What motivated you to come to Bristol and do this programme?
With a background in philosophy, economics, and statistics, much of my previous studies have centred upon examining potentially causal relationships in medical and social contexts. Whilst studying for an MSc in Statistics I became interested in ongoing work at the Integrative Epidemiology Unit at Bristol, and after completing my studies I was fortunate to be given the opportunity to work in the Bristol Population Health Sciences Institute. Having the opportunity to discuss and develop interesting causal inference techniques, and contribute to debates surrounding important healthcare issues has served as my primary motivation for undertaking a PhD at Bristol.
What is the key research question of your PhD research project and what have you found out so far?
My PhD primarily focuses upon identifying and correcting for bias in statistical analyses. One issue of particular concern is the role of pleiotropic bias in Mendelian randomization methods, which can result in biased causal effect estimates and misleading study findings.
At Bristol, we have been able to develop novel methods for overcoming such difficulties. One such method is Linear Slichter regression, which utilises gene-covariate interactions to adjust for pleiotropic bias. This work has been presented at the European Mathematical Genetics Meeting 2017 in Estonia, the Mendelian Randomization Conference 2017, and is currently available through BioRxiv.
I have also had the opportunity to work on statistical packages which facilitate analyses within genetic epidemiology. Thus far I have contributed towards the development of the mrrobust Stata package with Tom Palmer, and the Radial MR R package with Jack Bowden, which are also available online. Though there is much work to be done, it is exciting to be at the forefront of Mendelian randomization methods development here in Bristol.
Where do you think your research could lead and what are your future career plans now?
At present we are living in a very exciting time for conducting epidemiological studies. With the proliferation of genome wide association studies, researchers have more information than ever before, and new methods for making the most of such data are growing increasingly important. It is my hope that methods stemming from my PhD can help address such needs, and subsequently contribute to research on important issues within population health.