MRC Integrative Epidemiology Unit (IEU) Seminar Series

Title: Accounting for genetic confounding in observational studies

Abstract: The notion of genetic confounding can be dated back at least to a letter by Ronald Fisher contesting the causal role of smoking on cancer in the 1950s. Since then, a variety of methods have emerged to account for genetic confounding and confounding in general in observational studies. Recently, I published a Nature Review Genetics article on genetically informed methods for causal inference, giving an overview of these methods. I will present some of our published and current empirical research implementing such methods. In particular, I will first present two sets of contrasting findings from the twin differences design to clarify the role of bullying victimization and birth weight on mental health. I will then present ongoing (and still initial) work attempting to embed polygenic scores into a sensitivity analysis aiming to estimate the role of genetic confounding in epidemiological associations, with an empirical example using intergenerational data.

Biography: Dr Pingault aims to better understand how early genetic and environmental risk factors lead to the development of mental health difficulties in childhood and adolescence. His team adopts an interdisciplinary approach building on disciplines such as developmental psychopathology, epidemiological psychiatry and behavioural genetics. They implement innovative methods for causal inference in big datasets, building on statistical innovation and genetically informed designs to better understand causal pathways from early risk to later mental health difficulties. In collaboration with national and international colleagues, Dr. Pingault and his team then seek to further characterize these causal pathways by investigating possible underlying biological mechanisms (e.g. cognitive profiles, epigenetics).

All welcome