GLOSSARY
Life course epidemiology
D Kuh, Y Ben-Shlomo, J Lynch, J Hallqvist, C Power
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
J Epidemiol Community Health 2003;57:778–783
The aim of this glossary is to encourage a dialogue that will
advance the life course perspective.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
See end of article for
authors’ affiliations
. . . . . . . . . . . . . . . . . . . . . . .
Correspondence to:
Dr D Kuh, Medical
Research Council National
Survey of Health and
Development, Department
of Epidemiology and
Public Health, University
College London and Royal
Free Medical School,
Gower Street Campus,
1–19 Torrington Place,
London WC1E 6BT, UK;
d.kuh@ucl.ac.uk
. . . . . . . . . . . . . . . . . . . . . . .
A
life course approach offers an interdisciplinary
framework for guiding research on
health, human development and aging.
Psychologists,1
sociologists,2 3
demographers,4
anthropologists,5
and biologists6
have actively
promoted such an approach for many years. The
interdisciplinary research area of developmental
science,7 8
also brings together psychological,
cognitive, and biological research on developmental
processes from conception to death.
Epidemiologists have been more recent converts
to a life course approach.9–15
We have previously defined life course epidemiology
as the study of long term effects on later
health or disease risk of physical or social
exposures during gestation, childhood, adolescence,
young adulthood and later adult life.9 15 16
The aim is to elucidate biological, behavioural,
and psychosocial processes that operate across an
individual’s life course, or across generations, to
influence the development of disease risk. The
catalyst for a life course approach in epidemiology
stemmed from the revival of interest in the
role of early life factors in cardiovascular and
other chronic diseases,17
in particular the ecological
and historical cohort studies used to explore
the fetal origins hypothesis.18
According to this
hypothesis, environmental exposures such as
undernutrition during critical periods of growth
and development in utero may have long term
effects on adult chronic disease risk by ‘‘programming’’
the structure or function of organs,
tissues, or body systems. This idea of ‘‘biological
programming’’ was presented as an alternative
paradigm to the adult lifestyle model of adult
chronic disease that focuses on how adult
behaviours (notably smoking, diet, exercise and
alcohol consumption) affect the onset and
progression of diseases in adulthood.19 20
To counteract the increasing polarisation of
biological programming in utero and adult lifestyle
approaches to chronic disease aetiology, life
course epidemiology was built on the premise
that various biological and social factors
throughout life independently, cumulatively
and interactively influence health and disease
in adult life.9
A life course approach does not
deny the importance of conventional risk factors,
such as smoking and hypertension, which were
so successfully identified by the early post-war
adult cohort studies. Rather its purpose is to
study the contribution of early life factors jointly
with these later life factors to identify risk and
protective processes across the life course. So far,
life course epidemiology has paid particular
attention to the long term effects of childhood
and adolescent risk factors on later disease. This
is partly a response to the emphasis on adult
factors in most post-war aetiological models of
chronic disease. This is in contrast with the focus
of life span developmental psychology on adult
human development to counter the dominance
of child centred developmental psychology.1
Life course epidemiology attempts to integrate
biological and social risk processes rather than
draw false dichotomies between them. The
interests of life course epidemiology overlap with
social epidemiology, that branch of epidemiology
that studies the role of social factors in the
production of health and disease in popula-
tions.21
Life course epidemiology studies how
socially patterned exposures during childhood,
adolescence, and early adult life influence adult
disease risk and socioeconomic position, and
hence may account for social inequalities in adult
health and mortality. Socioeconomic factors at
different life stages may operate either via social
chains of risk or by influencing exposures to causal
factors at earlier life stages that form part of long
term biological or psychological chains of risk.
Differential health effects according to the
timing or duration of exposure to socioeconomic
circumstances may provide important clues to
aetiology.22–24
Life course epidemiology also provides
a fresh perspective on explanations for
secular disease trends10
and for gender,25
ethnic,26
and geographical12
inequalities in health.
Life course epidemiology implies more than a
longitudinal study. The first is a theoretical
model whereas the second is a study design.
The purpose of life course epidemiology is to
build and test theoretical models that postulate
pathways linking exposures across the life course
to later life health outcomes.16
These models
explicitly require the temporal ordering of
exposures and their inter-relationships. The
development of these models in life course
epidemiology provides a persuasive rationale for
time related study designs. It has led to a
growing interest in existing and new longitudinal
studies that capture certain time windows
or other potentially significant features of
the life course.27 28
They include new birth
cohorts29 30
and the revitalisation of old historical
cohorts.18 28 31
Some guidance on the practicalities
of conducting life course studies exists32–34
and more is planned.35
The implementation of life course models
present many methodological challenges beyond
study design16 24 36
that are also being addressed.
They include the analytical problems associated
with modelling repeat observations, hierarchical
data, latent exposures, or multiple interactive or
778
www.jech.com
onApril21,2021byguest.Protectedbycopyright.http://jech.bmj.com/JEpidemiolCommunityHealth:firstpublishedas10.1136/jech.57.10.778on22October2003.DownloadedfromonApril21,2021byguest.Protectedbycopyright.http://jech.bmj.com/JEpidemiolCommunityHealth:firstpublishedas10.1136/jech.57.10.778on22October2003.DownloadedfromonApril21,2021byguest.Protectedbycopyright.http://jech.bmj.com/JEpidemiolCommunityHealth:firstpublishedas10.1136/jech.57.10.778on22October2003.Downloadedfrom
small effects. These problems are common to epidemiology
and other related disciplines, but are particularly relevant for
testing life course models. Multilevel models,37
latent growth
models,38
Markov and graphical chain models39
are among
the techniques being used by epidemiologists for this
purpose. The problems of missing data, omitted exposures,
and measurement error may be particularly salient in
longitudinal studies where observations have taken place
over a prolonged time period.
Life course epidemiology borrows many of its concepts and
ideas from other scientific disciplines. Some of these ideas are
defined in this glossary for use by epidemiologists. As life
course epidemiology is a dynamic new field, terms are often
used loosely or inconsistently. Our intention is not to impose
definitions but rather to encourage a dialogue among
epidemiologists that will advance the life course perspective.
Clarification of terms will aid the development of life course
theoretical models and their operationalisation in terms of
testable hypotheses, analytical strategies, and the use of
appropriate statistical techniques.
There is a structure in the glossary that is hidden by the
alphabetical ordering of the concepts. Most concepts fall into
one of three categories. Firstly, there are the concepts
referring to the causal pathway in relation to time (see
accumulation, chain of risk, trajectory). Secondly, there are
concepts about the timing of causal actions (see birth
cohorts, critical and sensitive periods, induction and latency
periods). Thirdly, there are concepts referring to different
types of mechanisms (see embodiment, mediating and modifying
factors, resilience, susceptibility and vulnerability).
ACCUMULATION OF RISK
One of the earliest descriptions of accumulation of risk was
Riley’s concept of insult accumulation,40
the notion that life
course exposures or insults gradually accumulate through
episodes of illness and injury, adverse environmental conditions,
and health damaging behaviours. One of the purposes
of life course epidemiology is to test the extent of cumulative
damage to biological systems as the number, duration or
severity of exposures increase, and as body systems age and
become less able to repair damage. The accumulation of
different types of exposures (such as environmental, socioeconomic,
and behavioural) may cause long term damage41
with exposure risk being either independent (fig 1, model
(a)) or clustered (fig 1, model (b)). The latter is known as an
accumulation model with risk clustering.16
Life course epidemiology
shares with social epidemiology a particular interest in
exposures that cluster because they are often related to an
individual’s or family’s socioeconomic position in society.
BIRTH COHORT EFFECTS
Birth cohort refers to the location of an individual in historical
time as indexed by their year of birth. An environmental
change (such as an improvement or deterioration of living
standards) that affects the health of children may show up
several decades later as birth cohort differences in adult
mortality.42
Cohort differences in adult disease can also occur
if there are cohort variations in risk factors such as
childbearing characteristics or the adoption of habits such
as smoking43
that have long term effects on health. Birth
cohorts may also be differentially affected by rapid or
extensive social change.44
The impact of such change, creating
possibilities for turning points on life trajectories with short and
long term effects on health, often varies by age.2
The
changing size of birth cohorts (such as the post-war ‘‘baby
boom’’ generation) is itself one of the forces for social
change. Age-cohort models identify possible cohort effects.
More complex age-period-cohort models can be fitted to help
distinguish these effects but require assumptions in order to
estimate the parameters of the model because of the linear
dependency among the three factors.45 46
Cohort effects are
easiest to distinguish when disease trends have accelerated,
decelerated, or changed direction. Where they are steady and
linear they cannot be reliably distinguished from period
effects.43
CHAINS OF RISK MODEL
A chain of risk model refers to a sequence of linked exposures
that raise disease risk because one bad experience or
exposure tends to lead to another and then another.
Different types of chains can confer increased or decreased
risk. Our definition of chains of risk (or protective chains) is
based on the idea of ‘‘chain reactions’’ put forward by
Rutter47
to explain continuities between early experiences
and adult psychosocial function. It also resembles what has
been described as a ‘‘pathways model’’.13 48
Social, biological,
and psychological chains of risk are possible and involve
‘‘mediating factors’’ and often ‘‘modifying factors’’. The sequential
links are probabilistic rather than deterministic. It is
possible to conceive of two different types of chains of risk.
One model is that each exposure not only increases the risk of
the subsequent exposure but also has an independent effect
on disease risk irrespective of the later exposure (fig 1, model
c). When each adverse experience increases the risk of
disease in a cumulative fashion this is called an ‘‘additive
effect’’ and may be a special case of an accumulation model.
Figure 1 Life course causal models.
Life course epidemiology 779
www.jech.com
onApril21,2021byguest.Protectedbycopyright.http://jech.bmj.com/JEpidemiolCommunityHealth:firstpublishedas10.1136/jech.57.10.778on22October2003.Downloadedfrom
Alternatively, earlier exposures have no effect on disease risk
without the final link in the chain that precipitates disease
onset (fig 1, model (d)). Such a ‘‘trigger effect’’ describes the
situation when it is only the final link in the chain that has a
marked effect on disease risk.
CONTEXT
Context refers to the location of an individual by time and
place. Place refers to both geographical location and to group
membership in terms of family, friends or age, and on the
basis of class, ethnicity, residence, and gender that arise out
of the social and economic structure of society. Context may
affect exposure to risk and the individual’s response
strategies. Application of a life course perspective to
contextual as well as individual effects on health implies
understanding the effects of changing contexts over time on
the life course of individuals.
CRITICAL PERIOD
In the natural sciences a critical period of development refers to
a time window when intrinsic changes in the organisation of
living systems or sub-systems towards increasing complexity,
greater adaptivity and more efficient functioning occurs
rapidly and may be most easily modified in a favourable or
unfavourable direction.49
In life course epidemiology the
relevance of changes during a critical period is in respect of
their long term effects on disease risk many years later. Thus,
we define a critical period as a limited time window in which
an exposure can have adverse or protective effects on
development and subsequent disease outcome. Outside this
developmental window there is no excess disease risk
associated with exposure.
CRITICAL PERIOD MODEL
A critical period model of disease causation is when an exposure
acting during a critical period of development has effects on
the structure or function of organs, tissues or body systems
that are not modified in any dramatic way by later
experience, and that precipitate disease later in life. Also
known as ‘‘biological programming’’ or as a ‘‘latency
model’’,13
it underlies the fetal origins of adult disease
hypothesis. An expanded version of the basic critical period
model includes the possibility that the effect of an exposure
during a critical period of development on later disease risk
may be dramatically changed by later physiological or
psychological stressors. The critical period with later effect
modifiers implies investigating plausible interactive effects
between early and later life risk factors.
EMBODIMENT
Embodiment is a term common to both social epidemiology21
and life course epidemiology. It describes how extrinsic
factors experienced at different life stages are inscribed into
an individual’s body functions or structures. This may be
through developmental processes associated with critical
periods, habituation, learning, damage, and repair. The term
‘‘biological embedding’’48 50
has a similar meaning although it
tends to be applied to neurobiological or psychobiological
mediators of the early social environment on child development
and life course health.
INDUCTION AND LATENCY PERIODS
While different definitions of these terms appear in the
epidemiological literature, we favour those used by
Rothman.51
Induction period is defined as the time between
exposure and initiation of the disease process, hence it is a
characteristic of a relation between a specific exposure and a
specific disease. Life course epidemiologists have a particular
interest in causal relations with long induction periods.
Latency period refers to the period between disease initiation
and detection, and is a characteristic of the disease (onset of
symptoms) or the healthcare system (diagnosing the
disease). If an exposure has a long induction and/or latent
period, its public health importance may vary according to
when in the life course one is exposed. Variations in the time
between induction and latency periods make it difficult to
detect the true causal agents, and the strength of association
between an exposure and disease can be diluted or missed if
the wrong time frame is measured. Often it is difficult to
observe the time of disease initiation and this prevents
researchers from distinguishing empirically between the
induction and latent periods. Latency period should not be
confused with latency model, which is an alternative term for
a critical period model.
LIFE CYCLE/LIFE SPAN/LIFE COURSE
The concept of a life cycle has generally been used in other
scientific disciplines to describe a series of distinct, bounded
life stages which are socially or biologically determined.52
In
contrast the concept of the life span used in psychology
assumes that development and aging form a continuous
process from birth to death. The distinction between life span
and life course, the term used more commonly in sociology, is
mainly a matter of scientific history.
MEDIATING FACTOR
A risk or protective factor mediates the association between
exposure and disease when it chronologically follows the
exposure and is conceptualised as lying, at least partly, on the
causal pathway. Life course epidemiology requires an explicit
temporal theoretical model that distinguishes between
mediating factors (post-exposure of interest) and confounding
factors (conceptualised as prior to and/or tangential to
understanding the effects of the exposure of interest).
MODIFYING FACTOR
A risk or protective factor modifies the association between an
exposure and disease when the causal effect of the exposure
of interest differs across levels of the modifying factor.
Investigation of modifying factors provides information
about the nature of the causal process. Effect modification
is known as interaction—‘‘synergism’’ if the modifying
variable enhances the effect of the explanatory variable or
‘‘antagonism’’ if it diminishes it. Interactions are thought to
be common features of life course processes and should be
investigated where plausible biological, behavioural or social
hypotheses exist. Well defined theoretical life course models
are needed to interpret the biological or social significance of
observed interactions. Empirical assessment of interaction is
the source of much debate in epidemiology and needs to be
carefully considered in terms of additive or multiplicative risk
models.51 53
PLASTICITY
Plasticity is the potential for change in intrinsic characteristics
in response to environmental stimuli. It is measured by
within person variability. Like life course perspectives in
other disciplines,54
the search for the range of plasticity and
age associated changes and constraints is a fundamental
quest of life course epidemiology.
RESILIENCE
Resilience is a dynamic process of positive adaptation in the
face of adversity.55
The focus of research has been on the
intrinsic and extrinsic factors associated with educational,
emotional and behavioural resilience of children.55 56
There
has been less focus on health outcomes and on long term
outcomes in general.57
A question that has hardly been
780 Kuh, Ben-Shlomo, Lynch, et al
www.jech.com
onApril21,2021byguest.Protectedbycopyright.http://jech.bmj.com/JEpidemiolCommunityHealth:firstpublishedas10.1136/jech.57.10.778on22October2003.Downloadedfrom
addressed is whether resilience in development, and its
associated factors, have long term health benefits or costs.
For example, the fetal origins hypothesis assumes that
adaptations for survival made by the fetus in response to
adversity (maternal undernutrition) raise the risk of chronic
disease risk in later life.18
Life course epidemiology has
contributions to make to an integrated science of human
adaptation and development56
that also includes an aging
perspective. Both share a focus on lifelong biological,
psychological, and social processes and their interrelation-
ships.
SENSITIVE PERIOD
Like critical periods, sensitive periods are also times of rapid
individual change but there is more scope to modify or even
reverse those changes outside the time window. Thus a
sensitive period in life course epidemiology is a time period
when an exposure has a stronger effect on development and
subsequent disease risk than it would at other times. Outside
the time period any excess risk will be weaker.
SUSCEPTIBILITY
Within epidemiology susceptibility has been defined in terms
of Rothman’s ‘‘causal pie model’’,51
which describes susceptibility
as ‘‘a term that refers to the condition of having one of
two interacting causes already and therefore being susceptible
to the effect of the other.’’ (page 172). It is usually used
where host factors increase the likelihood that another
exposure will produce disease. Host factors refer to both
intrinsic factors as well as prior exposures that have become
embodied over time. Thus susceptibility is a process occurring
over time that may eventually lead to disease via the
completion of the last piece of the causal pie. This idea has
been given greater salience with the emergence of genetic
epidemiology and the growing attention to gene-environment
interactions.58
In the wider scientific literature susceptibility
is often used synonomously with vulnerability and
conveys a broadly similar process.
TIME
Time is a fundamental concept in life course epidemiology,
both in terms of life time (as indexed by the chronological
age of individuals) and historical time at the population level
(as indexed by membership of a birth cohort). Changing
individuals must be studied in a changing environment (see
context). The effect of an exposure on a health outcome may
be dependent on the duration or timing of exposure. Critical
and sensitive periods are so defined because the age of the
individual at the time of exposure is assumed to affect the
long term outcome. In other words, they represent qualitatively
different exposure-time interactions. For critical
periods there is no excess risk associated with exposure
outside this time window whereas for sensitive periods it is
merely weaker. This is similar to Elder’s ‘‘life stage
principle’’.2
Similarly, social roles and life events have
different social meanings and possibly differential health
effects depending on their timing and whether they are
typical for a person’s age, gender, or culture.
TRAJECTORY, TRANSITION AND TURNING POINTS
(ADAPTED FROM ELDER2
)
A trajectory provides a long term view of one dimension of an
individual’s life over time. These may be social states (such as
work, marriage, socioeconomic position), psychological states
(such as depression) or physiological states (such as lung
function). Implicit is the idea of a normative trajectory
around which individuals deviate. Transitions are short-term
and embedded in trajectories, marking a change in social,
psychological, or physiological states. A marked change of
direction is referred to as a turning point.
VULNERABILITY
Vulnerability is the opposite of resilience and refers to a
dynamic process of negative adaptation in the face of
adversity. The negative physiological or behavioural response
is shaped by prior embodiment of extrinsic factors as well as
intrinsic characteristics.
CONCLUSION
Life course epidemiology has blossomed in the past five years
as a natural response to the limitations of previous
aetiological models of chronic disease. We have welcomed
the opportunity here and elsewhere16
to clarify its purpose
and guiding principles as misunderstandings have inevitably
arisen. A life course perspective in epidemiology has been
erroneously limited to notions of risk accumulation,59 60
or
associated only with psychosocial theories of social epide-
miology.21
Rather we would argue that it represents an
attempted synthesis of previous models of disease causation
(such as adult lifestyle, biological programming, and social
causation) that is scientifically interesting, conceptually and
methodologically challenging, inherently interdisciplinary
and policy relevant. A life course approach has been adopted
in all but name by researchers whose focus was initially
restricted to fetal and infant life but who have now extended
their frame of reference to include postnatal growth and
development.59 61
Childhood indicators of adult disease risk
include childhood stature, particularly leg length,62
relative
weight,63
energy intake and IQ.64 64a
There is also more
willingness among those who focus on adult risk factors to
recognise the importance of early life factors.65
In epidemiology
there is a growing appreciation of the need for a temporal
perspective for conventional so called ‘‘adult’’ risk factors.66–68
Age at onset, duration and changes in levels of exposure to
conventional risk factors may alter their effects on adult
disease risk and impact on long term disease trends.69
What are the future challenges for life course epidemiology?
One challenge is the continued development of testable
theoretical models that elucidate the risk and protective
factors at each life stage, and the underlying biological,
psychological, and social pathways that link these together
across one or more generations. The life course models
defined in this glossary are not mutually exclusive and may
operate simultaneously. It may not be easy to distinguish
these models empirically24
nor to develop standardised and
acceptable methods of combining cumulative exposures.36
These are just two of a number of complex methodological
problems facing life course epidemiology. The second
challenge is the ability of life course epidemiology to
help explain temporal, geographical, and social patterns
of disease distribution. Thirdly, changing individuals need
to be studied in a changing world and we would like to see
a life course approach fully integrated into the study of
broadly defined contextual health effects. Fourthly, the
growing genetic knowledge and development of cheap and
reliable methods to analyse DNA on large population
samples, has opened up possibilities for studying the
interactive effects of genetic and environmental factors
across the life course. Lastly, and most important of all,
there is the need to translate our new knowledge into
interventions and policy recommendations designed to
improve the long term health of individuals, social groups,
and societies.14 70–72
ACKNOWLEDGEMENTS
The authors would like to acknowledge the European Science
Foundation Scientific Programme on Social Variations in Health
Life course epidemiology 781
www.jech.com
onApril21,2021byguest.Protectedbycopyright.http://jech.bmj.com/JEpidemiolCommunityHealth:firstpublishedas10.1136/jech.57.10.778on22October2003.Downloadedfrom
Expectancy in Europe. Many of the ideas in this paper were discussed
with colleagues in Working Group 1 (Life course influences on health
inequalities).
Authors’ affiliations
. . . . . . . . . . . . . . . . . . . . .
D Kuh, Medical Research Council National Survey of Health and
Development, Department of Epidemiology and Public Health, University
College London and Royal Free Medical School, UK
Y Ben-Shlomo, Department of Social Medicine, University of Bristol, UK
J Lynch, Department of Epidemiology, School of Public Health, University
of Michigan, USA
J Hallqvist, Department of Public Health Sciences, Karolinska Institutet,
Stockholm, Sweden
C Power, Institute of Child Health, Centre for Paediatric Epidemiology
and Biostatistics, London, UK
REFERENCES
1 Baltes PB, Lindenberger U, Staudinger UM. Life-span theory in developmental
psychology. In: Damon W, Lerner RM, eds. Handbook of child psychology
volume 1: theoretical models of human development. New York: Wiley,
1998:1029–143.
2 Elder GH Jr. The life course and human development. In: Damon W,
Lerner RM, eds. Handbook of child psychology volume 1:
theoretical models of human development. New York: Wiley,
1998:939–91.
3 Giele JZ, Elder GH Jr. Life course research: development of a field. In: Giele JZ,
Elder GH Jr, eds. Methods of life course research: qualitiative and quantitivae
approaches. Thousand Oaks, CA: Sage, 1998:5–27.
4 Uhlenberg P. Mutual attraction: demography and life-course analysis.
Gerontologist 1996;36:226–9.
5 Panter-Brick C, Worthman CM, eds. Hormones, health and behavior.
Cambridge: Cambridge University Press, 1999.
6 Henry CJK, Ulijaszek, eds. Long-term consequences of early environment:
growth, development and the lifespan perspective. Oxford: Oxford University
Press, 1996.
7 Magnusson D, ed. The lifespan development of individuals: behavioral,
neurobiological and psychosocial perspectives. Cambridge: Cambridge
University Press, 1996.
8 Cairns RB, Elder GH, Costello EJ, eds. Developmental science. Cambridge:
Cambridge University Press, 1996.
9 Kuh D, Ben-Shlomo Y, eds. A life course approach to chronic disease
epidemiology; tracing the origins of ill-health from early to adult life. 2nd edn.
Oxford: Oxford University Press (in press).
10 Davey Smith G, Gunnell D, Ben-Shlomo Y. Life-course approaches to socioeconomic
differentials in cause specific adult mortality. In: Leon D, Walt G,
eds. Poverty, inequality and health: an international perspective. Oxford:
Oxford University Press, 2000:88–124.
11 Davey Smith G, Ben-Shlomo Y, Lynch J. Life course approaches to inequalities
in coronary heart disease. In: Stansfeld SA, Marmot MG, eds. Stress and the
heart. London: BMJ Books, 2002:20–49.
12 Leon DA. Common threads: underlying components of inequalities in mortality
between and within countries. In: Leon D, Walt G, eds. Poverty, inequality and
health: an international perspective. Oxford: Oxford University Press,
2001:58–87.
13 Hertzman C, Power C, Matthews S, et al. Using an interactive framework of
society and lifecourse to explain self-rated health in early adulthood. Soc Sci
Med 2001;53:1575–85.
14 Graham H. Building an inter-disciplinary science of health inequalities: the
example of lifecourse research. Soc Sci Med 2002;55:2005–16.
15 Kuh D, Hardy R, eds. A life course approach to women’s health. Oxford:
Oxford University Press, 2002.
16 Ben-Shlomo Y, Kuh D. A life course approach to chronic disease
epidemiology: conceptual models, empirical challenges, and interdisciplinary
perspectives. Int J Epidemiol 2002;31:285–93.
17 Kuh D, Davey Smith G. When is mortality risk determined? Historical insights
into a current debate. Soc Hist Med 1993;6:101–23.
18 Barker DJP. Mothers, babies and health in later life. Edinburgh: Churchill
Livingstone, 1998.
19 Barker DJP, ed. Fetal and infant origins of adult disease. London: British
Medical Publishing Group, 1992.
20 Robinson RJ. Is the child father of the man? Controversy about the early
origins of cardiovascular disease. BMJ 1992;304:789–90.
21 Krieger N. A glossary for social epidemiology. J Epidemiol Community Health
2001;55:693–700.
22 Davey Smith G, Hart C, Blane D, et al. Adverse socioeconomic conditions in
childhood and cause specific adult mortality: prospective observational study.
BMJ 1998;316:1631–5.
23 Power C, Manor O, Matthews S. The duration and timing of exposure: effects
of socioeconomic environment on adult health. Am J Public Health
1999;89:1059–65.
24 Hallqvist J, Lynch J, Bartley M, et al. Can we disentangle life course processes
of accumulation, critical period and social mobility? An analysis of
disadvantaged socio-economic positions and myocardial infarction in the
Stockholm Heart Epidemiology Program (SHEEP). Soc Sci Med (in press).
25 Maughan B. Depression and psychological distress: a life course perspective.
In: Kuh D, Hardy R, eds. A life course approach to women’s health. Oxford:
Oxford University Press, 2002:161–76.
26 Hertzman C. The life course contribution to ethnic disparities in health.
Washington, DC: National Academy of Sciences, US (in press).
27 Eaton WW. The logic for a conception-to-death cohort study. Ann Epidemiol
2002;12:445–51.
28 Susser E, Terry MB, Matte T. The birth cohorts grow up: new opportunities for
epidemiology. Paediat Perinat Epidemiol 2000;14:98–100.
29 Golding J, Pembrey M, Jones R, et al. ALSPAC- The Avon Longitudinal Study
of Parents and Children. 1. Study methodology. Paediat Perinat Epidemiol
2001;15:74–87.
30 Smith K, Joshi H. The millennium cohort study. Pop Trends 2002;107:30–4.
31 Leon DA, Lithell HO, Vagero D, et al. Reduced fetal growth rate and increased
risk of ischaemic heart disease mortality in 15 thousand Swedish men and
women born 1915–29. BMJ 1998;317:241–5.
32 Farringdon DP, Gallagher B, Morley M, et al. Minimising attrition in
longitudinal research. Methods of tracing and securing cooperation in a 24year
follow-up study. In: Magnusson D, Bergmann LR, eds. Data quality in
longitudinal research. Cambridge: Cambridge University Press,
1990:122–47.
33 Wadsworth MEJ, Butterworth SB, Hardy RJ, et al. The life course prospective
design: an example of benefits and costs associated with study longevity. Soc
Sci Med (in press).
34 Stouthamer-Loeber M, van Kammen WB. Data collection and
management: a practical guide. Thousand Oaks: Sage, 1995.
35 Pickles A, Maughan B, Wadsworth MEJ. Methods, theory and analysis in life
course research. Oxford, Oxford University Press (in press).
36 Kuh D, Hardy R. Conclusions: linking the past, present and future. In: Kuh D,
Hardy R, eds. A life course approach to women’s health. Oxford: Oxford
University Press, 2002.
37 Goldstein H. Multilevel statistical models. London: Edward Arnold, 1995.
38 Muthen B. Latent variable mixture modeling. In: Marcoulides G,
Schumacker R, eds. New developments and techniques in structural equation
modeling. Mahwah, NJ: Lawrence Erlbaum, 2001:1–33.
39 Whittaker J. Graphical models in applied multivariate statistics. Chichester:
Wiley, 1990.
40 Riley JC. Sickness, recovery and death: a history and forecast of ill-health.
Basingstoke: Macmillan, 1989.
41 Power C, Manor O, Fox AJ. Health and class: the early years. London:
Chapman Hall, 1991.
42 Kuh D, dos Santos Silva I, Barrett-Connor E. Disease trends in women living in
established market economies: evidence of cohort effects during the
epidemiological transition. In: Kuh D, Hardy R, eds. A life course approach to
women’s health. Oxford: Oxford University Press, 2002:347–73.
43 Strachan DP, Perry IJ. Time trends. In: Kuh D, Ben-Shlomo Y, eds. A life course
approach to chronic disease epidemiology. Oxford: Oxford University Press,
1997:101–20.
44 Stein S, Quarraisha KA, Susser M. The life course of black women in South
Africa in the 1990s: generation, age, and period in the decade of HIV and
political Liberation. In: Kuh D, Hardy R, eds. A life course approach to
women’s health. Oxford: Oxford University Press, 2002:374–96.
45 Holford TR. Understanding the effects of age, period, and cohort on incidence
and mortality rates. Annu Rev Publ Health 1991;12:425–57.
46 Robertson C, Gandini S, Boyle P. Age-period-cohort models: a comparative
study of available methodologies. J Clin Epidemiol 1999;52:569–83.
47 Rutter M. Pathways from childhood to adult life. J Child Psychol Psychiat
1989;1:23–51.
48 Keating D, Hertzman C, eds. Developmental health and the wealth of nations:
social, biological and educational dynamics. New York: The Guilford Press,
1999.
49 Scott JP. Critical periods in organizational processes. In: Falkner F, Tanner JM,
eds. Human growth: a comprehensive treatise. Vol 3. Methodology;
ecological, genetic and nutritional effects on growth. New York: Plenum Press,
1986:181–96.
50 Hertzman C. The biological embedding of early experience and its effects on
health in adulthood. Ann NY Acad Sci 1995;896:85–95.
51 Rothman KJ. Epidemiology. An introduction. Oxford: Oxford University Press,
2002.
52 Leidy LE. Lifespan approach to the study of human biology: an introductory
overview. Am J Hum Biol 1996;8:699–702.
53 Hallqvist J, Ahlbom A, Diderichsen F, et al. How to evaluate interaction
between causes: a review of practices in cardiovascular epidemiology. J Intern
Med 1996;239:377–82.
54 Sherrod LR, Brim OG Jr. Epilogue: retrospective and prospective views of lifecourse
research on human development. In: Sorensen AB, Weinert FE,
Sherrod LR, eds. Human development and the life course: multidisciplinary
perspectives. Hillsdale, NJ: Lawrence Erlbaum Associates, 1986:557–80.
55 Luthar SS, Cicchetti D, Becker B. The construct of resilience: a critical
evaluation and guidelines for future work. Child Dev 2000;71:543–62.
56 Masten AS. Ordinary magic. Am Psychologist 2001;56:227–38.
57 Ryff CD, Singer B, Love GD, et al. Resilience in adulthood and later life.
Defining features and dynamic processes. In: Lomranz J, ed. Handbook of
aging and mental heath. New York: Plenum Press, 1998:69–96.
58 Khoury MJ, Stewart W, Beatty TH. The effect of genetic suseptibility on causal
inference in epidemiologic studies. Am J Epidemiol 1987;126:561–7.
59 Eriksson JG, Forsen T, Tuomilehto J, et al. Early growth and coronary heart
disease in later life: longitudinal study. BMJ 2001;322:949–53.
60 Marmot M, Shipley M, Brunner E, et al. Relative contribution of early life and
adult socioeconomic factors to adult morbidity in the Whitehall II study.
J Epidemiol Community Health 2001;55:301–7.
782 Kuh, Ben-Shlomo, Lynch, et al
www.jech.com
onApril21,2021byguest.Protectedbycopyright.http://jech.bmj.com/JEpidemiolCommunityHealth:firstpublishedas10.1136/jech.57.10.778on22October2003.Downloadedfrom
61 Forsen T, Eriksson JG, Tuomilehto J, et al. Growth in utero and during
childhood among women who develop coronary heart disease: longitudinal
study. BMJ 1999;319:1403–7.
62 Gunnell DJ, Davey Smith G, Frankel SJ, et al. Childhood leg length and adult
mortality—follow up of the Carnegie survey of diet and growth in pre-war
Britain. J Epidemiol Community Health 1998;52:142–52.
63 Kuh D, Hardy R, Chaturvedi N, et al. Birth weight, childhood growth and
abdominal obesity in adult life. Int J Obesity 2002;26:40–7.
64 Frankel S, Gunnell DJ, Peters TJ, et al. Childhood energy intake and adult
cancer—the Boyd Orr cohort study. BMJ 1998;316:499–504.
64a Whalley LJ, Deary IJ. Longitudinal cohort study of childhood IQ and survival
up to age 76. BMJ 2001;322:1–5.
65 Marmot M. Aetiology of coronary heart disease. BMJ 2001;323:1261–2.
66 Lynch JW, Kaplan GA, Salonen JT. Why do people behave poorly? Variation
in adult health behaviours and psychosocial characteristics by stages of the
socioeconomic lifecourse. Soc Sci Med 1997;44:809–19.
67 Harper S, Lynch J, Hsu W-L, et al. Life course socioeconomic conditions and
adult psychosocial functioning. Int J Epidemiol 2002;31:395–403.
68 Schooling M, Kuh D. A life course perspective on women’s health behaviours.
In: Kuh D, Hardy R, eds. A life course approach to women’s health. Oxford:
Oxford University Press, 2002:279–303.
69 Law M, Wald N. Why heart disease mortality is low in France: the time lag
explanation. BMJ 1999;318:1471–80.
70 Hertzman C, Wiens M. Child development and long-term outcomes: a
population health perspective and summary of successful interventions. Soc
Sci Med 1996;43:1083–95.
71 Aboderin I, Kalache A, Ben-Shlomo Y, et al. Life course perspectives on
coronary heart disease, stroke and diabetes: key issues and implications for
policy and research. Geneva: WHO, 2002.
72 Halfon N, Hochstein M. Life course health development: an integrated
framework for developing health, policy, and research. Milbank Q
2002;80:433–79.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Speaker’s corner
Understanding the functional components of public health surveillance
D
eveloping a public health surveillance system is similar
to developing a housing estate. Besides the actual
building process (setting up infrastructure for data
collection, analysis, interpretation and dissemination) there
are many other requirements to be identified in planning—
and care must be taken to ensure products will stand up to
the test of time.
The first step is to identify ‘‘wants’’ of homebuyers
(recognise and integrate all the ‘‘needs’’ of various stakeholders
in a manner that will lead to quality, efficient
outcomes). Consultation with government agencies, town
planners, and homeowners (relevant authorities and community
groups) will ensure planning processes are appropriate
to the ‘‘authorising environment’’.
It is then necessary to enlist architects (epidemiologists
and public health professionals) who will conceptualise the
project and describe scenarios in the form of drawings and
models (show the relevance to information needs and
applications to public health policy). Architects will also
scope engineering, structural and costing issues and other
aspects of the design that will be attractive to the client.
The stakeholder can now make serious decisions about
implementation—choice of builder (data collection contractors),
preferred building specifications (data collection
methodology), and submission of a planning permit (data
quality control plan).
A visit to an accountant and banker is now important to
ensure the project is economically viable (budget plan).
Processes such as these are all necessary to ensure
surveillance systems are sustainable and that core data
elements remain stable over time. As buildings must be
resistant to harsh environmental conditions, surveillance
systems must also be resistant to changes in the political or
bureaucratic environment.
Bringing in builders and technical teams is now easy—but
architectural supervision will be required throughout. Use
will be made of fast turnaround, automated building
machinery (software) and pre-fabrication where appropriate
to ensure efficient construction and fit-out of buildings
(analysis and reporting).
Following a successful building project, a sales office is
needed to attract homebuyers. Again, this process will be
made easier if early consultation and planning has been
thorough. Surveillance systems (like buildings) also require
marketing. Information must continually be communicated
to stakeholders in the most appropriate way.
The key message here is that architects (epidemiologists)
must engage with all stakeholders to ensure their needs are
met (product relevance and application) and that the quality
of life of the new residents (public health outcomes) will be
improved through construction of the new housing estate
(health surveillance).
B C K Choi
Population and Public Health Branch, Health Canada, AL no 6701A,
120 Colonnade Road, Ottawa, Ontario K1A 1B4, Canada;
Department of Public Health Sciences, University of Toronto;
Department of Epidemiology and Community Medicine, University of
Ottawa, Canada
M Ackland
Health Surveillance and Evaluation Section, Rural and Regional Health
and Aged Care Services Division, Victorian Department of Human
Services, Melbourne, Australia
Correspondence to: Dr B C K Choi;
Bernard_Choi@hc-sc.gc.ca
Life course epidemiology 783
www.jech.com
onApril21,2021byguest.Protectedbycopyright.http://jech.bmj.com/JEpidemiolCommunityHealth:firstpublishedas10.1136/jech.57.10.778on22October2003.Downloadedfrom
PostScript. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
LETTERS
If you have a burning desire to respond to a
paper published in JECH, why not make use
of our "rapid response" option?
Log on to our web site (www.jech.com),
find the paper that interests you, and send
your response via email by clicking on the
‘‘eLetters’’ option in the box at the top right
hand corner.
Providing it isn’t libellous or obscene, it
will be posted within seven days. You can
retrieve it by clicking on ‘‘read eLetters’’ on
our homepage.
The editors will decide as before whether
to also publish it in a future paper issue.
Social epidemiology,
intra-neighbourhood correlation,
and generalised estimating
equations
The recent editorial by Merlo1
offers an
interesting critique of the generalised estimating
equations (GEE) analysis of a paper
published in the same issue of the journal. In
the editorial, the author notes that the
paper’s GEE analysis treats ‘‘the intra-neighbourhood
correlation as a ‘nuisance’ that
needs to be adjusted in the analysis but not
explicity investigated’’ (page 550).
The editorial then becomes a call for an
alternative, more innovative approach in
social epidemiology: ‘‘Estimation of the
extent to which individuals within a given
neighbourhood are correlated with one
another in relation to health (that is, the
concept of intra-neighbourhood correlation)
has value in the context of ideas about the
efficacy of focusing intervention on places
instead of people’’ (page 551). The finale is a
logical conclusion that studies of intraneighbourhood
correlation may ‘‘...present
themselves as a new epidemiological
approach that may prove very useful in social
epidemiology’’ (page 551).
The author is appartently speaking of first
order GEEs but the JECH readership may not
appreciate that second order GEEs (GEE2)
treat the intra-neighbourhood and interneighbourhood
correlations into deliberate
objects of study and estimation.2
Although
we ourselves deserve absolutely no credit for
biostatistical innovations, the ‘‘alternating
logistic regressions’’ (ALR) approach3
we
use in our forthcoming article in this journal4
is a computationally efficient alternative to
GEE2 in the case of a binary outcome. As
such, it estimates the pairwise odds ratio,
which quantifies the degree to which health
conditions, behaviours, or perceptions might
cluster within neighbourhoods (or other
nested structures of community life) to a
degree other than one might expect if these
health conditions, behaviours, or perceptions
were distributed at random across neighbour-
hoods.
Because we believe our work is responsive
to the author’s call for a new approach in
social epidemiology that measures intraneighbourhood
correlation, we would welcome
an editorial comment on the potential
value (and possible shortcomings) of the
GEE/ALR approach we used in our forthcoming
article in JECH on clustering of
cocaine incidence in the United States. We
hope you will concur that our application of
the ALR approach is a step in the right
direction for research on contextual influences
and health.
K R Petronis
US DHHS, Substance Abuse and Mental Health
Services Administration, Rockville, Maryland, USA
J C Anthony
The John Hopkins University, Bloomberg School of
Public Health, Department of Mental Health,
Baltimore, Maryland, USA
Correspondence to: Dr K R Petronis, US DHHS,
Substance Abuse and Mental Health Services
Administration, 5600 Fishers Lane, 16-105 Rockville,
MD, 20857, USA; kpetroni@samhsa.gov
References
1 Merlo J. Multilevel analytical approaches in social
epidemiology: measures of health variation
compared with traditional measures of
association. J Epedimol Community Health
2003;57:550–2.
2 Liang K-Y, Zeger SL. Regression analysis for
correlated data. Annu Rev Pub Health
1993;14:43–68.
3 Carey V, Zeger SL, Diggle P. Modelling
multivariate binary data with alternating logistic
regressions. Biometrika 1993;80:517–26.
4 Petronis KR, Anthony JC. A different kind of
contextual effect: geographical clustering of
cocaine incidence in the US. J Epidemiol
Community Health (in press).
Author’s reply
I have read with great interest the comments
made by Petronis and Anthony on my
editorial.1
I have also read their forthcoming
article,2
and I believe they apply an analytical
approach that seems to be, in my opinion, a
step in the right direction for research on
contextual influences and health that focus
on investigation of clustering. I will be very
pleased to write a more extensive comment
and send it for consideration and possible
publication in the journal.
Measuring clustering with the aim of
obtaining substantive scientific information
is so far an uncommon approach in social
epidemiology and most multilevel analyses
have in fact been plain ‘‘contextual analysis’’3
focused on measures of association. This
seems to be true not only for studies using
GGE techniques but also for studies using
multilevel hierarchical regression. However,
the original standpoint of multilevel hierarchical
regression analyses is the investigation
of complex patterns of variation4
rather
than dealing with residual correlation.
Regarding the use of ‘‘GEE2’’ and measurement
of clustering, the comment of
Petronis and Anthony is certainly right.
From an epidemiological point of view the
most interesting question is the conceptual
rather than the mathematical approach used.
I agree with Petronis and Anthony in their
conceptual approach and I believe that they
put context back in epidemiology using
‘‘GEE2’’ techniques.5
The pairwise odds ratio
and other techniques for measuring neighbourhood
heterogeneity and clustering like
the median odds ratio and the interval odds
ratio6 7
deserve more development and
spreading.
J Merlo
Department of Community Medicine
(Preventive Medicine), Lund University, Sweden;
juan.merlo@smi.mas.lu.se
References
1 Merlo J. Multilevel analytical approaches in social
epidemiology: measures of health variation
compared with traditional measures of
association. J Epidemiol Community Health
2003;57:550–2.
2 Petronis KR, Anthony JC. A different kind of
contextual effect: geographical clustering of
cocaine incidence in the US. J Epidemiol
Community Health (in press).
3 Diez Roux AV. A glossary for multilevel
analysis. J Epidemiol Community Health
2002;56:588–94.
4 Goldstein H. Multilevel statistical models. London:
Hodder Arnold, 2003.
5 Carey V, Zeger SL, Diggle P. Modelling
multivariate binary data with alternating
logistic regressions. Biometrika 1993;80:517–26.
6 Larsen K, Petersen JH, Budtz-Jorgensen E, et al.
Interpreting parameters in the logistic regression
model with random effects. Biometrics
2000;56:909–14.
7 Larsen K, Merlo J. Appropriate assessment of
neighborhood effects on individual health—
integrating random and fixed effects in multilevel
logistic regression. Working paper 2003.
(Available on request to the author).
CORRECTION
An editorial error occurred in this article by
Kuh and colleagues (2003;57:778–83). In
figure 1 the top of the figure should read
Time [not Time (min)].
914 J Epidemiol Community Health 2003;57:914–914
www.jech.com