Over the last few years there has been increasing interest in conceptualizing disease aetiology within a life course framework.1,2 This approach is not new to Public Health or unique to epidemiology (see below). However, its current resonance and interest within epidemiology reflects the challenging theoretical framework this approach provides. This issue of the International Journal of Epidemiology has several papers with a ‘life course theme’. This accompanying editorial is intended to highlight what we believe are the key conceptual issues around life course epidemiology. We have chosen to use examples from chronic disease epidemiology, but this approach is also applicable within the context of infectious diseases3 and wider notions of health and wellbeing.4 We have defined a life course approach to chronic disease epidemiology1 as the study of long-term effects on chronic disease risk of physical and social exposures during gestation, childhood, adolescence, young adulthood and later adult life. It includes studies of the biological, behavioural and psychosocial pathways that operate across an individual's life course, as well as across generations, to influence the development of chronic diseases. Conventionally, chronic disease cohort studies recruit subjects in mid-life and follow them up for future disease end-points. The risk of developing disease is then related to baseline exposures or changes in exposure measures ascertained at further follow-ups. Even when baseline measures include early life exposures, such as birthweight and childhood socioeconomic position, these would usually be entered into a multivariable model without much attention to the temporal relationship between variables. Merely the collection of exposure data across the life course is not synonymous with a life course model of disease causation. Surprisingly few epidemiological publications explicitly state the temporal ordering of exposure variables and their inter-relationships, both directly or through intermediary variables, with the outcome measure. One example, where this approach was explicitly undertaken was a study testing the influence of early and later life factors on carotid intima thickness.5 This diagramatically ordered classes of variables across the life course. Such an approach is commonplace in structural equation modelling, path analysis and graphical models where prior conceptual representations, before statistical modelling, are standard practice.6,7 Figure 1 illustrates such a conceptualization with respect to adult respiratory disease and/or impaired respiratory function. This figure illustrates many potential pathways between intrauterine growth and adult disease. Such a life course model enables the researcher to explicitly test not only early life course exposures with later disease, but possible pathways with potential intermediaries or confounding factors. Path (a) would represent a predominantly biological pathway whereby impaired fetal development of the lung architecture is associated with future respiratory insults from infectious agents and greater susceptibility to impaired lung function in adulthood and/or chronic obstructive airways disease. Path (b) would be a predominantly social pathway whereby adverse childhood socioeconomic position influences adverse childhood exposures as well as adult socioeconomic position and smoking behaviour. Path (c) reflects a socio-biological pathway whereby adverse childhood socioeconomic position is associated with post-natal lung function and subsequently with poor adult lung function through its effects on immune function and the likelihood of exposure to infectious agents. Path (d) is a bio-social pathway so that repeated childhood infections results in adverse educational attainment and lower adult socioeconomic position. Even such a crude model highlights the complex inter-relationships and rather arbitrary differentiation between biological and social mechanisms. As Krieger8 asserts a ‘simplistic division of the social and biological will not suffice’. Even such a simplified model (see Strachan9 for detailed discussion) is daunting but importantly challenges both clinical epidemiologists and social scientists to operationalize exposures and conceptualize their inter-relationships across the life course. As well as integrating biological and psychosocial pathways, a life course approach essentially requires some understanding of the natural history and physiological trajectory of normal biological systems (Figure 2). As can be seen for lung function, (but it applies to many other continuous physiological measures e.g. muscle strength, cognitive function), different periods across the life course influence phases of biological development, stability or decline. This alternative diagrammatic representation of disease aetiology is important for understanding how exposures may differentially act in critical and/or sensitive periods (see below). For example, an exposure acting in early life may adversely affect lung function during the development period (line B) resulting in a diminished physiological reserve without having any appreciable effect on the rate of decline (line C) Given the wide range of potential exposures and the importance of the timing of an exposure, it is unsurprising that exposures may affect disease risk in more than one way. We have previously proposed a simple classification (Table 1) of potential life course models of health.10 Some have misinterpreted a life course approach as precluding a critical period model and only encompassing accumulation of risk.11 The ‘critical period model’ is when an exposure acting during a specific period has lasting or lifelong effects on the structure or function of organs, tissues and body systems which are not modified in any dramatic way by later experience. This is also known as ‘biological programming’ or it is sometimes referred to as a ‘latency model’12 and is the basis of the ‘fetal origins of adult disease’ hypothesis.13 For example, evidence suggests that poor growth in utero leads to a variety of chronic disorders such as cardiovascular disease, non-insulin dependent diabetes, and hypertension. Exposures acting in later life may still influence disease risk in a simple additive way but it is argued that fetal exposures permanently alter anatomical structures and a variety of metabolic systems.14 In its purest form, this model advocates that an exposure in a critical period results in permanent and irreversible damage or disease. A stark example of such an exposure would be maternal exposure to thalidomide in pregnancy and limb development. However, within the context of chronic disease, it is important to distinguish the effects of exposure on structure from those on function. Poor intrauterine development may have an adverse effect on the number of muscle cells that develop in the fetus so that at birth a small growth retarded fetus may have a permanent reduction in the number of muscle cells (structure). Such a child, however, may still compensate by muscle hypertrophy so that in functional terms there may be no evident difference. Such an adaptive facility would make evolutionary sense and may only be unmasked at older ages when the underlying structural deficits may become more important as adaptive processes begin to fail.14 The same is likely to apply to metabolic and hormonal systems, which may be up or down regulated during fetal life but are still capable of modification by many adult exposures. The second model extends the ‘critical period’ concept as it recognizes the importance of later life effect modifiers. For example, studies have shown that the relationships of coronary heart disease, high blood pressure and insulin resistance with low birthweight are particularly strong or sometimes only observed for subjects who become obese in childhood, adulthood or both.11,15–17 Such subtle differentiation is important as the critical period may again only be critical for those individuals who experience some other exposure. Whilst statistical interaction is not the same as biological interaction, it is plausible to envisage that a biologically compromised system may only result in pathology with the subsequent addition of other physiological or metabolic stressors. In contrast, factors that raise disease risk or promote good health may accumulate gradually over the life course, although there may be developmental periods when their effects have greater impact on later health than factors operating at other times (see below). This idea is complementary to the notion of allostatic load18 so that as the number and/or duration of exposures increase, there is increasing cumulative damage to biological systems. Environmental or behavioural insults may cause long-term, gradual damage to health in separate and independent ways, or they may cluster together in socially patterned ways. For example, a subject may experience a variety of independent exposures (such as a road traffic accident, subsequent unemployment and finally the death of a spouse), where each event was unrelated to the proceeding one. However, it is far more common for adverse exposures to be clustered. For example, children living in adverse social circumstances are more likely to be of low birthweight, be exposed to poor diets, experience passive smoke exposure, and have worse educational opportunities. In this case, understanding the effects of childhood social class by identifying specific aspects of the early physical or psychosocial environment (such exposure to air pollution or family conflict) or possible mechanisms (such as nutrition, infection or stress) that are associated with adult disease will provide further aetiological insights. Risk factors at different life stages may also accumulate over time because of ‘chains of risk’ where one adverse (beneficial) exposure or experience tends to lead to another, and so on. For example, unemployment will lead to financial insecurity, which in turn will increase the likelihood of marital conflict and possibly physical abuse leading to marital separation and divorce. These links are probabilistic rather than deterministic but are likely to be sequential. In this scenario it is possible to conceive that each exposure increases the risk of separation in a simply cumulative fashion (‘additive effect‘). Alternatively, it may be only the final link in the chain, physical abuse, which results in separation (‘trigger effect‘). From a preventive point of view, these chains of events help to identify points of intervention where chains of risk may be broken and a new life course trajectory established. Stopping the additive effect will have health benefits but residual damage will remain. Preventing a trigger effect will more dramatically abolish any adverse risk associated with exposures experienced earlier in the chain. The terms critical and sensitive periods are often used loosely in epidemiology without much distinction. In the natural sciences a critical period of development refers to a time window when change in the organization of living systems or subsystems towards increasing complexity, greater adaptivity and more efficient functioning is occurring rapidly and may be most easily modifiable by a variety of factors in a favourable or unfavourable direction.19 The intrinsic changes that occur in critical developmental periods are wholly or partially irreversible; sensitive developmental periods are also times of rapid change but there is more scope to modify or even reverse those changes outside the time window. We suggest the following distinctions for the use of these terms in life course chronic disease epidemiology. In this context, the relevance of changes during a critical period is in respect of their long-term effects on intermediate markers of disease risk or manifestation of disease, usually many years after the critical period. A critical period is defined as a limited time window in which an exposure can have adverse or protective effects on development and subsequent disease outcome. Outside this window, this developmental mechanism for mediating exposure and disease risk is no longer available. A sensitive period is a time period when an exposure has a stronger effect on development and hence disease risk than it would at other times; in other words the same exposure outside this time period may still be associated with increased risk but this association is weaker than during the sensitive period. In epidemiological terms both critical and sensitive periods may be understood as qualitatively different exposure-time interactions. For critical periods, there is no excess disease risk associated with exposure outside this time window, whilst for sensitive periods it is merely weaker. Whilst critical period exposures appear obvious if they act during fetal development, they are not limited in this way. For example, the elevated risk of multiple sclerosis amongst European migrants to South Africa is only observed if migration occurred after the age of 15 years.20 Sensitive period effects for chronic diseases are far harder to demonstrate empirically. Hall et al. discuss how clinical disease severity seen with infectious agents varies with age at exposure.3 Critical periods may be more evident for chronic disease risk associated with developmental mechanisms in biological subsystems whereas sensitive periods are likely to be more common in behavioural development. Most of us are familiar with the ease by which children learn a second language and the difficulties encountered in adulthood. Most existing life course studies have limited their scope in examining exposures within a single cohort. Whilst studies in the past have examined disease and/or risk factor concordance rates or correlations across generations, it is only more recently that investigators have started to test inter-generational exposure disease associations. Several studies have now shown that both maternal and paternal cardiovascular mortality are associated with offspring birthweight.21–23 Such associations are complex and may reflect long-term adverse socioeconomic circumstances, maternal health, determined by mother's own birthweight and childhood growth, lifestyle factors such as smoking, and genomic or epigenetic processes.22 Recently the potential for a life course approach to aid understanding of variations in the health and disease of populations over time, across countries and between social groups has been given more attention.24–27 Davey Smith and his colleagues25 suggest that explanations for social inequalities in cause-specific adult mortality lie in socially patterned exposures at different stages of the life course. A life course perspective is being increasingly used in developing the ‘social inequalities in health’ debate.28–31 More broadly, Keating and Hertzman26 argue that the fundamental processes such as neural sculpting that affect brain and behavioural development interact with the growing chaos in the lives of children and adolescents with long-term effects on human capital and hence the wealth of nations. Leon24 suggests that a life course perspective may help understand the underlying geographical patterns of mortality, particularly East-West differences. He states ‘the assumption that inequalities in health today, whether between or within countries are caused by contemporaneous differences in circumstances of life is not sustainable for a range of important diseases that appear to be driven instead by poor socioeconomic circumstances in early life and childhood’.24 The end of the 20th century has seen much debate around the need for epidemiological theory, particularly in relation to health inequalities.32,33 Alongside the development of life course epidemiology with its focus on time, has been a growing interest in eco-social or models that risk factors operating at a variety of from the to the life course they the need to have an approach between biological and social the of social into the they also the temporal relationship between exposures. The between life course and eco-social has been by in a life course approach the of psychosocial Whilst a life course approach psychosocial it is far than and to life course with and measures of into a We have previously that a life course approach to eco-social influences on In this respect we that both models may be increasingly used We not believe that life course a conceptual approach but rather that both temporal and are complementary and We would also argue that a life course approach is not limited to individuals within a single but biological and social of risk across It any exposure both within a structure as well as in relation to geographical and which may be unique to that cohort of We have the by and to these points (Figure and children are across both by common and/or social The potential of and influences are acting across time and across For example, adverse affect a and exposures time may be specific to a single cohort or period effects may be experienced by individuals The idea that childhood is important for adult health is not new in epidemiology or health but was the model of health in the of the 20th The idea that of risk to adult health in early life was given during the years following the The childhood origins of adult chronic and data from the origins of adult was a natural of epidemiology where cohort and long-term effects of early or exposures in adult and birth been to childhood The of the effects of the during the of on subsequent human development, particularly a life course and the for life course epidemiology when they that ‘the of each cohort at from the of at the time, and with the of and experience. The adult the of these favourable and unfavourable during has and mortality has their in a way specific to each in his of Epidemiology also the importance of inter-generational and childhood of adult disease (see by Davey In cardiovascular epidemiology, the adult lifestyle model the some but interest in early life factors was from the with natural history studies of adult risk factors (such as smoking, and in of This was then by that poor childhood living (see accompanying by and accompanying in this and impaired early to adult cardiovascular disease. The most model of human development has been the idea that ‘the few years of life have effects later development and adult The of development at the of the 20th century development as an of developmental where change was cumulative and usually The of the environment was to provide the and for growth and to Environmental insults or of during critical periods at the or during during childhood more have long-term effects on physical or development. was systems to be From the this early life was increasingly It was that of the rather than early evidence of long-term effects on adult of a critical period in behavioural development social or language was by a of some behavioural from the effects of critical period exposure or for irreversible biological changes during critical periods of growth was as a to this a life perspective in developmental in the where development was seen as a lifelong The of early developmental be again and again by later and the course of development into In the argued that ‘simplistic of effects need to be and by more notions of the over time but between early and adult psychosocial we have ‘chains of provide an for these whereby one leads to another, a good experience it more likely that one will be and have how the individual's life course is by and and social and changes in processes (such as the increase in life the cohort effects occur because of different ages and those who different are differentially exposed to and by social and such as the of the The life course perspective in has life to how the life course is in the and in
Stored only when both papers are in the corpus — 3,026 citations exist upstream.