milner, RDG (1989) Mechanisms of overgrowth. In Fetal growth. Proceedings of the Twentieth Study Group of the Royal College of Obstetricians and Gynaecologists, pp. 139148 [Sharp, F, Fraser, RB and Milner, RDG, editors]. London: Royal College of Obstetricians and Gynaecologists.Google Scholar
50. Weder, AB & Schork, NJ (1994) Adaptation, allometry, and hypertension. Hypertension 24, 145156.CrossRefGoogle ScholarPubMed
51. Julius, S, Majahalme, S, Nesbitt, S et al. (2002) A ‘gender blind’ relationship of lean body mass and blood pressure in the Tecumseh study. Am J Hypertens 15, 258263.CrossRefGoogle ScholarPubMed
52. Voors, AW, Webber, LS, Frerichs, RR et al. (1977) Body height and body mass as determinants of basal blood pressure in children – The Bogalusa Heart Study. Am J Epidemiol 106, 101108.CrossRefGoogle ScholarPubMed
53. Bansal, N, Ayoola, OO, Gemmell, I et al. (2008) Effects of early growth on blood pressure of infants of British European and South Asian origin at one year of age: the Manchester children's growth and vascular health study. J Hypertens 26, 412418.CrossRefGoogle ScholarPubMed
54. Bergman, RN, Ader, M, Huecking, K et al. (2002) Accurate assessment of beta-cell function: the hyperbolic correction. Diabetes 51, Suppl. 1, S212S220.CrossRefGoogle ScholarPubMed
55. Barker, DJP, Hales, CN, Fall, CHD et al. (1993) Type 2 (non-insulin-dependent) diabetes mellitus, hypertension and hyperlipidaemia (syndrome X): relation to reduced fetal growth. Diabetologia 36, 6267.CrossRefGoogle ScholarPubMed
56. Barker, DJ, Eriksson, JG, Forsén, T et al. (2002) Fetal origins of adult disease: strength of effects and biological basis. Int J Epidemiol 31, 12351239.CrossRefGoogle ScholarPubMed
57. Bavdekar, A, Yajnik, CS, Fall, CH et al. (1999) Insulin resistance syndrome in 8-year-old Indian children – Small at birth, big at 8 years, or both? Diabetes 48, 24222429.CrossRefGoogle Scholar
58. Adair, LS & Cole, TJ (2003) Rapid child growth raises blood pressure in adolescent boys who were thin at birth. Hypertension 41, 451456.CrossRefGoogle ScholarPubMed
59. Byrne, CD & Phillips, DI (2000) Fetal origins of adult disease: epidemiology and mechanisms. J Clin Pathol 53, 822828.CrossRefGoogle ScholarPubMed
60. Barker, DJ, Osmond, C, Forsén, TJ et al. (2005) Trajectories of growth among children who have coronary events as adults. N Engl J Med 353, 18021809.CrossRefGoogle ScholarPubMed
61. Forsén, T, Eriksson, J, Tuomilehto, J et al. (2000) The fetal and childhood growth of persons who develop type 2 diabetes. Ann Intern Med 133, 176182.CrossRefGoogle ScholarPubMed
62. Evans, KC, Evans, R, Royal, R et al. (2003) Effect of caesarean section on breast milk transfer to the normal term newborn over the first week of life. Arch Dis Child Fetal Neonatal Ed 88, F380F382.CrossRefGoogle Scholar
63. Victora, CG, Barrosa, FC, Hortaa, BL et al. (2001) Short-term benefits of catch-up growth for small-for-gestational-age infants. Int J Epidemiol 30, 13251330.CrossRefGoogle ScholarPubMed
64. Wells, JC, Chomtho, S & Fewtrell, MS (2007) Programming of body composition by early growth and nutrition. Proc Nutr Soc 66, 423434.CrossRefGoogle ScholarPubMed
65. Price, KC, Hyde, JS & Coe, CL (1999) Matrilineal transmission of birth weight in the rhesus monkey (Macaca mulatta) across several generations. Obstet Gynecol 94, 128134.Google ScholarPubMed
66. Ong, KK, Ahmed, ML, Emmett, PM et al. (2000) Association between postnatal catch-up growth and obesity in childhood: prospective cohort study. Br Med J 320, 967971.CrossRefGoogle ScholarPubMed
67. Law, CM, Shiell, AW, Newsome, CA et al. (2002) Fetal, infant, and childhood growth and adult blood pressure: a longitudinal study from birth to 22 years of age. Circulation 105, 10881092.CrossRefGoogle ScholarPubMed
68. Bouwens, L & Rooman, I (2005) Regulation of pancreatic beta-cell mass. Physiol Rev 85, 12551270.CrossRefGoogle ScholarPubMed
69. Leon, DA, Lithell, HO, Vågerö, D et al. (1998) Reduced fetal growth rate and increased risk of death from ischaemic heart disease: cohort study of 15 000 Swedish men and women born 1915–29. Br Med J 317, 241245.CrossRefGoogle Scholar
70. Elwood, PC, Pickering, J, Gallacher, JEJ et al. (2005) Long term effect of breast feeding: cognitive function in the Caerphilly cohort. J Epidemiol Community Health 59, 130133.CrossRefGoogle ScholarPubMed
71. Eriksson, JG (2007) Epidemiology, genes and the environment: lessons learned from the Helsinki Birth Cohort Study. J Intern Med 261, 418425.CrossRefGoogle ScholarPubMed
72. Martin, RM, Gunnell, D, Pemberton, J et al. (2005) Cohort profile: The Boyd Orr cohort – an historical cohort study based on the 65 year follow-up of the Carnegie Survey of Diet and Health (1937–39). Int J Epidemiol 34, 742749.CrossRefGoogle ScholarPubMed
73. Lumey, LH, Stein, AD, Kahn, HS et al. (2007) Cohort profile: the Dutch Hunger Winter families study. Int J Epidemiol 36, 11961204.CrossRefGoogle ScholarPubMed
74. Leon, DA, Lawlor, DA, Clark, H et al. (2006) Cohort profile: the Aberdeen children of the 1950s study. Int J Epidemiol 35, 549552.CrossRefGoogle ScholarPubMed
75. Wadsworth, M, Kuh, D, Richards, M et al. (2006) Cohort Profile: The 1946 National Birth Cohort (MRC National Survey of Health and Development). Int J Epidemiol 35, 4954.CrossRefGoogle ScholarPubMed
76. Power, C & Elliott, J (2006) Cohort profile: 1958 British birth cohort (National Child Development Study). Int J Epidemiol 35, 3441.CrossRefGoogle ScholarPubMed
77. Bhargava, SK, Sachdev, HS, Fall, CH et al. (2004) Relation of serial changes in childhood body-mass index to impaired glucose tolerance in young adulthood. N Engl J Med 350, 865875.CrossRefGoogle ScholarPubMed
78. Antonisamy, B, Raghupathy, P, Christopher, S et al. (2008) Cohort Profile: The 1969–73 Vellore birth cohort study in South India. Int J Epidemiol (Epublication ahead of print version).Google ScholarPubMed
79. Elliott, J & Shepherd, P (2006) Cohort profile: 1970 British Birth Cohort (BCS70). Int J Epidemiol 35, 836843.CrossRefGoogle ScholarPubMed
80. Victora, CG & Barros, FC (2006) Cohort profile: the 1982 Pelotas (Brazil) birth cohort study. Int J Epidemiol 35, 237242.CrossRefGoogle ScholarPubMed
81. Tudor-Locke, C, Ainsworth, BE, Adair, LS et al. (2003) Physical activity in Filipino youth: the Cebu Longitudinal Health and Nutrition Survey. Int J Obes Relat Metab Disord 27, 181190.CrossRefGoogle ScholarPubMed
82. Richter, L, Norris, S, Pettifor, J et al. (2007) Cohort Profile: Mandela's children: the 1990 Birth to Twenty study in South Africa. Int J Epidemiol 36, 504511.CrossRefGoogle ScholarPubMed
83. Golding, J, Pembrey, M & Jones, R (2001) ALSPAC – the Avon Longitudinal Study of Parents and Children. I. Study methodology. Paediatr Perinat Epidemiol 15, 7487.CrossRefGoogle ScholarPubMed
84. Victora, CG, Hallal, PC, Araújo, CLP et al. (2008) Cohort profile: the 1993 Pelotas (Brazil) birth cohort study. Int J Epidemiol 37, 704709.CrossRefGoogle ScholarPubMed
85. Eriksson, JG, Osmond, C, Kajantie, E et al. (2006) Patterns of growth among children who later develop type 2 diabetes or its risk factors. Diabetologia 49, 28532858.CrossRefGoogle ScholarPubMed
86. Eriksson, JG, Forsen, T, Tuomilehto, J et al. (2001) Early growth and coronary heart disease in later life: longitudinal study. Br Med J 322, 949953.CrossRefGoogle ScholarPubMed
87. Gaskin, PS, Walker, SP, Forrester, TE et al. (2000) Early linear growth retardation and later blood pressure. Eur J Clin Nutr 54, 563567.CrossRefGoogle ScholarPubMed
Figure 0

Table 1. Key historical and prospective cohort studies used in research on the developmental origins of adult disease

Figure 1

Table 2. Associations between indices of reduced or greater infant growth with later disease

Figure 2

Fig. 1. Schematic diagram illustrating the different emphases of randomised controlled trials (RCT) v. historical cohort studies. Disease is attributed to a high metabolic load (deriving from postnatal growth) being exerted on metabolic capacity (organ phenotype determined primarily by fetal experience). RCT investigate the effects of growth on metabolic load and show that rapid weight gain exacerbates metabolic load independent of birth size (A). In contrast, historical cohort studies show that holding current size (i.e. metabolic load) constant, disease risk is greatest in those born small, which can be attributed to their reduced metabolic capacity (B). (), Metabolic capacity; (), metabolic load.

Figure 3

Fig. 2. Schematic diagram illustrating the change in the target of growth with increasing age through the life course. Early-life growth primarily promotes metabolic capacity, whereas later growth induces metabolic load. Infancy reflects a time period when both outcomes can be influenced, with the relative effect on metabolic capacity v. metabolic load varying according to current nutritional status and previous growth patterns. Such a model can explain why both poor and rapid growth in infancy have been associated in different studies with poorer health outcomes.