Using Mendelian randomization to assess the causal effects of familial environmental tobacco smoke exposure.

Abstract

Disentangling the effects of genetic and environmental transmission is an important question for
both developmental psychology and epidemiology. For example, environmental tobacco smoke
exposure (ETS, when an individual is exposed to another individual’s cigarette smoking) appears to
increase liability to smoke and associates with ill health. However, it is difficult to establish causality
given problems of selection bias, reverse causation, and confounding. A popular approach in
epidemiology, called Mendelian randomization (MR), leverages the random inheritance of genetic
variants known to influence an environmental exposure to strengthen causal inference. This thesis
extends this quasi-experimental approach to assess the transmissibility of interpersonal effects, using
ETS as an illustrative example throughout.
One approach to using MR to study ETS is to select genetic variants which robustly associate with
someone’s reported ETS exposure. Chapter 2 illustrates one setting where this approach is
biologically plausible, looking at the effect of having more offspring on parental wellbeing. However,
Chapter 3 argues that selecting variants which associate with downstream traits, like ETS, probably
results in exposure misidentification, and therefore biased MR estimates.
An alternative approach leverages the random assignment to not inherited variants to study the
effect of the parental phenotype on their offspring’s phenotype. Half of the variants which increase a
parent’s risk of a phenotype will not be inherited by an offspring. These non-inherited variants
cannot affect the offspring via a direct genetic effect. Because the parents also inherited the variants
randomly, the not inheriting a variant is effectively randomization of the offspring to exposure to the
parental phenotype. Chapter 4 provides a theoretical motivation for this approach, while Chapter 5
and 6 use a related approach to study the effect of parental ETS exposure on offspring health.
Chapter 7 provides a philosophical discussion of what the aims of an MR analysis should be.

Date of Award	20 Jan 2026
Original language	English
Awarding Institution	University of Bristol
Supervisor	Marcus Munafo (Supervisor), Hannah M Sallis (Supervisor) & Stephen Burgess (Supervisor)