Life course epidemiology and analysis

Life course epidemiology approaches disease aetiology and prevention from the perspective of risk and protective factors that influence health and disease throughout the lifespan. The integration of a life course approach to epidemiologic research is central for identifying effective policies and programmes to promote population health and health equity. This chapter will introduce life course concepts and models and analytical approaches for research on life course determinants of health. It will discuss threats to causal inference, approaches for overcoming these difficulties, and future directions in life course epidemiology. For example, in addition to expanding epidemiologic research with a life course perspective to include people with diverse life experiences, new areas of development include life course research extending beyond one human lifespan to include intergenerational and transgenerational life course research, as well as the application of innovative methods.

1057Addressing challenges in life-course epidemiology: established and novel approaches using big data and twin/family studies

Focus and outcomes for participants Long periods between exposures and outcomes pose a number of challenges for life course epidemiological research, including unmeasured confounding factors (e.g.; familial factors) and mediation by other covariates, which make it difficult to unequivocally establish associations let alone causality. In this symposium we will present a number of different studies based on big data utilizing a variety of methods to overcome some of the issues encountered in research across long time frames or generations. Our focus will be on the different methods, the solutions they provide as well as their limitations. The methods presented were applied in the context of life course epidemiology and include: mediation analyses; genetic epidemiology; well-established and novel family designs including twins, siblings and cousins, and a method comparable to Mendelian randomization - ICE FALCON (Inference on Causation from Examination of Familial Confounding) which is part of a more general approach called ICE CRiSTAL (Inference on Causation from Examining Changes in Regression coefficients in STatistical AnaLsyes). The intended outcomes for the audience are to increase awareness of the challenges imposed by the data frequently used in this field of research and present possible solutions to (at least partly) address those. It is our intention to generate discussion and encourage other researchers to share their experiences and solutions to increase knowledge collectively. Rationale for the symposium, including for its inclusion in the Congress The main theme of the conference is ‘Methodological Innovations in Epidemiology’. Our symposium includes six different methods to strengthen causal inferences in epidemiology. While some of the presented methods are well established in classic epidemiology research (i.e. mediation analyses), others are more commonly found in different disciplines such as the expanding genetic epidemiology field (e.g. alternative twin designs and measured genetic risk approaches). In addition, we are presenting a new method for making inference about causation developed by Prof. John Hopper and Dr Shuai Li and co-workers called ICE FALCON, which applies to twin and family data and is part of a more general approach called ICE CRiSTAL. These methods use observational data to infer or rebut causality between measured variables, similar to Mendelian randomization (without relying on genetic information or strong assumptions). All the work presented is either nearing publication or has been published in the last two years and each presenter is intimately involved with the analysis and methods they will be presenting. Beyond a range of methods and study designs we have also a diversity of researchers and research questions in our symposium by including: researchers at different stages in their career and from around the world (ranging from early Postdoctoral Fellows over Senior Research Fellows/Assistant professors to Professors); a variety of research questions and diseases; and a range of population context including Indigenous Australians, babies, children, and adults, in order to appeal to a wider audience. Presentation program 6 talks of 8 minutes each with 2 minutes for questions followed by a general discussion facilitated by the chair. Names of presenters Dr Miriam A Mosing1,2

Life course epidemiology

Types of psychiatric disorders vary with respect to age of onset, temporal continuity, and impact. Life course epidemiology provides powerful tools for understanding these complexities. This discipline broadly distinguishes ‘sensitive period’ and ‘risk accumulation’ models. The former refers to optimum windows for exposure (e.g. early life for some psychoses, in contrast to proximal exposures for depression). Accumulation refers to additive or multiplicative effects of multiple exposures, exemplified by stress process and chain of risk models. The preeminent study design for these approaches is the prospective longitudinal birth cohort study, especially where multiple cohorts help to distinguish period and cohort effects. However, limitations such as balancing the need for repeated versus age-appropriate measurement, and non-random missing data, must be carefully considered. While the statistical workhorse for life course epidemiology is general linear modelling, this discipline also requires advanced tools such as random effects, path, latent class, and latent growth modelling.

Sources of confounding in life course epidemiology

In epidemiologic analytical studies, the primary goal is to obtain a valid and precise estimate of the effect of the exposure of interest on a given outcome in the population under study. A crucial source of violation of the internal validity of a study involves bias arising from confounding, which is always a challenge in observational research, including life course epidemiology. The increasingly popular approach of meta-analyzing individual participant data from several observational studies also brings up to discussion the problem of confounding when combining data from different populations. In this study, we review and discuss the most common sources of confounding in life course epidemiology: (i) confounding by indication, (ii) impact of baseline selection on confounding, (iii) time-varying confounding and (iv) mediator–outcome confounding. We also discuss the issue of addressing confounding in the context of an individual participant data meta-analysis.